Incrementality

In Drew’s newsletter last week, he talked about stacking offers to capture as much margin as possible. I wanted to extend that thinking by asking (and answering) the question: What makes an offer good? 

If the subject line didn’t give it away, the topic I want to cover today is incrementality.

When should we use discounts? When is 20% off better than 10% off? Discounts, while effective, can put you in a scenario where, if you don’t generate enough incremental lift in conversion, you may make more revenue, but less profit—or even less revenue and less profit. 

Excuse me while I get a little math-y, but this is how you can put yourself in such a scenario. 

First, let’s assume you test a 10% discount against a 20% discount. Let’s also assume that the average cart composition stays the same. In scenario 1, the number of orders increases by 40% when customers are offered 20% off. Such a discount will create ~30% more revenue and about 5% more profit. 

But, as in Scenario 2, if that richer discount only creates 20% more orders, you end up with about 6% more revenue and 10% less profit. (The picture looks worse still if the 20% discount only increases orders by 10%, as both revenue and profit trail the 10% discount in such a scenario.)

There are, to be clear, reasons to discount outside of boosting conversions to generate more profit dollars. So here are some thoughts about when incrementality alone isn’t enough.

You might want to clear inventory before it expires or goes out of season, for example. And, in that case, it’d be hard to argue that it’s better to let a product expire than to take less profit. 

Same, too, if you believe (or have data to show) that a larger discount now can improve 90-day LTV. Such a trade (less profit now for more profit from that customer in the next three months) would also prove that your richer discount isn’t just pulling forward existing demand.

You might also have data that shows you need to improve your discount offer just to maintain profit targets when other big players discount. (Amazon Prime Day just wrapped last week, for example.)

These reasons, of course, are better to have defined before you test your discounts, as it helps you decide whether a particular scenario outlined above is “worth it” and whether they’re worth potential tradeoffs (creating “bad behaviors” among consumers, devaluing the brand, etc).

That said, weighing tradeoffs may not be necessary.

A more granular view of stacking offers in the way Drew discussed last week could include serving specific discounts to different segments, in effect personalizing the discounts so that you only serve the richer discounts to those who need it and preserve your margin with those who don’t. (You can use Intelligems for this.)

The advantage here is what Drew wrote last week: 

The end result of this … is the quantity sold gets stretched to the global maximum (the maximum quantity that can be sold at the unit cost or above) and the price simultaneously gets stretched to the local maximums for each segment (what customers/various customer segments are willing to pay).

Meaning, you sell more products (like Scenario 1), but preserve as much profit as possible. So, you don’t have to choose at all. That’s a pretty big win.