How Discounting Affects Product Value Perception?

It is fair to say everyone loves a discount. There is data to support the claim from years of Black Friday discounting that not only makes customers skip sleep and stand in cold weather for hours but also generates considerable revenue for the retailers. Then there are the group buying discounts made popular by GroupOn.

At the individual customer level, previous research done on customer preference for discounting point to both rational economic and emotional reasons.  Kahneman and Tversky showed, customers prefer discounts even if they saved lower amount in absolute terms.

Thaler in his work on Mental Accounting and Consumer  choice provide evidence of emotional (non-financial) reasons for why we may be motivated by discounts.

So we all value deals for rational and irrational reasons. But how much do we value the product after the initial buying decision? More specifically, since we do not know the absolute values, how much relative value do we place on a product bought on a discount vs. an identical and substitutable product bought without the discount?

There are three distinct possibilities.

Hypothesis 1: If we treat our customers as rational (ignoring the contradiction), once they bought the product at whatever price, the costs are sunk. How much relative value consumers get from each product should depend only on its own merits and independent of initial discount. At the very least, the choice between the two should be no different from  a random choice.

Hypothesis 2: If we treat our customers consistent with their previous action of seeking discounts for non-financial reasons, we should expect them to value the product bought at a discount higher than that of the one bought at full price.  This hypothesis is further reinforced by endowment effect and cognitive dissonance (I must really value this product, otherwise I would not have stood in line for it).

Hypothesis 3: Customers value the full price product more than the one bought at discount (remember the preconditions – identical and substitutable products).

So I conducted an experiment. The data and analysis show that  the last scenario, Hypothesis 3, is more likely than the other two.

The experiment was designed to get customers to reveal their preference by asking them to give away one of the two identical products of same price:

1. One they bought at full price
2. Other they bought at 50% discount

As it turns out, customers are more likely to keep for themselves the full priced product and happily give away half-priced identical product.  So while discounting may bring in customers, it may hurt by depressing the relative value of the product to the customers.

What does this mean to you as a marketer?

1. Think again before running 50% deals on any group buying site or allowing your product to be bundled and sold at a discount by other channels.
2. Attracted by large user base from giving away your product for free? Consider loss in perceived value to the customer.
3. Know the customer’s end use of the product. If they are predominantly buying it for their own use, discounting most likely will not help improve lifetime value of the customer.  If they are buying it for others, discounting helps.

Want more actionable insights? Talk to me.

Further readings:

Kahneman and Tversky – Choices, Values and Frames, American Psychologist,  1984

Thaler – Mental Accounting and Consumer Choice, Marketing Science 1985

Note on the Experimental Method:

I relied on convenient sampling and applied Bayesian hypothesis testing. Compared to conventional, run of the mill hypothesis testing, say testing for statistical significance at 95% confidence level using Chi-square test for this case, Bayesian Hypothesis testing allows to test multiple hypothesis at the same time and help state the results in terms of probabilities instead of as absolute truths.

Results are subject to sampling errors, and do not take into account segmentation differences. This is also stated preference study and not a revealed preference study.