Editor’s note: Paul Zaff is founder and principal of BLT Research, Framingham, Mass.

It seems that the basic questions behind almost every recent legal or political scandal are: “What did he know and when did he know it?” While certainly not as salacious, there are two fundamental questions in marketing research, appearing at some point in most questionnaires regardless of the larger purpose: “What did they buy?” and What’ll they buy next?”

The first question refers to usage and is asked in terms of last purchase or ownership with regard to products and services. It can also appear as voting, readership, viewership, etc., depending upon the focus of the investigation. The second question is purchase (or future) intent and can take corresponding forms.

These are probably the two most common questions in the research arsenal and they serve multiple functions. At the least, they position the respondent for the remainder of the questionnaire, but more likely the usage and intention data is tracked and crosstabbed, and the usage often functions as the basis for separate sub-group analysis. And that’s that! The study and questionnaire go on to fulfill a greater purpose (i.e., advertising, tracking, multivariate or other sophisticated design) and the focus of analysis is spent on detailed findings relative to the primary points of investigation.

But these two questions can provide much more than just basic starting point or foundational information. With some additional effort they can yield dramatic insights into product/market conditions and behavior that make the existing research much richer in actionable content.

A procedure called PAT (probability analysis technique) is the device to maximize the output from these two basic questions. PAT is a means of taking information from last-purchased (or current used/owned) data and purchase intent data to develop a comprehensive framework for predictive changes in market (share) potential, net share change (draw from as well as loss to other products), loyalty, switching, cannibalization, and product segmentation.

The single caveat is that, as the name indicates, PAT requires a probability measurement for the purchase intent question. This should be in the form of a chip sort or a point allocation question. It can be used alone as the source of intent or, if precedence dictates, as an adjunct to traditional single or multiple-choice questions.

While PAT can be utilized with any form of interviewing, it is best served by some form of personal, online or self-administered techniques. The reasoning is that greater visualization is allowed for product alternatives and probability counts. The closer to real-life product alternatives, of course, the better, but telephone will work reasonably well when choices are of limited quantity and well known.

First and foremost, PAT is a simulation methodology. It provides the magic number(s) that marketing wants to know. The world of market research is replete with many far more sophisticated simulation techniques that project not only share but volume, and can be calibrated to consider a host of offering and promotional options. Suffice it to say that (within the +/- limits of virtually all such methodologies) PAT projects future market share for all products within a defined market. Consider it, if you must, a poor man’s substitute as it can be incorporated into ongoing research. This is useful when learning about a new market, measuring the impact of changes in product or promotion in an existing market, considering new product potential (both in terms of the new product as well as its effect on competition), and perhaps as a corroboration of other simulations.

A word about that magic number. Don’t bet the farm on it. This isn’t heresy; it’s common sense. PAT has proven to be accurate regarding winners and losers, but precision is a matter that must be considered with caution. The share numbers that PAT projects more practically indicate product share potential as viewed from a specific interview date, under a unique set of marketing and economic conditions. Actual product performance is really an abstraction, dependent upon the above conditions and the place of the product on the continuum from fast-moving consumer goods to long-term durables. The PAT projected share number provides a reliable order of magnitude for share change and, just as importantly, a frame of reference for the other analytical tools that are described below.

And a few more words about market share. PAT addresses market share from a user basis. In many cases, businesses measure on a unit basis. While user and units often move in lockstep, this is not always the case, particularly when there are significant price discrepancies within a market. Nevertheless, PAT is a valuable tool for predicting individual product strengths and weaknesses, and analyzing the reasons why.

In addition, PAT measures the size of the slice of the pie, not the pie. It works best when the pie is relatively stable. In markets that are rapidly growing, the sample must be adjusted to include non-users as a segment. Declining markets — shrinking pies — are more difficult because purchases may be discontinued (as in the case of obsolescence) or volume may be reduced through stretching (using products longer) or postponement, as often happens when economic conditions worsen.

PAT examines the share changes from the perspective of the market in two ways — a macro and a micro analysis. On a macro level, it supplies information relevant to product volatility and loyalty, each of which can impact profitability. At the micro level PAT provides predictive draw/loss data and guidance on segmentation and cannibalization.

