Strategy White Paper

How is your brand’s awareness trending over time, relative to competition? Awareness is the single most important marketing variable in many product categories.

How is your brand’s image evolving over time? Think of “image” as the character or personality of a brand’s awareness. The strategic management of brand image is one of the most important goals of marketing.

Bold Italic

What advertising messages do your consumers remember about your brand, and how do these messages change over time? Advertising messages tend to undergo learning and memory “distortion” as they are interpreted and remembered by consumers. Therefore, the only way to know for sure the “net, net” communication of your advertising is to track advertising message recall.

What variables define your optimum target market? Who are your brand’s heavy users, nonusers, light users? The identification and monitoring of your brand’s optimum target market is one of the easily calculated outputs of good tracking research. What are the demographics (and the correlates) that define the optimal target market for your brand? Which market segments should you focus upon?

What impact are your competitors having in the marketplace, and how are competitive activities influencing your brand? Overreaction and underreaction to competitive initiatives constitute some of the greatest marketing mistakes historically. It’s really important to know, as early as possible, whether a new competitive product or new competitive advertising campaign is a real threat, or just smoke and vapors. Good tracking research allows you to monitor and assess competitive threats—before it’s too late to react.

Telephone surveys and online surveys are typically the best way to track awareness, image, and advertising message recall. . Telephone surveys can be continuous (i.e., conducted every day) or pulsed (conducted at a point in time, such as the last week of each quarter). Some types of tracking research can be conducted by mail (e.g., recognition tracking, image tracking, or brand share tracking), and the quality of the data from mail surveys can be high. Mail surveys, however, are not very good at measuring awareness (because respondents can ask other household members or Google the answers).

Good sampling is essential.. The greatest (and often least visible) mistakes in tracking research are usually sampling errors. The sampling plan and management of the sample are absolutely crucial to consistently accurate tracking data. The samples from month to month and year to year must be identical in every way or else the resulting data will not be comparable. Here are some common sampling errors to avoid:

1. Sample definition too narrow. If your target audience is females aged 21 to 29, that’s fine for guiding media placement. All too often, however, the target audience becomes the sampling plan for tracking. Therefore, only females aged 21 to 29 are interviewed in the tracking research. Suppose your advertising turns out to be really effective among women aged 34 to 54 instead of women aged 21 to 29. You might have canceled a very effective campaign because it appeared to be failing among the target audience. Also, it’s possible your advertising is working among 21- to 29-year-olds, but driving all other age groups away. If we were only sampling the 21 to 29 age segment, we would have overlooked this critical failing.
   
   Remember, always define the sample for tracking research very broadly and inclusively. The purpose of tracking is to tell us what’s happening in the marketplace, and a too-narrow sample almost always defeats this objective.
2. Variable definition of sample. Never allow the things you want to measure to be a part of the screening criteria that admits someone into the survey. For example, you would never want awareness of a product category or awareness of a brand to be part of the screening criteria for a tracking survey if one of the purposes of the tracking research is to measure awareness. Likewise, you would never want “past-30-day usage” of a category or brand to be a part of the sampling criteria, if the purpose of the study is to measure changes in usage over time. Awareness and product usage are variables that can change as a result of your marketing activities or competitive initiatives, or they can change from season to season. As these variables change, they change the composition of the tracking sample and destroy the comparability of the survey data across time.
3. Sampling without replacement. If the universe is limited (say you are tracking attitudes among your 1,000 dealers) and you take dealers out of the sample as they are surveyed, then the composition of your sample is gradually changing as interviewing progresses—and this makes the surveys from one time period incomparable to surveys in another time period.
   
   Remember if the universe is small and limited, then sample with replacement. That is, once a respondent is surveyed, put that respondent back into the sample for the next wave of data collection. An alternative solution is to divide the original sample into discrete, matched subsamples, and then use one of these subsamples for each subsequent wave of surveying.
4. Randomize sample within quota groups. Even though most projects begin with a random sample, things can happen that destroy randomness. For example, most samples are organized by time zone (so that households across the United States have equal time to respond). Sometimes, as part of this processing to organize the sample, the sample is put into some type of order (area code, prefix, or alphabet). As a final quality-control procedure, always randomize the final sample within each quota group. Then, no matter how the sample is worked, you will end up with a random sample.
5. Limit sample to force “callbacks”. The research company must limit the size of the original sample so that the “callback” cycle is properly triggered. If too many households are put in the initial sample, then it is likely that no “callbacks” will ever be made. The study is completed before the original sample is expended. The recommended policy is to release sample in a controlled, metered fashion so that every household has an equal opportunity to complete a survey.

The questionnaire must remain the same from month to month and year to year. Changes in the questionnaire (even something as seemingly innocent as a change in question order) can create unexpected changes in the results. Simply changing one word in a question can change the results. Therefore, keep the questionnaire constant over time. If you want to modify, add, or delete questions in a tracking study, do it toward the end of the questionnaire—so that the changes will not distort the key measures in the first 80% of the questionnaire.

All data collection procedures and controls must remain constant over time. Changes in the minutia, scheduling, sampling, email reminders or “callbacks” can inject unplanned changes in tracking study results. The instructions and procedures for each specific tracking study must remain unchanged over time.

Editing, coding, data cleaning, and tabulation must remain constant over time. Changes in the way “no answers” or “blanks” are handled, changes in how many multiple responses are accepted, and a hundred other “minor” tabulation details can change tracking results over time.