CIMM UK Touchpoints Report

Agencies Carat

• i-Level
• MC&Co (Media Independent)
• MEC
• Mindshare
• MPG/Havas Media
• Naked Communications
• OMD UK

Media Owners

• Associated Newspapers (newspapers, online)
• IPC Media (magazines, online – part of Time Warner)
• ITV (largest commercial television company)
• News International (newspapers, online – part of News Corp)
• Sky TV (largest cable/satellite operator – managed by News Corp)
• The Guardian Media Group (newspapers, online, radio)
• ThinkBox TV (marketing arm of all major commercial TV operators)

Advertisers

• Central Office of Information COI (government advertising)
• Unilever
• Joint Industry
• Committees
• NRS
• UKOM
• RAJAR
• Other Organisations IPA, Telmar (Software bureaux), RSMB (contractor for data fusion)

Put together, the views and opinions gathered do not necessarily represent the position of the industry as a whole. The discussions and interviews were meant to help identify some of the key issues that dictate usage of, and attitudes toward TouchPoints by media and research professionals in the UK.

1.1. The development of TouchPoints

The IPA’s Media Futures Group started to look at these issues in 2003. Consultation with senior planning directors of member agencies identified that their “key demand was for the IPA to provide a tool to evaluate mix media schedules both for planning and post campaign purposes” (Robinson, Beck, Wilcox 2006). Accordingly, the project was given three clear remits:

• • to deliver new insights in how individuals consume media,
• • to bring together the industry currencies (and other data sources) in a single database,
• • to act complementary to existing currencies and not aim to replace them

The agreed solution was to develop a consumer-centric multimedia survey that could deliver new insights as a stand-alone study and at the same time act as a central Hub through which the various currencies (and other data bases) could be brought together via a process of data integration, that would preserve the original currencies and become the Integrated Planning Database (or Channel Planner as it is now called) for multi-media schedule evaluations. The first TouchPoints survey was launched in summer 2006, followed by TouchPoints 2 in 2008. TouchPoints 3 is due to be published in July 2010 (the Hub survey) and September 2010 (the Channel Planner).

1.2. TouchPoints – its purpose and role in the planning process

The IPA TouchPoints data was conceived so its two key elements, the Hub Survey and Channel Planner can be used across all stages of the development of brand communication strategies (see Table 1). In practice the currencies in their original format will continue to be a key tool to be used by the various channel specialists (print, broadcast, digital etc) for any fine tuning of the print, TV or online schedules. What it new (and unique) with TouchPoints is that results can now be fed into the Channel Planner to get a more holistic estimate by producing net unduplicated reach, frequency and GRPs for all channels combined. A full list of the channels covered by TouchPoints and the available planning inputs can be found in Appendix 3.

Table 1: TouchPoints throughout the planning stages

1.3. Methodology

After intensive testing and based on experience with the BBC’s Day in the Life time-budget studies (carried out by TNS using a PDA device the collect electronic diary data) it was decided to collect data for the Hub Survey via a detailed self completion questionnaire (Appendix 2a) combined with a seven-day electronic diary using a PDA device, which respondents used to record the following in half hour intervals (Appendix 2b):

Table 1: TouchPoints throughout the planning stages

There was no face-to-face contact with respondents. All recruitment and briefing was done via telephone and questionnaires and PDA’s were sent and collected using standard mail delivery – only 6% of PDAs ‘got lost’ in this process. The resulting data nevertheless required some extensive editing and of the 8,000 original panellists only 5,010 (67%) were accepted for final analysis, which was based on close to 1.7m half-hour time inputs. The second stage comprises the development of individual exposure probabilities for each respondents and media vehicle and the integration of the industry currencies onto the hub survey (Table 2).

Table 2: Currencies and other data sources used in TouchPoints

This phase of TouchPoints is probably one of the most ambitious data integration projects ever attempted. More detailed descriptions are available from various sources (Baker 2006, IPA 2006, Robison, Beck, Wilcox 2006) and in chapter 2 of this report. One key difference to other fusions of media data is the availability of media usage data in the Hub survey, which has provided additional fusion hooks and helped to preserve more of the discriminatory power of the donor (i.e. currency) surveys.

Due to the complexity of this task TouchPoints data is released in two stages. Results from the Hub survey (since TouchPoints 2 these already includes the integration of TGI data) are usually published two to three months before the Channel Planner with the full fusion of the currencies becomes available to subscribers.

1.4 Changes and developments

There have been no principle changes to the survey since it was first introduced in 2006 but the IPA has always been open to feedback from end users and where necessary used learnings from one survey to make improvements to the next (for a detailed review of the changes see Robinson 2009).

Fieldwork for TouchPoints 1 and 2 was carried out by TNS, which had been involved in other surveys using PDAs as a method of data collection. For TouchPoints 3 the contact was switched to IpsosCT, which is also responsible for the National Readership Survey and thus could offer to re-contact NRS respondents to recruit the PDA panellists.

For all three TouchPoints surveys the data integration process has been the responsibility of RSMB, one of the market leaders in data fusion and weighting. One major change regarding the sequence in which currencies are integrated has been made in 2008: since TouchPoints 2 the Hub survey when launched already includes the TGI product usage data, making this new ‘superhub’ even more useful for subscribers from the start. With TouchPoints 1 users had to wait three months until the full Channel Planner (then know as the Integrated Planning Database or IPD) became available to get the full benefit of the Hub data.

For TouchPoints 2 both the self-completion questionnaire and the diary questionnaire have been improved, specifically to include more detailed information on digital media and on the usage of traditional media’s websites. New questions on video and audio streaming and VOD have also been added. In addition the shopping sections have been expanded to include a wider range of outlet types. The lifestyle has been reviewed and statements relating to environmental issues and attitudes toward new technologies have been added. For the Channel Planner a modelled ‘currency’ for online search was developed “to allow users to input gross impressions by search engine and product category to quantify search’s contribution to multi-media schedules” (Robinson 2009).

TouchPoints 3 promises further improvements and refinements, such as an extended section of the important Word-of-Mouth channel, which has been developed in cooperation with the Keller Fay Group, a US specialist in WOM measurement. The additional data will provide a better understanding of the role of ‘influencers’ and their relationship with media and demonstrate the role of advertising as a generator of valuable word of mouth activity.

1.5. Support and Outlook

Support for the TouchPoints initiative has not been universal. It is clear that in the run-up to the launch of TouchPoints 1 in 2006 expectations had probably been raised to unrealistically high levels and in practice everyday work with the data turned out to be considerably more complex and time-consuming than many subscribers had anticipated (for more details see chapter 3). After the initial euphoria support for and usage of the data seemed to fall off and a number of subscribers decided not to continue with their funding.

“We just felt we didn’t really make enough return on our investment in TouchPoints 1 as we were then too much focused on our print offers and hadn’t really integrated other platforms fully. Today this view may be changing again” (head of research/media owner)

“When Touchpoints 2 came out the whole insight team was charged to use the data as much as possible. When we reviewed the situation we realised that TouchPoints can’t always give us what we wanted, which is rich data for all our television channels, our website and our interactive TV platform. In the end TouchPoints doesn’t give us enough additional information to benefit our sales efforts. We already have to buy BARB and TGI data and it is difficult to justify the extra budget” (head of commercial insight/TV channel)

A further very important factor that had a negative impact on the popularity of the second release was the looming economic downturn, which hit the UK industry and consumer confidence hard and even at the time of writing this report the jury is still out as to when real and sustained growth will return. Overall advertising expenditure fell by for two consecutive years, according to the latest IPA/Bellwether Report, hitting the marketing and research budgets of media owners very hard, which partly explains their decision not to renew their subscription. Another reason for the limited support by media owners has to do with the way many sell space or airtime.

“TV companies use BARB and TGI to sell against each other and they don’t generally sell against other media, such as print or radio.” (head of research, TV)

“We used to sell our titles rather than newspapers as a channel” (head of research, media owner)

Advertisers never showed much direct interest in TouchPoints but in fairness it should be pointed out that in the UK very few advertisers get directly involved in media research and

planning. Rightly or wrongly, most matters concerning media decisions and media selection have been transferred almost entirely to media agencies. Advertisers rarely participate in any of the ‘Joint Industry Committees’ that are responsible for audience measurement and largely leave representation to their association, the Incorporated Society of British Advertiser (ISBA). Currently the sole exception with regards to TouchPoints is the Central Office of Information (COI), which handles all official government advertising, such as health warnings, drinkdriving campaigns, anti-knife initiatives aimed at young men and many others. The COI has mandated that every agency working on or pitching for its business must use TouchPoints data at some stage, either for strategic insights or cross channel planning. The COI analysts clearly derive huge benefit from the usage of TouchPoints and usage has further been encouraged by the organisation’s head of marketing and communications, who was a prominent member of the IPA Media Futures Group that developed the TouchPoints idea in the first place back in 2003-04. Some of those interviewed for this report had doubts as to whether the organisation would have taken an equally strong position on TouchPoints without his presence.

