POINT-OF-SALE (POS) RESEARCH IN THE SPORTING GOODS INDUSTRY

In theory, point-of-sale (POS) research is a valuable source of marketing information. By offering a high degree of accuracy for whatever data are captured, POS can overcome some weaknesses of the consumer research survey.

POS research is not error-free, but barcode scanning does ensure fairly precise reporting and tabulation of category, price, retail distribution channel, brand, model, size and other encoded product information. Totally objective computerization avoids human reporting and its inherent shortcomings, exempting POS research from criticisms leveled against methodologies based on the respondent's memory of a purchase experience. Consumer research can be invaluable, but memory-based reporting (even a daily diary) is subject to various forms of conscious and unconscious distortion — including forgetting, "telescoping", boredom, fatigue, disinterest, carelessness, or simply the need to be "included" in a survey.

On the other hand, while POS research can generally claim high-grade accuracy, such studies (with rare, futuristic exceptions) don't tell us anything about the buyer at the cash register. Consumer surveys can provide an endless array of demographics, psychographics, product usage, sports participation, attitudes, health habits and other consumer behavior — in addition to the core marketing elements of category, price, channel and brand. A consumer survey is also projectable to the entire market or population; as a practical matter, the POS study is not.

But the really prohibitive problem with most POS studies is that they do not generally represent all retail distribution channels, and even more seriously — due to a lack of respondent cooperation — cannot even adequately sample a representative part of a given distribution channel.

A well-known industry study exemplifies both deficiencies. Its data collection points (retail stores) include Kmart but not Wal-Mart; JC Penney but not Sears; Footaction but not Foot Locker, etc. Even worse, the paucity of Specialty retailers is a giant black hole in the study coverage: Exercise Equipment, Golf Pros, Bicycle Stores, Gun Dealers and Ski Shops — among other specialty outlets — are poorly (if at all) represented. In such a study, the greater the importance of the specialty retailer, the greater the distortion of brand share and all other reported data. Consequently, industries with a heavy reliance on mass merchants and discount stores (athletic footwear and camping, for example) may find a vaguely realistic portrait of the marketplace. But in Golf, a category heavily vested in specialty channels, Callaway for example — the undisputed market leader — is notoriously absent from the top brand rankings.

In addition, only a small fraction of all sporting goods categories are addressed, and among those measured, product breakdowns are vague or insufficient.

Some years ago, a POS study of Specialty Retailers in Sporting Goods had the opposite problem: it emphasized smaller retailers at the expense of large chains, and was further handicapped by a geographically skewed distribution of stores dominated by the northeast.

Yet for all these deficiencies, (and laying no claim to a sample weighting scheme!), POS studies purport to measure "brand share", "total size of market" and "top sellers". In reality, actual market volume may be several times that depicted for most categories, and no single market is accurately portrayed.

To make matters worse, unmodulated year-to-year "comparative" data are typically shown without any adjustment for new stores added in the current period. Naturally, a fair contrast with the prior year must show "same store" comparisons — but methodological propriety is rarely a concern of the research entrepreneur.

On the positive side, any designer of syndicated research appreciates the brute labor and painstaking effort demanded by such a gargantuan project…and can only applaud the effort. But at present, POS research is based on a large, amorphous chunk of the market — an undefined fragment of the retail universe. Tabulations of a partial indescribable, constantly changing sample will be misleading — especially when misinterpreted as representing the entire sporting goods market, or even an individual category. We can never divine the true market share of Reebok, the number of mountain bikes sold through discounters or how many dozen golf balls were purchased at pro shops — least of all, across the entire U.S.

The results of a POS study (in theory) can be legitimately presented in one of two ways — but neither one is feasible. The first remedy is to vastly increase the scope of research coverage by enlisting near-perfect retailer cooperation in all product categories. The second is to somehow weight the raw data, projecting it to a known retail universe. Unfortunately, the complexities and uncertainties of this option are even more daunting, if not insurmountable. To project to a universe (our sample balancing target) we need to know the size and composition of that larger market reality…but we don't.

Sample balancing in a consumer study is manageable, because we are dealing with a person as the fundamental, statistical unit. The Census Bureau produces the national demographic targets that we need to aim for, so we can "weight" individual respondents (people) by gender, age, income, geography, etc. In a retail environment, where the store is the fundamental unit of statistical analysis, certainty dissolves into chaos — because there is no comprehensive Retail Census of all distribution channels. How much weight do we assign a single tennis store? A tennis pro shop in a country club? A tennis department at Sports Authority? At Wal-Mart?