THE TOP 10 INSIGHTS
From the WARC Conference on Advertising Research
June 5-6, 2001 New York
1. It is more important to measure feelings than awareness and comprehension. At least that was the position of most speakers at this conference. The point was first made by Chuck Young, Ameritest CEO, and then repeated by Robert Heath, UK author of "The Hidden Power of Advertising" being published this month. Heath went the furthest with a lucid rationale for virtually ignoring comprehension. Today, people do not expect much difference between brands. So they pay little attention to claims in ads. Further, they do not even try to make rational purchasing decisions. They buy on feelings and intuition, which can be built with the impressions made from low involvement exposure to advertising, something that is not tapped by the typical pre-test measures of ad recall and main point feedback under conditions of forced exposure, but it is something that can be measured by more subtle diagnostic measures of reactions and feelings about the product and the advertising. They are stored in Episodic, as distinguished from Semantic memory, two frequently-heard buzz words rapidly replacing the old right brain / left brain concepts, but with essentially the same meaning. Millward Brown’s Nigel Hollis said TV, radio and cinema involve episodic memory while magazines, newspaper and the internet are semantic, so they should be tested differently.
2. Persuasion scores may or may not be the answer. Pre-testing firms that put more emphasis on awareness and comprehension (ARS and Ipsos-ASI) made a vigorous case that whatever is driving sales, they capture it with their measures of the shifts the advertising causes in brand preference. Tom Tindle (ARS) told of over 2000 cases they have complied to show their persuasion scores predict market performance. David Brandt showed how Ipsos-ASI combined Recall and Persuasion scores to provide their most predictive measure that across 31 ads explained 77% of the variance in sales impact. It appeared everyone at the conference would agree that pre-testing usually helps, but lower attendance at the pre-testing session suggests researchers remember the lukewarm endorsement pre-testing got from the two industry-wide studies during the last decade: the ARF’s Validity Study and IRI’s Adworks 1. Heath summed it up by saying pre-testing is always an option, but tracking is virtually mandatory.
3. Unconventional pre-testing helped Colgate overtake Crest. Uwe Munzinger described his "Ad Plus" procedure from Icon that was recommended to Colgate by overseas colleagues. They show a clutter of commercials imbedded in program material and combine traditional measures of awareness, communication and persuasion with a set of diagnostic measures designed to identify the images and impressions left by low involvement processing, a function similar to those advocated by Ameritest and Heath. The result was the selecting and perfecting of the brushing sound that people keep hearing all day to convey long lasting protection.
4. Test lots of alternative ads. This may have been the one topic on which there was universal agreement. Tom Tindle of ARS, reviewing the ad development process, cited an Irwin Gross quote from 1967 to emphasize this is a point that has long been recognized. Developing and testing alternatives is a key to successful advertising. In the Colgate case history, Jim Figura said introducing Colgate Total eight years earlier overseas was a major advantage. (It took that long to get FDA approval for a US introduction.) They had experience with a pool of overseas commercials to use in developing their eminently successful US commercials. Randy Kosloski described how switching to lower cost online surveys was making it possible to pre-test more print ads at General Mills.
5. The most effective advertising reaches people just before they buy. Erwin Ephron stated this as the central principle of his "recency planning." Living up to his reputation as the ultimate authority, he gave what was hands-down the conference’s most entertaining presentation in the form of the Gospel according to Erwin, complete with hymns. He cited the new evidence from Adworks 2 to show the first exposure to advertising has the biggest effect. Sitting only two feet away was Conference Chair Mike Naples, author of the most authoritative book on Effective Frequency back in 1979 that said it was probably the third exposure that had the biggest effect. With only the slightest hint of sarcasm, Ephron went on to cite the new evidence showing advertising reach builds fastest at the lowest TRP levels and continuous advertising is more effective than flighting. His conclusion: lower levels of continuous advertising throughout the year will reach more people at the time they buy, and produce more sales per GRP, than the usual bunching up of GRPs into sporadic flights of heavier advertising.