Given the stance of the COI and the fact that the organisation has been one of the top three UK advertisers (in budget terms) over the past 10-15 years it is not surprising that many agencies are pitching for its business (and indeed campaigns have traditionally been spread over a large roster of advertising and media agencies). As a consequence almost all major network agencies as well as some of the independent outlets are subscribers to TouchPoints.

“We work on COI business and the COI mandates the use of TouchPoints”

is one of the most regularly quoted reasons why the survey is used. In addition some of the Media Futures Group members are heading key agencies and as such operate as champions for TouchPoints within their companies.
There is of course nothing wrong with this – most major initiatives rely on individual champions in order to succeed in early years.

“Take-up of TouchPoints may not have been as rapid as some in the industry had hoped but there can be little doubt that the COI’s insistence of using the data has been a massive driver”(business director/agency)

It will be critical for TouchPoints 3 to reduce the reliance on individual champions and ensure that the survey creates its own momentum to drive it forward. It remains to be seen whether the ‘tipping point’ can be reach during this year on not. The feedback from the industry on this is generally cautiously optimistic.

Part 2. The Hub plus Fusion Concept

Data integration has a long history in the UK. Since the early 1990s the Target Group Index (TGI) has been used as the donor survey to be fused with BARB, to enrich the official TV currency with the wide range on product and brand information contained in TGI. Thirteen common hooks were used (demographic, TV hardware ownership, weight of viewing). The degree of regression to the mean (RTM) varied – for consumer goods (asked only of housewives there was virtually no RTM; lifestyle and/or attitude related purchases asked of all adults showed less retained efficiency (in some extreme cases there was a 23% RTM) but the overall “RTM was an encouraging low 14%” (Bedwell 1991). Subsequent BARB/TGI fusions retained this level of efficiency “today the BARB/TGI fusion is viewed as highly successful” (Baker 2007) and new data is released quarterly.

The fusion process deployed for the TouchPoints survey is considerably more complex as it involves a multitude of individual databases and a more detailed description of the technical issues of the TouchPoints fusion can be found in various IPA material and conference papers (Robinson, Beck, Wilcox 2006; IPA 2006, Wilcox, O’Sullivan 2007, Baker 2006)

The TouchPoints survey itself (i.e. the central Hub survey) is a single source database measuring multi-media consumption habits of just over 5,000 respondents. In some cases the media data from the Hub survey is very similar to the currency, for example for radio both TouchPoints and the currency (RAJAR) collect half-hour listening data; for print the selfcompletion questionnaire was used to record newspaper and magazine reading replicating the NRS questionnaire. In other cases, such as Outdoor the Hub only provides a surrogate measure in from of time spent travelling by mode of transport.
Furthermore, respondent level data fusion is only possible with those currencies that report at individual respondent level (TV, radio, print, TGI). Alternative integration techniques had to be used for other currencies.
Last but not least the significant variations in sample size posed another problem. The Hub survey is based on approx 5,000 respondents, but others are considerably larger (the BARB panel reports on 11,000 individuals, TGI is based on 25,000 respondents, the NRS sample is around 36,000).

2.1. Expansion of the Hub survey

A first step in the integration process required the expansion of the TouchPoints sample by re-weighting it to an appropriate host survey - one based on a large and random sample. The 50,000-strong BARB establishment survey fulfils both criteria.

(a) As a random sample it overcomes any possible bias within the smaller TouchPoints sample by weighting the data to the correct population profile. For example the original sample of TouchPoints 1 overstated ‘technophiles’ with 70% claiming to have an internet connection at home (whereas the BARB ES showed only 51%). Clearly some re-weighting was required in any case.

(b) Due to the size of the BARB ES the full sample of the various currencies can now be used in the subsequent fusion with this recipient file, ensuring a rather more even use of donor respondents. For example, the NRS sample is seven times larger than the TouchPoints sample and a standard fusion would have resulted in using less than 15% of the currency sample and its effectiveness would be severely reduced

It was therefore decided that the Hub survey should donate media and other diary measures to the Establishment Survey, whose demographic and socio-economic structure is an industry standard. For every TouchPoints respondent there exist a larger number of respondents in the Establishment Survey who have the same demographic profile. The exact number of matches found varies because the Establishment Survey has a more representative profile. The common variables included various standard demographic and socio-economic measures combined with media measures such as multi-channel home, number of TV sets at home, internet access, and broadband connection. On average Hub donors were used approx ten times.

It must be stressed that this process of weighting and replicating the original TouchPoints sample does not in any way increase the effective sample of the Hub survey; it simply means that overall the sample is now more balanced and the currency samples can be better preserved in the next stage of the integration process.

2.2. The Integrated Database

In order to allow cross media schedule comparisons the various currencies needed to be integrated into the TouchPoints Hub survey. Because currency surveys are conducted in different ways no single integration technique was available for this task and different methods had to be used:

• Respondent level fusion was used for print, radio, television, and product usage surveys

• calibration was used to integrate data for cinema and posters

• profile matching was used for the regional newspaper survey

• data for online usage, direct mail, and SMS was already contained in the Hub survey, as there were no industry currencies were available during TouchPoints One and Two (online usage data from the newly established UKOM is due to be fused with TouchPoints 3 in fall of 2010).

Whilst demographics are generally strong discriminators of behaviour for all media there is still a concern that fusion on purely demographic variables does not always perform well against life-style lead products, such as cosmetics resulting in higher regression to the mean. The TouchPoints Hub survey provides media-based variables that can be used as additional hooks together with demographic and geographic variables. The principle of the fusion process is to match respondents in the currency (donor) survey with respondents in the Hub (recipient) survey who have the same demographic and media imperative profile. These donors’ media currency data is then assigned to the TouchPoints recipients and replaces their original TouchPoints media data. (Wilcox, in IPA 2006).

The end result of the fusion procedure is an integrated database with the following information now available for each respondent:

Standard demographic, geographic and geodemographic classifications

Product usage and ownership from TGI

Media usage records from the individual (and now integrated) currencies

2.2.1. Personal probabilities

For multimedia reach and frequency planning the different contact definitions that have evolved over time for each media channel required a solution. With the exception of BARB (a panel with long-term viewing records for each panel member) other currencies have only short-term contact measurement (reading in recent publishing interval for print, a one-week quarter hour diary for radio), and therefore depend upon some form of probability expansions to estimate the achieved longer-term contacts over a campaign schedule. As each media currency has a different probability model for this expansion, the solution was to adopt - as a common denominator - personal probabilities, which cover all media channels. If a person does or doesn’t make contact with a specific media event (e.g. a TV spot, yesterday’s issue of a newspaper, a radio station in a particular quarter hour) then their personal probability is one or zero, but this says nothing about that person’s probability of making contact with the equivalent event on another day or in another week. What needed to be calculated were the underlying probabilities of making contact with each media event, i.e. a number between 0 and 1. Then if that media event is repeated a number of times, a binomial expansion can estimate a particular person’s probability of making 0, 1, 2, 3……contacts. For TouchPoints this approach is used for television viewing, readership, radio listening, direct mail and SMS text messaging. Other media are not available for planning at such a granular level and a different approach was used, based upon underlying rates of exposure; for example the number of poster sites passed in a week. By scaling these rates to the weight of a particular campaign, the Poisson model can be used to expand to a particular person’s probability of making any given number of contacts. Personal probabilities for outdoor, cinema

and online are calculated in this way. (A more technical summary of this process has been written by Wilcox, in IPA 2006).