The macro look - market switching and loyalty

As a macro analysis PAT explains share movement from the perspective of market volatility and direction. Switching and loyalty are the drivers of market share change. Switching is the transmission and accelerator of share dynamics; loyalty the shock absorber.

All changes in unit share (with the exception of consumption changes) are the result of customer switching. This change can result from either customer movement from competing brands or first-time users of a product category (in which sense they are switching from non-usage). When a product has more new users - “comers” - than “leavers,” it must exhibit share growth.

PAT considers two types of switching - market switching and product switching. The former identifies the profile of a product within the context of its market (macro). It is a measure of volatility and the direction of share change.

There are two measurement components to market switching. First, there is a switching index, which measures share volatility - high indexes indicate potential for larger share changes, low indexes mean far less dramatic changes. Note the use of the word “potential.” A high switching index means that there are a lot of customers changing brands, but there is another factor that determines brand growth or decline.

This is the switching ratio (direction), which records the amount of customers coming to and going from a product. If there are more switchers into a product than leaving it, then the market share must increase, irrespective of the level of volatility. Similarly, if more leave than enter (a negative ratio), the market share must decline. Thus, the level of switching indicates potential volatility (the accelerator); while the direction of switching is a measure of growth/decline (the transmission). Together, they explain share change.

Loyalty is, of course, the tendency for consumers to stay with a product (repurchase), and PAT measures, for all products under consideration, levels of future loyalty. Strong loyalty also implies advocacy, which when put into effect is the best and most cost-effective means of promotion. Products with a high loyalty reading are more likely to exhibit share growth, but if not, then it acts as a buffer for any share decline. Most importantly, loyalty is a critical indicator of profitability because it’s far less expensive to make that next sale to an existing customer than it is to educate and convince a non-customer. Changes in loyalty can be tracked over time and measurement of this attribute is an often overlooked component of positioning and advertising tests.

It is a given that different types of markets exhibit different levels of loyalty, but PAT shows that there can also be very different levels of loyalty among products within a particular market. This may be a function of a given product’s promotion, perceived value, utilization or life cycle stage.

While an individual consumer is either very loyal or likely to switch, on a market basis, switching and loyalty are mutually exclusive. That is, a product can have high loyalty as well as high switching. This is often the case for strong newer products or existing products with high projected growth rates. On the other hand, a product with low levels of switching (even if negative) and high loyalty may not have the growth potential, but still be quite profitable as a harvest candidate.

Figure 1

In Figure 1, A and D have high loyalty, high switching indices and a positive switching ratio, and they are the primary prospective share gainers in this market. F also has high loyalty, but low and negative switching - it is in the harvest mode. Product E has a high switching index, but its ratio is decidedly negative and it can expect severe share problems.

The test also measured the market efficacy of two new products K and L. One was never launched; the other came to market and was withdrawn.

High levels of switching that are generally neutral indicate that a product is potentially volatile. This is important because lack of projected share change doesn’t necessarily mean stability. High switching indicates that the product has to fight harder to maintain share and this impacts profitability. The message here is that two hypothetical products with identical shares and no predictive share changes may still have very different profiles with regard to profitability. A deeper analysis can reveal each product’s true characteristics and perhaps its changes over time as a result of different promotions or its life cycle position.

Thus, PAT can examine a product on a market basis by deducing share change, loyalty and volatility, and it does so for all products in the market. This is useful for examination of ongoing markets as well as determining new product potential.

The micro look - product switching, segmentation and cannibalization

On a product basis (as opposed to the market basis) PAT can examine the dynamics of share change as well as product segmentation. In all markets, products will experience some levels of switching with virtually all other products. Higher switching levels between specific products are indicative of a true segment. As above, switching (this time between specific products) can range from positive to neutral to negative. If two products have a high switching index (they form a segment) but there are equal quantities coming and going (draw/loss) then the net share change will be neutral between them, but the interaction can still be potentially unstable. Strong new promotional efforts by one will disproportionately affect the other.

