If the only goal of your research is to determine which price point yields the most favorable rating, then any inherent
problems with measuring purchase probabilities are minimized. The simple mean scores comparing the stated likelihood of
purchase, or interest in the product, can be compared across the different price points. The most favored rating
identifies the best price point.
Rarely, however, is this level of interest sufficient. Most of our clients want to know not only which price
point is best received but also what will generate the most revenue. Furthermore, if there is any question
whatsoever about the possibility of introducing the product, they want accurate information so that they can
assess what their return on investment will be for this new product, and if that is sufficient to warrant the
product development, marketing, and sales effort that will be required. Under these circumstances, an accurate
assessment of the likelihood of purchase is required.
What this implies, of course, is that survey data as provided by most respondents are not accurate in
this regard. This is true. Under most but not all circumstances, respondents will in the aggregate
over-state their likelihood of purchase. The aggregate over-statement of purchases masks some interesting sub-trends. If
we ask the question about the likelihood of purchase and use a response scale of 0-10, where 0 means absolutely
will not purchase and 10 means definitely will purchase, we see that:
A small percentage of respondents who state that they absolutely will not purchase the product (0 on the response scale)
will in fact ultimately purchase the product, i.e., they have understated their stated likelihood of purchasing.
This might be a quite small number (e.g., one-half of 1%), but the total volume of sales attributable
to this group can, in the final analysis, be significant.
The primary reason for this pattern is that some respondents simply cannot envision themselves
purchasing a product without having seen it, heard their friends talk about it, and the like.
A very large percentage of those persons who give a 10 on the response scale, implying that
they definitely will purchase, will not do so within the time frame specified in the question. Oftentimes
over 50% of respondents who state that they definitely will purchase will not actually do so.
There are many reasons for this pattern. Looking forward, some respondents cannot accurately
forecast their time horizon. Many respondents want to appear agreeable during the interview,
and thus provide the answers that they believe the interviewer wants to hear.
What this implies, of course, is that producing an accurate sales forecast from survey data requires
that questions about the likelihood of purchase must be calibrated to more accurately reflect the actual
likelihood of purchase. Most typically, the calibration involves transformation from the linear 0-10 scale
onto an S-curve function with modest purchase probabilities at the low end of the scale, rapid acceleration
of the curve through the intermediate portions of the scale, followed by a plateau effect at the upper
end of the response scale.
At MRA, we use industry sources, existing research, and data from the survey to calibrate
purchase probabilities. The end result is a sales forecast with alternative scenarios that
allows you to make informed decisions about potential that any new product introductions might have.
No other research company can bring you this level of precision and
expertise. Please contact us to
find out how our state of the art solutions can help your marketing decisions.
Proceed to Previous Page