2.3 Quality of the TouchPoints integration process

The IPA commissioned consultant statistician and fusion specialist Ken Baker, to assess the quality of both the Hub expansion and the subsequent currency fusions. With regards to the Hub survey he found that, age aside, the resulting degree of exact matching was equivalent of 17 out of 21 characteristics being perfectly matched for each respondent. Exact matching is more problematic for variables such as age (especially for actual age but less so when standard age groups are used), social class, or terminal education age (TEA), but once the criteria are widened to ‘matching exact plus one cell out’ the degree of fit rises to 80% and over. “This is totally in line with the degree of exact matching with classic fusions such as BARB/TGI” (Baker 2006). As to the various currency fusions the results are generally very encouraging:

• RAJAR: most matches achieved 80% or higher efficiency on exact matching; standard age groups achieve 61% (increasing to 90% for exact match plus one out). “The matching process worked as well as a typical fusion” (Baker 2006)

• BARB: matches achieved a slightly lower efficiency compared to radio, according to Baker is at least in part due to the relative small size (just under 8,800) of the BARB donor file, which required each donor to be used on average 5-6 times but he nevertheless concludes that “the match rate is acceptable in comparison with other fusions” (Baker 2006).

• NRS: Again Baker’s conclusion is positive (“the general level of matching on common variables is up to the level of a normal fusion”) but he does point out that the algorithms matched ‘disappointingly’ on age and social grade.

• TGI: of the 26 common variables used, on average each respondent was matched perfectly on 20 of these variables, making “this the most accurate fusion to date … exceeding the original BARB/TGI fusion” (Baker 2006)

Overall, Baker’s analysis of the TouchPoints fusion finds little evidence of any systematic RTM and “the overall estimate of RTM is 7%, i.e. 93% of the discriminatory power has been retained by the fusion” (Baker 2006). No further independent assessment has been commissioned for TouchPoints 2 but the IPA is now considering the need for another appraisal in connection with TouchPoints 3 later this year. (Please see Appendix 4 for a more detailed version of Ken Baker’s fusion assessment and Appendix 5 for examples comparing results between the original currency and TouchPoints)

2.4. Lost in the fusion

Fusions can never completely replicate the original database and some data invariably changes or is missing because it is not transferred into the fused database. Generally speaking only data that is required for schedule inputs is available within the fused data. For media planners (and vendors of airtime and space) the preservation of the currencies is vital to encourage usage of TouchPoints. This was an important objective for the survey and has certainly been achieved, especially for newspapers, magazines, and radio. For television the fused data on TouchPoints 1 showed a 10% increase in total viewing hours over the original BARB results, which was somewhat surprising, given the use of media imperatives, i.e. only matching heavy/medium/light/non-viewers of specific channels. To correct this all results have been checked and calibrated back to their ‘correct’ levels achieved in the original currency survey and “for RAJAR and NRS the calibration factor is largely minimal” (Baker 2006). Calibration was used to adjust results, if necessary, against GRPs, coverage and frequency distribution.

With TouchPoints 2 the fused BARB television data shows a much-reduced and smaller uplift of only four percent in TV viewing (prior to final calibration). Some BARB TV metrics are missing in the fused database: the original BARB data shows individual spot ratings; in the fused database this information is no longer shown. Time-shifted viewing is included after fusion but can no longer be separated out. For newspapers, too, some granularity has been lost. There have been no reports in any systematic change of net reach scores but some readership profiles can be affected by the fusion - after all, it is impossible to control for all variables. Also, the original NRS file contains readership scores based on Monday – Saturday, Monday – Friday, Saturday, and Sunday reading; the fused NRS file only reports on six-day (Monday – Saturday) and Sunday reading. In addition there is no data on ‘amount read’ in the fused database. For radio only those stations that achieve a minimum weekly reach of 250,000 listeners (approx. 80) have been transferred – the original RAJAR survey reports on well over 300 stations.

With regards to outdoor media the fusion generally worked well in terms of reach but less well with regards to frequency distribution. Mid-range frequencies (3-6+ exposures) in the fused database are noticeably lower than on POSTAR – this applies to both TouchPoints releases. The Outdoor measurement in the UK has been re-vamped since TouchPoints 2 (for example introducing GPS technology) and there is cautious optimism that the new data set (due out this summer) will improve results.

One final, if self-evident comment: media currencies are updated regularly. The NRS and RAJAR report quarterly, BARB reports on a continuous basis – overnight results are available

daily at 09:30 every morning. For the fusion a data file that most closely coincides with the TouchPoints fieldwork is chosen for the fusion onto the Hub survey. For television this means the fused BARB data represents scores averaged over a 12-week period. Seasonality, which can have a significant impact on media consumption, tends to get lost and with regards to TouchPoints 2 the available data is now over two years old. This may well be a crucial issue once the new online data becomes available for fusion via UKOM. Too much is still happening in the digital and online world and more regular updates of TouchPoints may have to be considered for the future.

Part 3. TouchPoints’ reception and usage in the UK advertising market

The following comments are largely based on discussions and interview with users as well as non-users of TouchPoints during April and May of 2010 (see chapter 1 for a full list).

3.1. General Acceptance of the TouchPoints Concept

Earlier in this report (see 1.5. Support and Outlook) of the results from these discussions have already been covered, mainly TouchPoints’ heavy reliance on individual champions and the economic downturn, that hit the survey (and budgets of potential subscribers) after the second data release in 2008 and made it more difficult for the IPA to increase its subscriber base.
Nevertheless, none of the thirty individuals interviewed, whether they were currently users or non-users of TouchPoints raised any principal objections or doubts about the validity of the concept and all expressed the view that the survey is on its way

3.2. Data usage

The degree to which the TouchPoints data has been used over recent years depended on a variety of factors and many of the key reasons that are mentioned as detrimental have little to do with the survey itself. Some of these factors are nevertheless important issues that have an almost disproportionate (negative) influence on end users’ attitude towards TouchPoints and those responsible ignore them at their peril.
Only one person interviewed expressed some general doubts about the usefulness of the survey.

“Five years ago, when I was still working in a traditional media agency I would have loved to have TouchPoints but now it does not really fit in with this agency’s philosophy. The hub element does not cover anybody under the age of 15, and young people are an important segment of the population for many of our campaigns. The Integrated Planning Database is only good if the aim is to maximise reach and frequency across different channels. Our campaigns are aimed to change behaviour and maximising effectiveness is more important than maximising net reach” (senior strategist/agency).

3.2.1. The Hub Survey

It is very obvious that the data from the Hub survey is much more widely used than the IPD or Channel Planner. Some claimed that Hub usage accounted for up to 75-80% of all their TouchPoints analyses, but there is some evidence that over the years the balance started to shift to a more even usage of all TouchPoints’ elements.
The Hub survey contains a number of key elements that make this part of the database particularly attractive:

• The time factor, which brings to life a typical day of British consumers, showing what they do and where they are at any given part of the day. The data shows what is being done throughout the day, as well as activities that peak at certain times during the day. Comparisons can be made between different target groups and data is available for every day of the week, allowing comparison between working days and weekends.

• Activities, including the consumption of a wide range of media channels can be split into solus activities and ‘multi-tasking’ and further refined by analysing whether the respondents were alone or in the company of others.

• Respondents also recorded how they felt at specific time intervals (good mood vs. bad mood, relaxed vs. alert). Some respondents felt that this question could be better phrased, indeed the antonym to ‘relaxed’ should be ‘tense’, ‘anxious’, or ‘stressed’. When the data is included in an analysis results are not always particularly enlightening. Some people may be in a bad mood at certain times on certain days but it doesn’t really explain why this should be and consequently this data seems to attract less usage.

Most of those interviewed mentioned two, equally important benefits they derive from the use of the Hub data.
Roughly half of them claimed to have gained new insights into consumer behaviour and media consumption pattern, all of which contributed to significantly to the overall quality of their strategic planning and the development of ideas.

“We use the Hub data for segmentation work and target group information as well as a lot of the time-diary data. We get a lot of insights that are useful when pitching for new business” (planner, agency) “There were learnings that we did not expect, for example in the area of parallel media use between television and online. There is some interesting data on how people download TV programmes and videos, use VOD and mobile access. Using TouchPoints we could recommend a TV plus online campaign, whilst previously they only used TV” (planning director, digital agency) “Initially we only used the Hub data but this has changed now. We use it for insight in people’s activities and media usage and for broader channel selection (insight manager, agency) “The Hub was very good; especially the day-part information gave us more insight and a lot of interesting information for our sale presentation” (head of research, media owner) “We are currently not subscribers and the Hub data could be very useful for our marketing and promotion efforts” (head of research, TV) “With TouchPoints 3 coming out soon the Hub data should provide us with a lot of trend analysis over the past five years. It also gives us very granular data on media usage throughout the day and across the week” (head of research, YV)

At the same time there were many (generally more senior planners or insight/ research managers) who felt that whereas few truly new insights were delivered the data was nonetheless important as TouchPoints provided valuable empirical evidence to back up longheld opinions and assumptions.