6. Others say "Not so!" Advertising’s effect can last a long time. Bruce Goerlich of the Starcom MediaVest Group presented data from SMG’s Brand Library showing advertising’s effect is more accurately reflected by adstock models: those based on measures of advertising’s half-life, the point in time when advertising has had half of its eventual effect on sales. He said Ephron’s recency approach assumed a very short half-life of a week or less, and he agreed if that is the case, you need continuous advertising. But 90% of the brands his firm had studied had a half-life of two weeks or longer. Over 50% had a half-life of 6 weeks or longer, and much less need for continuous advertising. Earlier, David Brandt of Ipsos-ASI had cited John Philip Jones data as showing essentially the same thing: the adstock effect is real and can last a long time.
7. Marketing mix models and tracking provide the best measures of advertising impact. This appeared to be another issue on which there is virtually complete agreement. Tracking shows what actually happened when people were exposed to the advertising under normal, real-world conditions, and modeling is providing much better estimates of the change in sales that is actually due to advertising. Ron Bates from Kraft confirmed a point this author has been making for years. Early models that only used GRPs to represent advertising didn’t fit. They didn’t show advertising accounting for much sales because they assumed bad advertising was contributing as much to sales as good advertising. On the other hand, detailed scanner based price data showed clear relationships to sales, so for years Kraft was cutting back on advertising, and doing more of what they could prove was effective. Now they are incorporating data on the quality of their advertising derived from pre-testing and continuous tracking studies. This is leading to better evidence that advertising has an effect, and growing ad budgets at Kraft. Bates sees it as a continuing process. Some at Kraft are still reluctant to increase ad expenditures to levels research now says are optimal. But he quickly adds those who have, have faired the best.
8. Recognition is the best tool for showing what people notice and what they ignore. The Ameritest procedure has people sort out photographs they recognize as being from a commercial they were shown earlier. It shows the parts that made an impression far better than verbal questioning. Robert Heath, as part of his premise that it is the images and impressions that do the selling, said recognition is the tool to use to see if the process is working. It will show if the people that noticed the advertising actually have the images and impressions you are striving to convey.
9. Descriptive models are better than predictive models. This was also alluded to by a number of speakers after the principle was first articulated by Kraig Schulz of Mindshare. He said a good descriptive model is constructed to see if it confirms some theory, so it is primarily a tool to find out how the process works. But he emphasized the theory is confirmed by showing the model could predict what happens, but prediction was not the purpose of the model. When it is, we run the danger of getting statistically advanced models that tell us little about how marketing works. There was also considerable agreement that advertising quality is best reflected in models as a multiplier effect: sales per 100 GRPs.
10. Long term effects of advertising are many times greater than short term effects. Nigel Hollis of Millward Brown presented data showing they were 6 times greater. Mike Hess of PDI/Knowledge Networks reported long term effects 2-3 times greater. The PDI modeling of short and long term effects was based on panels built from over 5 million households that have joined frequent shopper programs. He found those who first tried the product because of deals or coupons tended not to buy again until deals or coupons were again available. Those who buy for full price were much more likely to buy the brand again, and to pay full price when they did. The result of a study of 12 brands in 11 categories over 78 weeks showed the short term effect of advertising accounting for 7% of sales, but 19% after the eventual long term effects are allowed for.
10.5 Make the brand part of the culture. That was the strategy advocated by Rupert Newton of M&B. Faced with the product parity and instant imitation described by others he described how Channel 4 in the UK established itself with Big Brother, one of the initial ventures into voyeurism TV. John Shaw, head of planning at Wieden + Kennedy, agreed saying great creative work affects popular culture and cited the agency’s "Just do it" for Nike as an example.
Papers from this conference are available from WARC at (202) 778-0680 Don Bruzzone, June 2001
Bruzzone Research Company · 2515 Santa Clara Avenue · Alameda, CA 94501- 4692 · (510) 523-5505 · www.Bruzzone-Research.com ·