Download Conjunctions and linking words and more Essays (high school) English Language in PDF only on Docsity!
University of Wollongong Research Online
Faculty of Business - Papers Faculty of Business 2017
Methodological Guidelines for AdvertisingResearch
John R. Rossiter University of Wollongong , [email protected]
Larry Percy Copenhagen Business School
Research Online is the open access institutional repository for the University of Wollongong. For further information contact the UOW Library:[email protected]
Publication Details Rossiter, J. R. & Percy, L. (2017). Methodological Guidelines for Advertising Research. Journal of Advertising, 46 (1), 71-82.
Methodological Guidelines for Advertising Research Abstract In this article, highly experienced advertising academics and advertising research consultants John R. Rossiter and Larry Percy present and discuss what they believe to be the seven most important methodologicalguidelines that need to be implemented to improve the practice of advertising research. Their focus is on methodology, defined as first choosing a suitable theoretical framework to guide the research study and thenidentifying the advertising responses that need to be studied. Measurement of those responses is covered elsewhere in this special issue in the article by Bergkvist and Langner. Most of the frameworks are derivedfrom the authors' own published work, although other frameworks are noted where appropriate. Disciplines Business Publication Details Rossiter, J. R. & Percy, L. (2017). Methodological Guidelines for Advertising Research. Journal of Advertising, 46 (1), 71-82.
This journal article is available at Research Online: http://ro.uow.edu.au/buspapers/
In the present article the authors have selected what they consider to be the seven most important improvements of a methodological nature that need to be implemented in today’s advertising research. The initial emphasis in each is on theory and on mistakes to avoid and the closing emphasis in each comes in the form of a research guideline. By way of overview, the seven methodological concepts highlighted in this article are:
- The probabilistic nature of the four buyer response steps – exposure, ad processing, brand communication effects, and brand purchase – and whether or not the steps need advertising repetition.
- The three separate sequence of responses, at each of the last three buyer response steps, that must be taken into account when planning and testing advertising.
- In measuring the ad processing step, the need to estimate the prospective buyer’s probability of paying initial attention to the ad.
- In measuring the brand communication effects step, the need to measure brand awareness as the necessary precursor to brand attitude.
- Clarification that the Rossiter-Percy Grid applies not to whole product or service categories but to the target audience’s purchase decision for a particular brand.
- The necessity of pretesting benefit claims, not benefits.
- The future of ad testing. Methodological issues in advertising research may be seen as largely a case of choosing a reasonable model or framework with which to plan the advertising research study. Almost any framework is better than none but surprisingly few have been proposed in the literature, with the consequence that advertising research, at least in the academic world, today resembles a scrambled set of jigsaw pieces that teachers and future managers are
supposed to somehow put together. In direct contrast with others’ efforts, the advertising planning frameworks in what has become known as “the Rossiter-Percy model” are the only comprehensive, systematic, and conceptually consistent frameworks available in the advertising textbook and journal article literature. Following, in the authors’ opinions, are the seven most important ones.