“We didn’t necessarily gain a lot of new insights about our readers but had many of our assumptions confirmed. In an age of accountability this benefit should not be underestimated” (strategic insight specialist, media owner).

“We nearly always use the Hub survey. The time diary is particularly useful and over time we got quite good at using it to its full potential. On balance the data is still used more often to back something up but sometimes we get utterly new and fresh insights just by playing around with the data” (insight manager, agency)

3.2.2 The TouchPoints data integration

Despite consistent and repeated prompting virtually nobody of those interviewed had any issues with the general concept of data fusion or the method chosen by RSMB for TouchPoints.

In the last few years people have become used to what data fusion can do. It is now a very commonly used technique by many of our clients who fuse their own customer segmentation data with TGI or other surveys. It is the best and most cost effective option we currently have” (insight manager, agency)

“Data fusion is now very accepted as long as it is transparent and accountable. There are always plus sides and downsides but with TouchPoints we are definitely gaining more than we are losing. We are no closer to a single currency for everything but TouchPoints is getting us there. If the survey did not continue it would be a big step backwards” (head of commercial insight, TV)

In the case of the more junior media planners this is not really surprising, as they often lack the basic knowledge of statistics and are neither trained (nor encouraged) to get involved in technical discussion. The main response from almost everybody was that they trust a survey that has the backing of the IPA and those more senior planners and insight managers trust RSMB (an industry-wide acknowledged specialist in TV research, data weighting, and fusion) and the positive assessment of the fusion process given by Ken Baker, an independent fusion specialist.

“I trust the IPA and RSMB to deliver the best possible quality data” (insight manager, agency)

“Because the data comes from the IPA, a reputable and established body we and our agency feel comfortable using the information” (head of marketing services, advertiser)

“I go with the experts like (Steve) Wilcox and (Ken) Baker. What’s more, the results make sense, and if something makes sense it is probably right. At top level there is absolute consistency between TouchPoints and the original BARB data. It is very close .. I am very pleased” (head of research, television)

“I trust the IPA and their research contractors to get it right” (senior strategist, agency)

A couple of those interviewed were honest enough to admit to a lack of understanding of the fusion process:

“It’s all a bit complicated and nebulous so people tend not to worry about it (fusion and data collection method) too much” (planner, agency).

Only one of those interviewed voiced some generic concern about data integration, referring to the danger of ending up with

“bland data and regression to the mean. For example I would have preferred the original hub – the expansion to the BARB Establishment Survey has probably diluted the results. Too much fusion prevents any unique insights from coming out” (insight manager, agency),

but when pressed could not give any concrete example of this happening with TouchPoints.

In general, the UK advertising industry seems to take a very pragmatic approach and understands that there is always a trade-off between different research approaches.

“In the end it is an issue of practicality vs. perfection. Sometimes we need to make compromises to move things on. Too often we only talk for years about new ideas and concepts and never do enough. The ambition to bring together all platforms was in important one. Single source is a wonderful concept but it becomes apparent very quickly that it is not a practical way forward”. (head of research, media owners)

3.2.3. Sampling Issues

Another issue that was mentioned frequently was the sample size of the (original) Hub Survey. Most users of TouchPoints data are (at least vaguely) aware of the difference between the original (5,000) and the expanded (50,000) hub sample. Unfortunately the software providers (Telmar, IMS, and Kantar) show the expanded Hub with an unweighted count of 50,000 for the total population (and whatever equivalent for selected target groups). If users remember (there are no prompts to remind them) they usually divide the number shown by 10 to get an idea of the actual hub sample. It is of course by now means guaranteed that every planner does this on every occasion. It is quite possible that someone sees an unweighted count of 400 and embarks on an elaborate analysis not realising the actual sample size. (The author believes that in future the software bureaux will rectify this and some warning will be given on screen).

The limited sample has other consequences. Regional planning can be difficult (especially for narrowly defined target groups) and generally speaking TouchPoints is being used mainly to evaluate national campaigns.

“TV is planned and bought mainly nationally these days and regional sample sizes are not a big issue. For network audiences the Hub sample is sufficient” (head of research, TV)

“We don’t really do much regional planning so this is not a massive issue for us” (planning director, digital agency)

“Sample size is OK for top line level but for smaller media owners or specific targets it can become an issue” (planner, agency)

“There are a lot of things you have to bear in mind when working with TouchPoints data. You have to remember to divide the grossed-up sample by ten and sometimes people simply forget. There is currently nothing to flag this up”. (insight manager, agency)

Only one of those interviewed voiced some generic concern about data integration, referring to the danger of ending up with

“bland data and regression to the mean. For example I would have preferred the original hub – the expansion to the BARB Establishment Survey has probably diluted the results. Too much fusion prevents any unique insights from coming out” (insight manager, agency),

The hub sample also limits the usage of TouchPoints for media owners. This is less of an issue for radio stations, which are often bought in groups by genre rather than every station individually. As to print media TouchPoints clearly favours newspapers with larger readership (such as the Sun, News of the World, Sunday Times Daily Mail) but for titles such as The Guardian, Observer and The Times sample size can be a real issue and limits how deep they can drill into the data. That said none of the current non-users claimed that sample size was a major factor in their decision not to subscribe.

There seems to be some confusion as to the apparent (50,000 as shown), actual (5,040 Hub) and effective sample size (never officially calculated, but Baker 2006 gives an estimate of just 3 ,400) and a number of people expressed the need for better guidance on actual sample size and the impact on confidence limits. The much used ‘divide by ten rule’ may indeed be a little simplistic but is currently the best advice available.

3.2.4. Frequency of data release

Overall users had mixed comments on this. In an ideal world most agree that TouchPoints data should be collected throughout the year and released at regular intervals (the BARB/TGI fusion is now updated quarterly). At the same time everyone is realistic enough that this would have huge cost implications and few are hopeful, that the necessary budgets can be found in the current economic climate.

Attitudes may of course change once the new Nielsen online measurement results become available from UKOM. Usage patterns in the digital online world are still evolving and a number of users expressed their concern.

“Two year old data is not ideal but we can’t use this as an excuse not to use TouchPoints. We have to be realistic.” (planning director, digital agency) “Working with two year old data for our rapidly growing online platform is getting problematic” (strategic insight specialist, media owner)

“In the end we expect TouchPoints to give us some broad brush strokes, for example on a typical Monday evening more people like to watch television in combination with other media and these patterns don’t actually change that much. It is also down to the planner’s intelligence to get the best interpretation of the data” (business director, agency)

“We recently had a client who wanted to integrate their latest customer segmentation data into TouchPoints but decided against when they found out that the data was

almost two years old an they did not want to wait six month until the next release” insight manager, agency)

“Once we have the new online currency fused into TouchPoints the frequency of data release may well become more of an issue” (planner, agency)

“Every two years TouchPoints needs a major re-fresh. But I am not sure whether media usage changes significantly enough to warrant more frequent updates” (head of marketing services, advertiser)

In future there may well be increasing demand for more frequent updates as it is doubtful whether the new online data when fused into the 2010 TouchPoints Hub is still of much use to digital planners in 2012.

3.2.5. Digital channels

Most of those interviewed accepted the limitations of information regarding the digital world. In the absence of an industry currency in the UK, data had to be collected within the Hub survey itself. Given the richness of the hub data this in itself was a major step forward in understanding how the British public use the Internet and mobile devices. Visits to some 50 named websites, 40 site genres, and usage of five major search engines are measured. The self-completion questionnaire includes a detailed section on mobile (cell) phone usage and the electronic diary records mobile phone usage as well a reception of commercial text messages (see Appendix 2a and 2b).

Not surprisingly, some found the digital channel data of TouchPoints 1 and 2 of limited use.

“Many of the questions relating to online and digital come across as fairly generic. The time diary stuff is great for a very general picture of internet usage, but it is not detailed enough. As an agency we would look at 300-500 websites for a client but on the TouchPoints Hub survey there are just around 30-40 websites. We would know what there is to know from other sources anyway and there is nothing really new. Constructing a media plan cross channels worked reasonably well for TV, radio or print but not so well for online”. (planning director, digital agency) “For many of the communication channels we are using, especially online and social media the Channel Planner is of limited use” (senior strategist, agency)

There is now great expectation that the new online currency (Nielsen’s NetRatings) will provide significantly more data on online usage.