The micro analysis allows detailed examination of draw/loss data for every product. All prospective changes in market share are the result of gains or losses from competitive products, both within- as well as outside-segment, and this information helps identify the key contributors to change. This analysis also measures the effects of cannibalization for those brands with multiple products in a market. And as PAT predicts new product potential, the draw/loss analysis shows the impact of introductions on all other products in the marketplace.

Figure 2

As Figure 2 shows, Product C was being considered to counter the success of Product F, which had quickly achieved a prominent market position. If launched it would help blunt the competitive gains. PAT projected that C would achieve a substantial introductory share, drawing from all market products, but primarily from F, which would lose 1.5 share points to its new competitor, even though it would still grow by 3 points. The new product would cannibalize 38 percent from other company products, an excellent option given its premium pricing.

Further, this type of analysis showed that the company behind the new product would hold total share if C was introduced, but would lose over 4 points in total market share without it.

The existence of a segment (high switching indexes between specific products) means that the brands must concentrate their marketing efforts against others in that segment in order to maintain or grow their competitive position. It falls to other forms of research to identify the characteristics and drivers that affect or differentiate that segment.

While product managers obviously know segments intuitively, PAT will often show that there are real and quantifiable segments within the intuitive subsets, thus narrowing the competitive focus. Further, segment growth, by definition, can only come from outside the segment, and PAT can show the key target products outside the segment for share gains or protection. Consider a product that is segment-neutral - holding its own within its segment, gaining as much share as losing to/from competition - but grows in total share because it is more effectively drawing from outside the segment. Here PAT can provide an entry point for more sophisticated research techniques by identifying the proper targets for attribute or motivational (growth driver) analysis, and after the fact, it can measure their potential effectiveness.

Figure 3

In Figure 3, Product A is projected to grow 0.7 share points, from 2.6 percent to 3.3 percent. Management had intuited that the market segment for A included products B through J. They were correct, as inter-product switching activity - indicated by circle size - was higher in this segment. But there was clearly a more relevant segment within the intuitive segment (C, D, G and H). And, more importantly, potential share gains were greatest from outside the segment, particularly from L. This is because it is a larger share product and the switching, although lower than segment levels, is decidedly positive toward A. This is a target product as opposed to a segment product. PAT identifies where to concentrate.

In addition to predictive share and loyalty/switching data, detailed analysis of the usage/intent questions can examine inter-product performance including real product segmentation and cannibalization. Traditional purchaser studies consider draw only, but markets are dynamic and every existing product experiences losses as well. True product segmentation results from analysis of the net effects of both draw as well as loss. Products within a relevant segment have high interaction. It may or not follow that there is a net share gain of magnitude for either product from the other, but if they are highly interactive then there is a competitive segment situation that must be addressed.

Similarly, PAT can examine cannibalization. This is most practical for new product introductions, but can also be pertinent when there are special promotions for existing products.

The next logical step in the PAT analysis is the measurement of what-if scenarios in the marketplace. Given that the product inter-dynamics are already known, then PAT can measure the effects of reasonable alterations in pricing or promotion outlays. Again, there are other more sophisticated simulations for this purpose, but the data is here and readily available.

A price reduction will grow share, but the degree will depend upon the switching levels, and in turn impact elasticity. Because PAT examines the total market, changes in pricing or promotional budgets will affect not only the target product, but also all other products in the market, and PAT will show the impact. One interesting possibility here is that if a major segment player makes a significant change, say increasing advertising, then that brand will benefit with share improvement, and its draw from competition will be enhanced. But, the net share effect on any particular competitive product may not be negative. It is quite possible that the segment may be enhanced (the river raised) and a competitor, while losing more to the initiator, will draw more from outside the segment, thus gaining total share even in light of increased competition.

Richer, more meaningful

In summary, most studies already being done contain the foundation for a richer, more meaningful and actionable analysis. Because of its inherent simplicity and the fact that it can be used as an adjunct to existing research, PAT is readily affordable. The market share projections are often considered the meat of the analysis, but depending on the type of product or service, they are only the starting point for the true value of PAT. It is a learning tool and provides valuable insights for all types of products (packaged goods to durables) and services. In this tough economic climate, it’s incumbent upon research specialists to get the most bang for their buck and maximize the value of their data.