1. Buyer Response Steps and Repetition The first important concept to be understood in conjunction with the planning of advertising research is that advertising, to be successful, must pass through four basic buyer response steps – exposure, ad processing, brand communication effects, and brand purchase behavior or other purchase-related actions (to keep things simple, the following discussion refers to all behaviors targeted by advertising as “brand purchase”). As McGuire (1969) explains in connection with persuasion, these steps should be regarded as a chain of multiplicative probabilities , thus p(exposure) × p(ad processing) × p(brand communication effects) = p(brand purchase). The chances of advertising resulting in brand purchase are radically reduced by a low probability at any or all of the first three buyer response steps. Even the last step, brand purchase, is probabilistic since, because of possible derailment at the point of purchase, purchase cannot be guaranteed even after an ad has been effective on all prior steps. Consider, for example, the second step, ad processing. Ads that appear on informational websites, like Google or the various news sites, have a very low probability of being processed by those who visit the page because visitors with an information-seeking mindset have learned to ignore them. Let’s say that a banner ad is placed in an online newspaper visited by 10% of the population, that it is noticed and read by 5% of visitors, and that a typical 1% of these visitors are successfully communicated to by the banner ad and click through to the advertiser’s website. The probability chain is then .10 × .05 × .01 =
gone through only once, that is: exposure ad processing brand communication effects brand purchase. (b) Ad repetition prior to action – this is the scenario needed for new product advertising. The potential buyer has to cycle through the first three buyer response steps at least several times before action can take place, thus: repetition (exposure, ad processing, brand communication effects) brand purchase. This ad repetition sequence has major implications for advertising research, particularly for the pretesting of TV commercials. Although this never happens in academics’ or practitioners’ pretesting methods, informational new product commercials have to be exposed at least twice in a solus setting and about four times in a clutter setting, and transformational new product commercials at least three times and about six times in a clutter setting, before they can be properly evaluated for their ad processing probability and their brand communication effects probability. For those readers unfamiliar with the careful and specific Rossiter-Percy definitions: informational advertising addresses advertising situations where the primary motive for brand purchase is negatively originated, originating from a consumer-perceived problem or anticipated problem, so that the advertising promises negative reinforcement – called “negative” because it negates a problem – by buying this brand as the best solution to the problem; whereas transformational advertising addresses advertising situations where the primary motive for brand purchase is positively oriented, so that the advertising promises more positive reinforcement – sensory, intellectual, or social – from buying and using this brand than is promised by other brands. (c) Full sequence recycling after trial for repeat- purchase products and services – this is the type of repetition needed for established consumer packaged goods and regularly accessed services. Continued advertising is needed to keep the buyer buying the brand in the face of advertising for competing brands; thus, repetition of all four buyer response steps – full sequence recycling – occurs (exposure, ad processing, brand communication effects, brand purchase). In the established-brand scenario,
current TV commercials and print ads have to be retested with brand purchasers to see whether the brand’s ads are “wearing out” so that executional variations can replace them. (d) Recycling on the last two buyer response steps with no advertising needed – this is the ideal situation because it saves the advertiser money. In this no-advertising scenario, the current buyer of the brand remembers it or encounters it at the point of purchase and the situation becomes repetition on the final two steps only (brand communication effects, brand purchase). A very valuable purpose of advertising research is to track the post-purchase period to see what happens and whether another wave or “burst” of advertising needs to be aimed at current brand buyers. The twin problems for advertising researchers, then, are that they study advertisements without regard to the buyer response steps that ads have to pass through and they also fail to account for advertising repetition. Advertising repetition is basically a media scheduling problem, about which academics, apart from those few like Peter Danaher who consult to leading advertisers on media planning, have had nothing to say for years. Most critically, media planning theory has been neglected in journal articles and especially in textbooks, where media planning theory remains in a primitive and, given that this is where most of the advertiser’s budget goes, unsatisfactory state (referring only to the simplistic media concepts of reach and frequency and the all-too-aptly-named Gross Rating Points). The exception is again the Rossiter and Percy textbook. Media-planning frameworks introduced in the 1997 edition (see also the Rossiter and Danaher 1998 book on advanced media planning, which contains an easy-to-use CD software package) include the concept of reach patterns and a management-judgment formula for estimating the required minimum effective frequency per advertising cycle.