“Digital media was poorly represented until now but this will be rectified with TouchPoints 3 and UKOM data” (research manager, agency)

“Online plans can be quite complicated and you have to take them back to a much more generic level within the Channel Planner and that doesn’t really give us what we want. This will hopefully change with the arrival of UKOM data” (insight manager, agency)

“I really hope to see better digital and mobile data in the future” (insight manager, agency)

“We wanted to see Nielsen and comScore data fused into TouchPoints before but as they as they did not have currency status we would have had to do this on our own. With UKOM there will be a huge change in the data for digital planning” (head of research, media owner)

But not all are completely convinced that digital and online planning will improve as much as many hope:

“Until last year most TV channels had a separate sales operation for their online offers, and often sold against each other. Now these departments are being integrated which should make TouchPoints more valuable to individual channel operators to sell across both platforms. But I was shocked by how poor UKOM still is, especially the total lack of demographic data” (head of research, TV)

According to James Smyth, UKOM’s director of operations Nielsen is now in the process of re-contacting their 35,000-strong panel to collect members’ basic demographic and location details, which are required before any results can be fused with TouchPoints 3 later this year. Other issues, for example which set of Nielsen data is to be fused are not yet resolved and this is very much work in progress at this stage.

3.2.6. Working with TouchPoints data

Given that the release of TouchPoints 3 is now imminent there has recently been more modest use of the survey due to its age. This is simply a temporary blip and was to be expected. There are much more serious issue affecting the usage of TouchPoints: most of those interviewed spoke of

• a longer than expected learning curve

• difficulties understanding the data available

• complex coding requirements

• poor software support

3.2.6.1 Understanding the multi-media world

As with any new initiative end users need to be willing to learn and embrace the likely benefits offered. Expectations were high in the run-up of the first launch in 2006 but potential users quickly ran into difficulties analysing the data. Many found the huge amount of information contained within TouchPoints difficult to understand and struggled with complex coding requirements. This had a major impact on usage, especially the use of the Channel Planner:

“The IPD was used less often, partly because there initially there was little interest from agencies but also because it was difficult to use. It was not easy to explain the procedures to junior staff and only experienced planners and research executives really understood what they were doing. In the end we had to create a position for a specialist within our ad planning group” (head of research, media owner).

“The IPD is just too complicated to handle unless you use it very regularly, which we don’t” (head of research, TV)

“We found planning for regional media very cumbersome” (planner, agency)

“TouchPoints provides too much data – we already have an extensive proprietary research programme including a 30,000-strong return path panel and I simply do not have the staff to work with yet another large database. From what I hear it requires a fair amount of training and learning in order to take full advantage of the benefits offered by TouchPoints” (head of research, TV)

“Our sales people are print and online specialists. They were hesitant to use other channel data in presentations as they do not really understand the planning parameters for example for television and they feel uncomfortable discussing TouchPoints data and multi-media campaigns with their agency contacts. … I now for a fact that my TouchPoints charts are sometimes deleted by our sales executives” (strategic insight specialist, media owner)

“Sales staff usually insisted that a member of the insight team would attend agency presentations that included TouchPoints material” (head of research, media owner) “Our planners have direct access but we had to develop some internal tools and templates to simplify their task – a kind of TouchPoints Express. More complex analyses are almost always referred back to a specialist within the insight group” (insight manager, agency)

There clearly is some disconnect between the media research community and senior planners, who have a better understanding how the communications model of different media channel works, and the planners, buyers and vendors of space and airtime who work at the sharp end of the business. The latter may pay lip service to media neutral planning but few have any in-depth understanding. The industry may well be multi-media minded, many individuals are mentally still in their traditional silos, whether by choice (or inertia) or because the way many media agencies are structured forces them into this position.

Not all of those questioned agree and there are those who are more positive in their approach and use the Channel Planner not as a barrier but as a tool to educate people about multimedia issues and bring them out of their traditional silos:

“The Channel Planner is a great tool to get more people to become more multi-media literate and has helped to drive changes within our agency. Of course, a lot of time had to be invested to get individuals to understand the data” (insight manager, agency) “We encourage everybody to use it – from planning directors to media planners. Within our agency the more senior planners and business directors are encouraged to challenge the more junior colleagues in their teams to use TouchPoints. The pressure to learn has to come from all directions” (business director, agency) “TouchPoints 1 was a real struggle and we new that the data wasn’t used correctly. With the next data release we decided to invest in a lot of internal training. We wanted to avoid the creation of specialists and provide a level playing field for everybody” (insight manager, agency)

3.2.6.2. Software and Systems

Without doubt the most frequent problem areas mentioned concerned the software packages offered by the three main bureaux services in the UK: Telmar, IMS and Kantar/Choices (New Age Media Systems started operations in the UK a couple of years ago but at this stage does not have any clients that use TouchPoints).

The issue became so serious that the IPA commissioned an independent media systems specialist Robin Angell to assess the functionality of all three packages as well as the general service and help-desk facilities on offer. Significant differences in service levels quickly became apparent and not surprisingly there is a strong correlation between data usage and bureaux service used. For many the experience has been (and often still is) very frustrating.

“With our previous provider there was a lot of inputting and rather little output. Users were very worried about getting things right. We recently switched and our new provider offers more scope, better training and is much more user-friendly” (planner, agency)

“My staff was getting very frustrated as it was so difficult to get data out of the system, despite great efforts and help from our software provider. I think with the time element TouchPoints is more a three dimensional study and the bureaux found it difficult to accommodate the data within their standard packages. They tried to shoe-horn this three dimensional data into a system that wasn’t developed for it” (head of commercial insight, TV)

“We had some training from the IPA and our software provider but there was a need for a lot of internal training and we had to adapt our own systems to handle TouchPoints data more easily” (insight manager, agency)

“We had real problems using the data with our software provider. We discovered so many errors that our provider should have edited before the data was released to us” (head of research, TV)

“The software package has been one of our major issues and the interface between us and the data isn’t exactly easy. It takes a long time for anyone to learn how to use it correctly” (business director, agency)

“Unlike TGI or the currencies, TouchPoints is not a tool that the typical planner uses every single day and therefore it doesn’t easily become second nature. You almost have to learn it from scratch every time it is used. Our supplier has simplified things with new coding procedures and working with the data is a little easier now” (insight manager, agency)

“We had some major teething problems with our software provider. There still is a lot of complex manual one-by-one coding, which is extremely time-consuming. It takes quite a while to find your way around the data” (planning director, digital agency) “As an agency we went through an 18-24 months learning curve with a lot of training from our software provider and development of our own internal systems to generate charts” (research manager, agency)

The lack of standardisation across al bureaux sometimes causes an additional problem, when different agencies work on the same client, for example with one looking after ‘traditional’ media channel and another agency being charged with digital work.

“We also find it difficult to share data with other agencies that work on the same campaign. We can’t always exchange the codes for the target group definition when they use a different software house” (planning director, digital agency)

3.2.7 Demand for TouchPoints data

The Central Office of Information (COI) is so far the only advertiser whose head of marketing has mandated the use of TouchPoints data and thus helped to encourage many agencies to subscribe to the data. Gradually agencies began to us TouchPoints analysis when working on other clients’ briefs and especially when pitching for new business it the survey’s popularity started to grow as subscribers slowly learned how to analyse the data.

“Apart from COI other clients, especially Kellogg’s and Channel Five (a UK commercial TV broadcaster) are now very interested and we use the data quite regularly” (planner, agency)

“Slowly we see growing demand from clients to see TouchPoints analysis. At our agency it is a must to use the data in all client pitches and for approximately a third of our current clients we use TouchPoints quite regularly” (research manager, agency) “What we learned by working with TouchPoints helped us to move up the food chain with a number of clients and also in discussions with offline agencies. I would not say that clients or potential clients are expecting to see TouchPoints data but it certainly helps to include it” (planning director, digital agency)

“Many advertisers still don’t really care – they expect their agencies to provide insights and trust the agency uses a reliable source. They don’t really care which survey is used. As to media owners’ sales presentations I never really see them playing TouchPoints data back to our planning group” (business director, agency)

“We use the data quite regularly now. The big advertisers, such as Unilever, are beginning to have higher expectations that TouchPoints data is included in our presentations, but it has at times been a slow process” (insight manager, agency)

“There are growing expectations in the market place and TouchPoints is well received by advertisers, who often haven’t seen the data, even from their own agencies” (head of research, TV)

“The way the data is presented by our agency when we have our annual review of activities is quite easy to understand and clear. But I also see a fair number of sales pitches from media owners and the inclusion of TouchPoints data is not very common” (head of marketing services, advertiser)

It is certainly true that awareness of TouchPoints is high, but usage levels are still more modest – so whereas some subscribers now use the data regularly, even enthusiastically others are still much more cautious.

