But will they buy it?

Editor’s note: Ed Erickson is senior vice president, regional manager at Socratic Technologies, Inc., Chicago.

Some of us can remember recruiting hundreds of consumers to hotel ballrooms to watch commercials projected on a big screen to provide researchers with feedback via a handheld dial known as an audience meter. Economic downturns, research budget cuts and a host of other forces conspired to curtail the use of this useful but expensive form of research.

Today, we are seeing a resurgence of interest in the technique. The increasingly widespread adoption of broadband Internet access has prompted an increase in the use of “perceptometers” which track moment-by-moment reaction to television ads using Flash or streaming video (Figure 1). This use of technology has eliminated most of the cost and logistical issues associated with the old central-location studies, making it a viable ad testing option. Since we first demonstrated these techniques publicly at the ARF Week of Workshops in 2000, a number of varieties of this online methodology have emerged.

Along with this rise in the use of the technique comes the obvious question: What do these curves mean? Looking at a curve move up and down, one can quite easily pick out the parts of the ad that a viewer did or didn’t like. However, most would agree that the measure of an ad’s value isn’t whether or not people liked it, it is whether or not people took the desired action as a result of seeing it.

Asking the simple question “So what does a curve that looks like that mean?” led us to the line of research discussed here. Admittedly, we’ve related self-reported purchase intent to the curve profiles, not actual purchase behavior, but the relationships we’ve uncovered are important nonetheless.

The research

We examined data collected on 14 different television ads in mature consumer product and service categories. Each ad was for a well-understood, low-to-moderate involvement product or service and was comparable in terms of the viewer’s level of effort required to process the information. For each ad, we collected second-by-second spontaneous reaction, followed by an overall rating for purchase intent, relevance and believability. In all cases, respondents were screened to be in the relevant market, so any effects of surveying those with no interest in or use for the product or service advertised were eliminated.

We found that the relationship between the curve and the summary measures of purchase intent, relevance and believability could be best described using the average levels and slopes of three segments of the spontaneous reaction curve as well as the overall average level and slope of the curve (Figure 2).

Not surprisingly, purchase intent, relevance and believability are all strongly correlated. However, it isn’t reasonable to expect that a judgment is made about the three items simultaneously but rather in sequence. We believe the sequence of processing the ad is relevance, believability and then purchase intent.

The relationships uncovered resulted in three important findings:

1. Viewers make an almost immediate decision about the relevance and believability of an ad.

2. The final third of an ad is the most influential in terms of ultimate judgment of purchase intent, relevance and believability.

3. Women are more likely to process more of the information in an ad than men when making purchase decisions and evaluating relevance and believability.

Snap judgments

We examined the spontaneous response curves over the first third of an ad (five seconds for 15-second ads, 10 seconds for 30-second ads) and found a strong relationship between the level and slope of the curve and all three of our summary measures. Higher average levels and the more positive slopes of the curves indicated greater relevance, believability and purchase intent. Figure 3 shows how the initial reaction to the ad relates to the summary measures. Purchase intent, relevance and believability relate similarly to the curve profile, so only a single figure is shown for brevity.

The data clearly show that viewers are making snap judgments about the ads. Analysis of about 1,500 individual curves showed that they either rose at the start of the ad or they didn’t rise at all. We did not see any examples of a high rating on any summary measure accompanied by a curve that remained low and flat for any extended period during the ad.

We did not examine specifically what about the ads resulted in the viewer’s reaction but we clearly saw a pattern across several different product and service categories. These findings suggest the need for a strong “What’s in it for me?” opening to an ad.

It is interesting to note that, while viewers seem to decide if an ad is relevant to them very quickly, the initial reaction alone has little predictive power with respect to relevance, purchase intent or believability. In fact, the ability to predict these measures improves more than fourfold when the rest of the ad is considered.

What they see last sticks with them

Our analysis of the perceptometer curves and the summary evaluations of the ads shows that the final third of the ad is the most influential on purchase intent, relevance and believability. As with the initial reaction, a higher average level and a more positive slope for the segment of the spontaneous reaction curve is associated with greater purchase intent, relevance and believability.

The average level of spontaneous reaction during the final third of the ad was the single most important measure in predicting the respondent’s summary evaluation of purchase intent, relevance and believability. The influence of the final third of the ad was between two-and-a-half and five times greater than the second most important factor.

In Table 1 and Table 2, the least influential predictor is set equal to 1. The values for the other predictors represent the number of times more influential each predictor is relative to the least influential. Items without numbers do not influence the rating.

The key finding here is that, likely due to the low-involvement nature of the products advertised, the summary judgment is very strongly influenced by the last thing the viewer sees. The lesson for marketers is to finish strong with their ads.

Men and women process information differently

Experience tells us - and academic research confirms - that men and women process marketing information differently. Women tend to process greater amounts of information and apply greater deliberation to decisions than men. We found that men’s summary measures are influenced almost exclusively by the final third of the ad, where women’s summary measures are influenced by the holistic impact of the entire ad.

The lesson we draw from this finding is that advertising aimed at women needs to build its argument throughout the ad, while ads aimed at men need to be much more focused on a strong finish.

Implications

These findings have some important implications for both the conduct of ad research and the development of advertising creative.

First, we can use the relationships explained here to determine question flow in a survey. It is a simple matter for any competent research firm with reasonably sophisticated online technology to detect and act on the pattern in the real-time perceptometer data stream. This pattern recognition can then be used to direct the line of questioning. For example, when a classic low-believability curve is detected, the follow-up questioning can explicitly address reasons for the lack of belief. Alternatively, when a high purchase intent curve is detected, follow-up questioning can be directed to understand timing and dollar value of purchases.

Second, these results provide some insight to marketers in terms of ad execution. The findings of this research clearly show the snap judgment that is made with respect to relevance of an ad. This argues strongly for ads to open with a compelling answer to “What’s in it for me?” We’ve also shown the importance of the final impression, suggesting that marketers should close with a compelling reason to take action. Finally, due to gender differences in information processing, ads targeting women should take care to build the argument for the product or service throughout the ad, while those targeting men should focus more heavily on the strong close.