the rather mindless rote learning of the brand’s key benefit – is critical for an ad aimed at low-involvement brand choice, whereas the more carefully considered Acceptance responses
- spontaneous cognitive responses made during ad processing – are critical for an ad aimed at high-involvement brand choice because they could just as easily be rejection responses, pushing the probability of successful ad processing down to zero. Also, Emotional responses, both positive and negative, have to be measured very specifically – not as overall affect or liking-disliking – and they have to be analyzed as an emotion shift (a negative emotion shifting to a slight positive emotion for informational advertising, and a neutral state or slight negative emotion shifting to a strongly positive emotion for transformational advertising). The lesson here is that advertising researchers cannot be said to have studied ad processing properly unless they measure all four processing responses, analyze them at the individual respondent level rather than aggregating them across respondents as is typically done, and analyze emotional responses as requiring a specified shift. The results of ad processing feed into the next sequence, that of brand communication effects. These communication effects, and there are five of them, are called Category Need, Brand Awareness, Brand Attitude, Brand Purchase Intention, and Purchase Facilitation. (Rossiter and Percy’s textbook is the only book that has identified all five communication effects, specified their accompanying options as communication objectives, and offered single-item measures of them.) Brand communication effects are learned in parallel, not in sequence, although they operate pretty much in sequence (as a hierarchy) during the brand choice process. Advertising for a familiar brand, which would usually represent a low- involvement brand choice, needs only to reinforce two communication effects: Brand Awareness, and Brand Attitude conditional on – that is, elicited from – prior Brand Awareness. Advertising for an unfamiliar brand in a category that is anything other than totally riskless, signalling a high-involvement brand choice, has to address all five
communication effects: it has to create or remind the prospective buyer of the Category Need, instil Brand Awareness connected to the Category Need, develop a favorable Brand Attitude conditional on the prior Brand Awareness, and encourage Brand Purchase Intention conditional on the presence of Category Need, and usually has to provide Purchase Facilitation with perhaps a promotion offer and a convenient means of obtaining the product. The problems in both academic and applied ad testing for brand communication effects are that no thought is given to which of the brand communication effects need to be measured and that in testing ads for high-involvement brand choice, such as direct-response ads (with digital media’s direct-response ads joining traditional media’s direct-response ads to form what is now the largest category of advertising), the first and last of the communication effects, Category Need and Purchase Facilitation, are neglected. The third and final sequence of effects to be taken into account, and in fact it has to be considered first when planning advertising research, is the buyer stage sequence. This sequence operates for individuals most often in hierarchical order. Brand-related communication effects feed into the buyer behavior hierarchy, which consists of five stages: Not in the Market, Need Arousal, Search and Evaluation, Purchase, and Usage. One of the most common problems, virtually universal in academic advertising research, is the testing of ads without knowing what particular stage the test respondents are in. As an example of this mistake, researchers in an advertising research article published not too long ago in JCR tested ads for unfamiliar imported beers with a U.S. undergraduate college student sample that consisted primarily of underage females, most of them, of course, Not in the Market, and, not surprisingly, the bipolar brand attitude ratings that the researchers measured clustered around the neutral midpoint. Another example is advertising intended to reach brand buyers not prior to purchase but actually while they are in the Usage stage; this post-purchase advertising can reduce possible dissonance and reinforce choice of the brand for next time.