“The truth is that if I do not include any TouchPoints charts in our sales presentation I do not get media planners questioning their absence or requesting data” (ad marketing executive/media owner).

“Today TouchPoints is no longer a novelty – everybody knows the survey. We currently do not subscribe and the fact that we don’t use any data does not create problems with agencies or clients” (head of research, media owner)

Part 4: Other Consumer-centric Measurement Initiatives

Many media currencies have long suffered from a lack of demographic and product usage data - TV audience measurement especially could offer more and more detailed commercial ratings in many markets but generally lacked product usage or brand preference data required by many advertisers. Fusing such data (often from TGI-type surveys) with the TV currency data has already a fairly long history. In the UK the TGI/BARB fusion has now been practiced for around 20 years. Even in emerging markets, such as Brazil, Argentina, Chile, and Peru IBOPE have started to offer fusion of their TAM people meter system with the local TGI database.

In fall of 2005 the World Federation of Advertisers (WFA) in Brussels published its Blueprint for Consumer Centric Holistic Measurement. The WFA strongly believed that advertisers’ mass marketing model was losing its effectiveness and needed to evolve; with measurement systems required to leave their single media silo approaches and accommodate multi-media measurement. In particular “advertisers (were) unable to measure and track target groups’ holistic and synergistic multi-media behaviour. New technologies to enable fast and low cost data collection are not being used”, so the document.

The WFA singled out a number of initiatives that fulfilled at least some of its Blueprint objectives, including Project Apollo (US), TouchPoints (UK), and Project Cross Medias (France). Some of these attempts have been abandoned, others have by now developed beyond their early pilot stages. Interestingly, many of these studies show that traditional media reaching large audiences still play a major role in people’s media consumption and it is somewhat debatable whether the mass-market model is really broken. The following gives a basic overview of some of these initiatives based on their current (2009/10) status. The list should not necessarily be interpreted as fully comprehensive as this would go beyond the scope of this project.

4.1. Italy: Eurisko Media Monitor

The Eurisko Media Monitor (EMM) has so far published results in 2006, in 2007 and a full scale, near-continuous survey was launched in March 2009. The 2006 study was covered TV, radio, print, the internet, and direct marketing, in 2007 exposure to supermarkets and outdoor was added based on respondents' declarations.

EMM uses a sophisticated personal meter passively identifying TV and Radio by sound recognition, and recording voice declarations of the interviewee about his/her use of the other media (internet, direct mail, etc) and the meter’s bar-code scanning capabilities are used for print readership. The latest version of the device, called ‘Dialogatore’ has all the features of

the previous, plus a touch screen. This enables questions to be sent directly to respondents to be answered in real time.

In future there will be seven waves of interviews per year (28 days per wave) and results will be released 2-3 times a year. A sample of 1,000 adults (aged 14+) is recruited seven times a year respondents are invited to participate for 28 days.

EMM provides minute-by-minute information, either passively (TV, radio) or by answers to questions on the touch screen, or else by scan/voice declaration. The media channels covered are TV, Radio, Print, Outdoor, Internet, Cinema, In-Store, and postal direct marketing. Demographic criteria are collected during the initial interview but consumption of goods and services, other context information (location, with whom, mood, attitude) are not as yet collected.

Support is coming mainly from large advertisers (Bayer, Coca Cola, Fiat, ENI, Kraft Foods, L’Oreal, P&G, Unilever, etc) and the main media agencies. Support from media owners is patchier.

Unlike TouchPoints EMM uses a single source approach. It could theoretically also be used as a ‘hub’ although so far it has not been used to fuse data with other sources of information.

4.2. France: Le Project Cross Média

A comprehensive, hub style initiative, Le Project Cross Média, has been tested in 2008. The new survey is now based on 10,000 respondents and data is collected in two waves per year.

The survey is conducted jointly by Médiamétrie, (the commercial media research venture owned by advertisers, media companies, and agencies) in partner-ship with AudiPresse (Print audience survey) and Affimétrie (Outdoor audience survey).

The study's purpose is to be a hub to which the various sources of information about the other media will be gradually connected, to form a single integrated database. The plan in 2009 was to report results annually and to gradually integrate other media (outdoor is likely to be the first) and information about consumption of products and services (such as TNS’ SIMM and Nielsen HomeScan data).

The universe is the adult population (15+). Unlike TouchPoints the hub information is collected through a single interview (i.e. not through a multi-day panel) over two waves of interviews per year. The hub does not contain time-budget data and information is focused on consumption habits of the main vehicles within each media channel (TV, radio, dailies,

magazines, cinema, internet). Media vehicles are defined as TV channels, TV programs, radio stations, magazine and daily newspaper titles, internet sites and portals. Context information (where, with whom, moods and attitudes) is not collected, but questions are asked about the person's areas of interest, sports activities, holidays, music, videos, computer/video games. Demographic and geographic information as well as usage of communication equipment are also recorded.

The fusion of the hub with the various currencies results in a virtual database of 62,000 individuals which have the same socio-demographic characteristics as the 10,000 original hub respondents. The French cross media survey does not publish total contacts achieved with a cross media schedule, only the unduplicated net reach and net contributions by channel.

4.3. Germany: Mindset

In 2007 Germany experimented with a TouchPoints-type approach. Mindset was initiated by Mindshare’s Frankfurt office and supported by TV sales house IP, publishing company Burda/Focus, and outdoor specialist Jost von Brandis. MindSet had the potential to be a hub for fusion with other sources but has not been used for this purpose.

Mindset's main purpose was to describe the activities, including multi-media behaviour (simultaneous or exclusive) of people, along with information about their situation (where, with whom, how the person feels) and their exposure to advertising. MindSet was financed by Mindshare, IP Deutschland and other Print publishers. The fieldwork was carried out by Research International during fall of 2006, using a PDA device, which 1,923 respondents were asked to carry for 72 hours combined with a detailed face-to-face interview.

According to Mindshare a further release of Mindset for 2011 is under discussion. In 2009 the agency tested the use of Smartphones as a data collection device for children aged 6-13, who were given the device to collect data about their media usage and other activities over three days.

4.4. Poland: Touchpoints

In 2009 AGB-Nielsen Media Research piloted a hub-style survey for advertisers, media agencies, and media companies with approx 150 respondents. The concept is based on an adaptation of the UK TouchPoints survey. If successful this initiative will be a good demonstration of the "scalability" of the TouchPoints concept, i.e. its ability to be adapted to smaller advertising markets.

The fieldwork is carried out by Millward Brown SMG using a pre-programmed mobile phone with PDA capabilities, which respondents had to carry for seven days. Additional data was collected via initial face-to-face interview.

If necessary funding targets can be achieved a full TouchPoints survey is planned for 2010 with a total of 4,000 respondents

4.5. Australia

The Australian market offers two services that can be considered, at least partially, to fulfil the WFA’s Blueprint requirements. Both are private, commercial initiatives of leading research companies.

4.5.1 Panorama

Panorama basically consists of hub plus fusion results from three sources:

• A hub survey: currently conducted online over 40 weeks providing information on print readership, cinema attendance, use of the internet and time spent out-of-home. The survey also provides information on the consumption of broad product/services categories, as well as attitudes and lifestyle questions. The Universe is adults (14+) and the sample size approximately 22,000 per year.

• The traditional TAM results, from the OzTAM (conducted by AGB-Nielsen Media Research) people meter panel

• The Nielsen Media Research Radio Diary Sample.

The results of the fusion are provided by Nielsen Media Research to clients ten times a year. The hub data is a rolling annual average; the TV fusion data are averages of the last fourweek ‘survey’ from OzTAM; and the radio data is taken from the most recent radio survey. The official ‘currency’ for print readership data is provided by the Morgan Survey and therefore not available for fusion and Nielsen use their own data for print readership. The methodology used puts Panorama into the hub + fusion category although the hub provides considerably less information than the British TouchPoints study as it does not use a time-budget approach, nor does it provide information about context and mood/attitudes at any specific point in time.

4.5.2 Morgan Research

Morgan Research is a single source study, based on face-to-face interviewing coupled with a traditional ‘pen and paper’ diary. It is run as a traditional commercial venture, with limited input from industry bodies. The survey provides the currency for readership data but also collects

data on the general use of radio, TV, online (duration and type), cinema, direct mail, plus the consumption of a vast range of products/services, attitudes and lifestyle. The annual sample is 55,000 and respondents are invited to keep a one-week diary. Results are delivered four times a year, based on rolling annual averages.

4.6. Japan

The Audience and Consumer Report (ACR) is conducted by VideoResearch, the same Company that provides TV audience measurement from people meters as well as surveys for newspapers (J-Read) and magazines (Magascene). ACR as a consumer-centric, holistic measurement initiative consists of the combined results of two methodologies applied to the same sample:

• A classic TGI (Target Group Index) Survey, providing information about the use of various media, purchasing behaviour of goods and services, demographic, lifestyle and attitudinal indicators are also covered. The survey is conducted in seven major urban areas covering roughly 50% of the population aged 12-69. The sample size is 8,700 per year.
The media covered are TV, radio, newspapers, magazines and internet. Additionally monitored are the use of transit systems, shopping habits (department stores and supermarkets), leisure facilities, and amusement parks. Consumption of some 200 categories of goods, products, and services is measured.

• A ‘time budget’ style panel survey is conducted on a sub-sample of the TGI survey. Currently 1,200 respondents from the Tokyo area are invited to provide information on their activities, for seven days, including use of media, and some indicator of context (location, alone or with others).

• The Panel component seems to have the potential to become the ‘hub’ of a larger cross media measurement system, enabling the fusion with other surveys and proprietary research, similar to the IPA TouchPoints’ approach in the UK.

4.7. Switzerland

At the time of writing this report Publica Data and Mediapulse, who are responsible for the TV and radio audience measurement in Switzerland published the results of a new time-budget study in Switzerland. It is not clear whether this study could be used as a hub survey for future currency integration.

4.8. Denmark: Mulit Media Mennesket

In 2009 TNS Gallup, which provides audience measurement of almost all major channels went ahead with a multi-media survey inspired by and closely modelled on the IPA TouchPoints study in the UK. RSMB was selected to undertake the data integration. The Hub data is based on 2,000 electronic diaries of Danish adults aged 15-65. The Integrated Planning Database brings together data from Gallup TV Meter, Gallup Radio Meter, the Index Danmark/Gallup (print, outdoor, direct mail) and the Foreningen af Danske Interaktive Medier (FDIM) Internet survey. The survey went life at the end of 2009.

Part 5. Conclusions and recommendations

This investigation managed to identify a number of issues that had an impact on the usage and acceptance of TouchPoints in the UK. Some of these are more significant than others. What is quite apparent is that the most critical issues that may have had a negative impact on TouchPoints have little to do with the survey itself and are largely outside factors.

5.1. Conclusions

• Overall, TouchPoints has been well received. Initial expectations may have been unrealistically high but after its second data release the survey started to gain traction again.

• One of the most critical issues turned out to be the software support that end-users often did not get from their specialist bureaux. A failure to provide an agreed minimum service level and a lack of accepted standards resulted in serious barriers and discouraged many to use the data to its full extend.

• There is no serious concern or criticism from end-users with regards to the Hub data. Some of the improvements that the IPA introduced with TouchPoints 2 have worked and overall there are no critical issues at this stage.

• In the UK market few experts, if any, voice any principal opposition to the practice of integrating data from two or more different sources. Most people are realistic and accept the trade-off between the granularity of silos and the added benefit of multichannel planning delivered by fusion. Until someone can deliver an alternative for the same price this is unlikely going to change.

• Ideally, the Hub sample should be enlarged but this has significant budget implications and does not seem to be a top priority for current users. As a minimum requirement thought better guidance as to the actual sample size, its impact on the robustness of the data, and clearer screen prompts should be made available.

• Frequency of data release has not been a major issue for TouchPoints 1 or 2. With the future inclusion of an online audience currency demands for more regular updates will grow louder.

• Digital and online channels had no accepted audience currency measure until earlier this year, when UKOM launched the first release of the expanded Nielsen data. The jury is still out how well UKOM data will integrate into the Channel Planner but for the market it is important to bring the online world up to currency level.

• Being a multi-media consumer (as most of us are these days) is one thing, truly understanding the implications of multi-media usage for planning purposes is quite another. Knowledge levels are not as high as many in the industry may have assumed; inertia and a resistance to embrace new ideas are more pronounced than initially expected. Thos who persevered soon realised that there are no

insurmountable challenges but overall the learning curve has been steep and acceptance slow. A clear and consistent training programme, whether organised internally or via relevant industry bodies is an absolute must to drive understanding and usage.

• Demand for TouchPoints has been growing, albeit slowly over the last five years. There is still more push by champions rather than pull by the marketplace and demand for the data is probably a little lower than some might have hoped. It is important to remember that new initiatives often take time to gain general acceptance. The TGI survey is today a standard device in any planner’s toolbox - it took more than ten years before it was embraced by the industry in the UK.

• The overall success of TouchPoints is reflected in the growing global interest in the ‘Hub plus Fusion’ concept, and similar initiatives are now either being tested or already fully implemented in a growing number of markets.

5.2. Key lessons and recommendations:

• There is a danger to get bogged down in too many technical details at an early stage. The often-posed question 'How good is a fusion?' is the wrong question. The key issue to ask is 'whether the use of fusion provides something important that is otherwise not available?'

• It is important to produce the new tool first, but marketing, sales, training and education programmes, as well as the necessary resources need to be part of the whole process. For end users there is much more involved than simply signing up to another survey.

• If the ambition is to generate more insight how consumers use media throughout the day and across the week a well-designed survey that combines a time-budget study with additional data, such as the TouchPoints Hub survey will be sufficient and in itself provide a rich source of knowledge and understanding

• If the ambition is to bring together all platforms then the use of the Hub survey as a vehicle for integrating the currencies, data fusion along the line of the TouchPoints model is a practical and achievable solution. As long as everybody understands the inevitable limitations and as long as there is a desire to work on overcoming any limitations CIMM should go ahead. Don’t waste time pondering about Gold Standard or Holy Grails – that can be reserved for learned papers at conferences.

• Do not let the assumption that there are limitations stop you from embarking in this journey. Nothing is perfect – if want perfection stop using the current currencies immediately.

• There needs to be a trade body or industry organisation that takes a very pro-active role in promotion and consistent education. If left entirely to their own devices some users will become frustrated and even give up.

Part 6: Selected Bibliography

The topic of data integration has been included in each for the Worldwide Readership Research Symposia (WRRS) since the Salzburg Symposium in 1985, when ‘intermedia comparisons and fusions’ made their first appearance. All conference papers are available free of charge at www.readershipsymposium.org and can be sourced by author, title, or topic. The topic has also been presented at numerous ARF, ESOMAR, and other conferences during the past ten years.

Specifically for TouchPoints a wide range of background documents, briefings, and case studies can be found on the IPA/TouchPoints website (www.ipa.co.uk/Content/TouchPoints-Site-Home). Most documents on this site are freely available to non-members. The most comprehensive bibliography on the topic of data integration has been compiled by Kantar’s Roland Soong. His ‘Data Fusion Bibliography’ lists over 1,000 papers and articles from 1950 to 2006 and can be found at www.zonalatina.com/datafusion.doc

Papers listed in italics refer directly to the UK TouchPoints project.

Antoine J (1985): A Case Study Illustrating the Objectives and Perspectives of Fusion Techniques, Salzburg Symposium 1985

A report on the fusion of the 1984 French press and broadcast surveys and a discussion of the major learnings with extensive tabulations.

Baker K (2006): Development of the TouchPoints Planning database – An Appraisal IPA Website

The paper gives a detailed appraisal of the general issues concerning data fusion and a description of the various fusion methods used to integrate each currency

Baker K (2007): Data Integration Methodologies in Market Research – an Overview International Journal of Market Research, Vol 49/4 (2007)

Battais L (2009): Measuring the Audience of a Medium beyond its Original Form ESOMAR Multimedia Measurement 2009

The paper describes the latest French initiative to provide a multi-media audience measurement tool, using a similar approach to the UK TouchPoints model and its Hub survey.

Bedwell R (1991): Fusion – Britain’s Latest Experience WRRS 1991

The paper reviews the fusion of a static (TGI) and a continuous (BARB TV ratings) database. The paper covers various issues, including regression to the mean and finds that whilst overall results were encouraging purchase decisions that are related to lifestyle and/or attitudes can be more difficult to predict.