can still use the figures from 40 years ago in the first edition of the Rossiter-Percy book which were reported in index form, though easier to understand are the same figures in the 1997 book reported as probabilities. Advertisers in non-mass media – notably outdoor, directories, and in digital media – are not so fortunate. Your authors had to find needles in haystacks to get reasonable attention estimates for ads in these media, especially for the quickly diversifying forms of digital advertising. Another problem was that the attention- estimating methods for these non-mass media ads were never as clean as the day-after recall method for broadcast or the ad recognition method for print. However, for managers faced with deciding on media plans, almost any reasonable estimate of ad-unit attention differences is infinitely better than no estimate at all. The importance of ad unit differences in attention-getting capacity should be apparent from the probabilities estimated in Table 1. A 30-second TV commercial, for example, has an estimated .65 probability of being paid attention to (but this drops to about a third of that if the program being watched has been time-shifted). A 60-second TV commercial, often used to launch a new product or service, has about a .76 probability of being paid attention to and even though you don’t get twice the “bang for your buck” the audience increase might be worth it in terms of increased prospects. A 1-page consumer magazine ad, on the other hand, has an estimated .49 chance of being noticed to the extent that the consumer can later recognize the ad as being in that issue, but only among those who subscribe to or purchased the magazine; secondary or “pass along” readers have only about a 5% chance of seeing a given ad, making total circulation figures for magazine readership uninformative unless the advertiser knows the extent of primary readership. Outdoor ads, a non-mass medium form of advertising which includes indoor posters, receive a tremendous boost in attention from a stand-alone location as opposed to being placed among other outdoor ads; the attention probability advantage of stand-alone placement, .80 vs. .40 for pedestrians and .53 vs. .38 for
drivers, would in most cases be well worth the extra cost. As a final example, companies’ websites, which of course have largely replaced brochures, have a 100% probability, obviously, of having the homepage being attended to by those who voluntarily visit the site, but only about a 25% chance of the visitor going to the next page and a diminishing probability of opening subsequent pages. Table 1 about here Attention probability findings make boring reading, quite frankly, and apart from the few who are consultants to proprietary advertising research companies, academics have neither the resources nor the interest in doing this type of research. However, advertising textbook writers among academics should at least report the research that has been done , as well as emphasize the critical importance of it for advertising effectiveness (due to the fact of attention’s fractionating effect on the probability multiplication chain outlined earlier). And advertising academics who engage in real-world consulting should definitely point out these attention norms to their advertiser clients, if only because attention has such a large effect on overall advertising response.
4. Brand Awareness: The Gatekeeper Communication Effect While Attention serves as the gatekeeper before other ad processing responses can occur, Brand Awareness plays a similar role among the communication effects at the next step of buyer response. Among the communication effects, most advertising researchers skip over Brand Awareness and study only Brand Attitude , or Ab as it is commonly referred to. Consider the situation of an ad for a new or fictitious brand. The researcher will conclude that the ad they are studying has been effective if it appears to increase attitude toward the brand, usually by comparing the post-exposure attitude rating given by the experimental
response because that is the cue that should be emphasized in the brand’s advertising. To measure Brand Recognition in an advertising pretest, the researcher has to show the respondents a typical competitive brand display and asked them to point out as quickly as possible those brands that they recognize, allowing 10 seconds, a generous search time for any brand display in the real world, for this task. If this test is conducted in up-front research, the researcher can ask respondents how they recognized each brand, because that will further indicate the precise recognition stimulus or stimuli to include in the advertising. There is also the occasional situation of auditory recognition of the name when the brand name is spoken, as when ordering a beverage in a restaurant or when a friend or acquaintance in conversation is recommending alternatives. Brand Recall is the appropriate form of brand awareness to measure when brand choice is initiated prior to the point of purchase (again, see examples in the Rossiter-Percy Grid). The initiating cue here, of course, is the very first communication effect, Category Need. The brands are not available to be recognized, so they have to be recalled from long- term memory in response to the category cue. Notice that brand recall cannot, as some proprietary ad-testing company researchers seem to believe, be tested by seeing whether test respondents can recall the brand name from the ad ; it matters only that they can recall it from the category cue. Most often, what has to be recalled is the brand name so that the prospective buyer can go online or look up a directory to find out where the brand can be bought. Likely the biggest mistake made in all forms of advertising is to emphasize or repeat the brand name only , without connecting it to the product or service category. In an ad test, category-cued brand name recall can and should be tested before administering the brand attitude measure. A pre-post design is ideal here but a fair approximation can be made from an experimental-control design by comparing the two groups’ recall of the target brand at the
individual level and then recording as successful brand recall the proportion of respondents who recall the target brand in the, say, the top three when given the category cue. Brand Recall-Boosted Recognition, which occurs when the prospect has to recall the brand first and then go looking for it in a competitive display, is an increasingly prevalent brand awareness situation. Not only are supermarket and drug retailers becoming larger in their category displays but so also are department stores and clothing stores, with a flood of name brands and retailer brands for consumers to choose from. In this brand awareness situation, advertising has a much more complex task: it has to stimulate brand recall by connecting the brand tightly to the category and also has to emphasize the appropriate brand recognition cue. In measuring this compound type of brand awareness, the recognition test obviously has to come first, followed by the recall test. The researcher then has to analyze the results in reverse order and then score recognition conditional on recall. Brand-awareness measurement is nuanced and difficult, so it is not too surprising that academic researchers take the easy way out by not measuring it at all , and that proprietary ad testing and tracking services take the wrong way out by measuring brand awareness incorrectly. The Rossiter-Percy approach is the only approach that draws attention to this major advertising research problem and offers a theoretically sound solution.