Beeftink B (2007): The IPA TouchPoints Initiative – in Practise WRRS 2007

A review of some of the practical findings and insights from the first TouchPoints Survey from 2006

Carpenter R, Wilcox S (1995): Data Fusion in the British National Readership Survey WRRS 1995

The paper describes an experiment where the annual NRS sample was split in two and magazine readership data transferred from one half (the donor survey) to the other (recipient survey), thus allowing direct comparison between ‘fused’ data and the original, actual data (fold over test). Average issue reach and duplications for the

fused magazines were very close to those in the original donor survey but some calibration was required to bring AIR scores exactly into line.

Collins J, Doe P(2009): Integrating Television and Print Audience Currencies with Consumer Behaviour Data WRRS 2009

The paper discusses two fusion projects – the integration of NPM television measures into MRI (for target evaluation against MRI’s behaviour and attitudinal measures) and the integration of MRI consumer, print and other media measurements into NPM (for television analysis against non demographic targets). The fused databases are described as accurate and actionable

Dodd M, Doe P (2009): Three-screen viewing behaviour complements traditional TV Admap, February 2009 (pg 18-20)

The article describes how American consumers are using TV, internet and mobiles, based on data from meter panels, single-source measurements, and data fusion.

Ephron E (2002): The Promise of Fusion Admap, December 2002

The paper gives as short review of one of the first US fusion projects (MARS/NTI)

Frankel M, Baxter P (1988): Fusion, Integration, Ascription, and Imputation WRRS 1988

A review of the status and direction of various statistical techniques and fusion algorithms and their impact for magazine readership measurement

Ford J, Perjés T (2009): Measuring the Combined Audience of Multi-Platform Titles WRRS 2009

Review of a large fusion project in Hungary using a 60,000-strong online panel (measuring online readership) as a donor base and the 16,000 sample national readership survey as a recipient base, preserving the ‘gold standard’ of both individual currencies.

Galdames P, Balmaceda O, Carranza E (I2005): Estimation methods for media audience duplication ARF/ESOMAR Worldwide Audience Measurement 2005

Hussein I (2006): IPA TouchPoints Admap, July 2006, (pg 34-36)

IPA (2006): The IPA TouchPoints Initiative – Handbook for Users

A detailed user guide for TouchPoints 1, including an overview of the TouchPoints integration process by Steve Wilcox (RSMB)

Jephcott J, Bock T (1998): The application and validation of data fusion Journal of the Market Research Society, 1998, Vol 40/3 (pg 185-205)

An overview of a wide range of fusion methodologies and validation methods, including a summary of the challenges facing survey designers

Licastro G (2009): The Eurisko Media Monitor ESOMAR Multimedia Measurement 2009

The paper describes a single source approach using mixed mode data collection (media usage, product purchase) and the reaction in the Italian market after the first data release.

Loerke S (2008): Is there Life after Apollo? The Advertisers, December 2008

The article summarises the current global state of holistic and consumer-centric media measurement.

Mareck M (2007): TouchPoints – A step toward a Multimedia Planning Nirvana? PubliGroupe Forum 2007 (Zurich, Geneva)

Mareck M (2008): Apollo We Have a Problem Research World, May 2008 (pg 8-11)

Marks R (2005): Time as the Unifying Metric ARF/ESOMAR Worldwide Audience Measurement 2005

The author argues that time-budget studies are the way forward for audience measurement in the digital, multi-media age, providing both deeper insight into how consumers use media and a central platform, or hub to fuse individual currencies.

Napoir D, Mattlin J, Ivins B (2005): Calibrated Fusion Evaluation WRRS 2005

The paper provides a short overview of the integration of MRI and comScore data in the US and the importance of pre-fusion calibration.

O’Brien S (1990): Target group ratings – fusing of BARB and TGI Admap Feb 1990, pg 38-40

Robinson L, Beck J, Wilcox S(2006): Creating the Missing Link – the IPA TouchPoints Initiative Esomar Worldwide Multimedia Measurement 2006

A detailed paper reviewing the history and development of TouchPoints, the data collection methodology and the fusion process.

Robinson L, Turner D (2007): Delivering the Dream – the IPA TouchPoints Initiative ESOMAR Worldwide Multimedia Measurement 2007

A review of TouchPoints1 including a number of examples and case studies and a summary of changes planned for TouchPoints2

Robinson L (2009): The IPA TouchPoints Initiative – its Effects on the Market Place and its Future Plans ESOMAR Multimedia Measurement 2009

A summary of the learnings from the first two TouchPoints studies and the implications for TouchPoints 3

Scheler H, Wiegand J (1985): Experiments in Fusion in the Official German Media Research WRRS 1985

The paper describes a series of experiments to develop an alternative to single source audience data for print, radio, and TV consumption. The aim was to build a partnership model, whereby each of the three channels is measured separately and results later ‘merged’ for intermedia comparison. As a result the technical committee

of AG.MA decided that in principle the fusion process as tested worked well and should be introduced into the new model.

Sharot T (2007): The design and precision of data-fusion studies International Journal of Market Research, Vol 49/4 (2007), pg 449-470

Soong R, de Montigny M (2003): Fusion-on-the-Fly for Multimedia Applications WRRS 2003

A discussion of various fusion methods with focus on how to optimise fusions for specific target groups ‘on the fly’ compared to syndicated fusions such as the Nielsen Television Index (NTI) and the MARS (pharmaceutical) surveys. The authors predict that customised fusion-on-the-fly which generates multiple fusion databases is likely to cause confusion as syndicated data fusion provides more consistency for all users.

Soong R, de Montigny M (2003): Does Fusion-on-the-Fly Really Fly? ARF/ESOMAR Worldwide Audience Measurement 2003

A similar paper to the one given at the Boston Readership Symposium but with additional tabulations.

Soong R, de Montigny M (2004): No Free Lunch in Data Fusion ESOMAR/ARF Worldwide Audience Measurement 2004

The authors discuss various integration techniques and conclude that there is no single method that works best in all situations and conclude that for any specific problem the best approach is to develop the most appropriate solution.

Soong R (2006): The Data Fusion Bibliography www.zonalatina.com/datafusion.doc

One of the most comprehensive bibliographies on the topic of data integration with approx 1,000 listed books, articles, and conference papers from 1950 - 2006

Thadani R, Sinha A (1988): A Study on the Validity of Data Fusion: the Indian Experience WRRS 1988

The paper describes an experiment fusing readership data with television viewing data. Conclusions include that results from fusion are very similar to those of single source data, but tend to work better for higher titles/TV programmes with higher audience levels and that it is difficult to replicate results for marginals.

Video Research Inc (Japan): Audience and Consumer Report www.videor.co.jp/eng/products/marketing/acr.html

English language summary of this Japanese Media/TGI survey

Walsh P, Zack B (2003): Validating Telmar’s Multibasing Technique WRRS 2003

The paper describes a fold over test using MRI’s 2002 Doublebase and compares results using MultiBasing and Weighted Profile Matching. The paper includes a short summary of other data integration techniques

Walsh P (2001): Multibasing – Data Integration without Regression to the Mean WRRS 2001

The paper argues that traditional fusion methods using respondent level survey data rely too much on demographics, which often account for too little in variance for many product categories. The author proposes the use of clusters of respondents to identify

relationships that are based more on respondents’ interests than their demographic characteristics.

Wilcox S, Johnson H (1987): Multi-media Reach and Frequency Analysis WRRS 1997

The paper describes various objectives of mixed media R&F analysis based on examples of mixed print plus TV schedules, using the fused TGI/BARB database.

Wilcox S, O’Sullivan N (2007): TouchPoints – Integration for Multi-Media Planning Vienna Symposium 2007

Summary review of the UK TouchPoints Initiative and its different elements and the actual fusion process

Withers H, Dow H (2007): Canadian Evaluation Fusion versus Single Source WRRS 2007

An experiment by the Canadian Media Directors’ Council tested two fusion techniques and concluded that neither succeeded in fully replicating the original currency relationships.

World Federation of Advertisers (2005): Blueprint for Consumer Centric Holistic Measurement,
WFA 2005, (www.wfablueprint.org)

Appendix 2a and 2b: Questionnaires

TouchPoints 3: Self- completion questionnaire

www.ipa.co.uk/Content/TouchPoints-Site-TouchPoints3-Self-Completion-Questionnaire

TouchPoints 3: Electronic diary (PDA)

www.ipa.co.uk/Content/TouchPoints-Site-TouchPoints3-eDiary-Questionnaire