5. Misuse of the Rossiter-Percy Grid The Rossiter-Percy advertising planning grid is probably the present authors’ best- known contribution to the advertising research literature. However, it is almost always misunderstood and misused. A new version of the grid that hopefully will clear up these misunderstandings and prevent the grid’s misuse is shown in Figure 2. Figure 2 about here
change or major damage to the company’s reputation, would be making a low - involvement decision when repeat-buying their usual brand. Several crucial problems in advertising research stem from ignorance or misuse of the Rossiter-Percy Grid. The first problem is that advertising researchers never specify, let alone screen for when testing ads, the brand-based target audience and therefore cannot properly classify the advertised brand as being low involvement or high involvement. (Advertising researchers need to screen for five possible target audience types as listed in the grid, which fall into either the low-involvement or high-involvement classification.) The second problem is that without specifying the brand-based target audience advertising researchers cannot identify those communication effects that have to become communication objectives for the campaign. (A low-involvement advertising situation calls for the correct type of brand awareness and either brand attitude increase or brand attitude maintenance as the two communication objectives, whereas a high-involvement advertising situation usually calls for category need generation, the correct type of brand awareness, brand attitude creation or change, brand purchase intention formation, and purchase facilitation – thus all five communication effects become communication objectives.) Finally, advertising researchers should take note that the study of creative execution tactics , which includes the appropriate type of presenter to use and indeed whether or not to use a presenter, makes absolutely no sense without first classifying the advertising situation into the cells of the Rossiter-Percy Grid. (On this latter ground, researchers should note that almost every empirical study in the advertising research journals can be discounted as incomplete and uninformative.) The Rossiter-Percy Grid, for which no pallid substitute will suffice, is therefore the single most important planning framework for advertising researchers to consider as well as for real- world advertisers to use.
6. Brand Attitude: Informational or Transformational The Rossiter-Percy textbook, not so much in the first edition but certainly beginning with the second edition, makes extremely important contributions to the theory of Brand Attitude. There are actually three main contributions: a brand attitude model, an explanation of how informational advertising works, and an explanation of how transformational advertising works. (a) Rossiter and Percy provided a model, a “free standing” emotion-added extension of the pioneering Fishbein model, showing how brand attitude can be created, increased, or changed downwards in some cases, in two different and not necessarily incompatible ways: by instilling the learning of brand benefit beliefs (via so-called informational advertising) or by associatively pairing the brand with specific favorable emotions or even with an overall feeling of positive affect (via so-called transformational advertising). This is a logical, quantifiable model. We know from our industry contacts that many managers use it, but we have seen not one academic use it in an advertising research study. (b) Informational advertising instils brand benefit beliefs by making benefit claims , not by trying to communicate the benefit literally. Indeed, brands competing in the same product or service category often address the same key benefit , but advertising agencies are differentially effective in creating belief in this benefit for the brand simply because they make a more persuasive benefit claim. Advertising researchers, commercial or academic, invariably fail to understand the difference between a benefit, objectively stated and general, and the all-important benefit claim, subjectively stated or visually portrayed, and specific. To the present authors’ knowledge, the sole exception has been the professional advertising researcher Howard Moskowitz, who for many years has been testing the persuasiveness of specific visual, verbal, and musically conveyed benefit claims.
