Are Promotions Growing Your Business or Losing You Money?

Promotions are a ubiquitous part of almost every business.

They can help deepen the relationship with existing customers, increasing engagement and revenue; they can drive adoption of new products and services; and importantly, they can be used to reduce customer churn. 

While promotions can be a powerful tool in your toolbox, they tend to be difficult to pull off correctly. There are three big questions most marketers face when evaluating promotions:

• First, which customer segment should I be targeting?
• Second, is this the best promotion I can be offering at this time to drive adoption (or some other desired outcome)?
• And finally, is the price incentive in this promotion worth the increase in customer adoption? Or in other words: is this actually making me any money?

These are the questions that often keep marketers up at night. 

For example, let’s imagine a DTC company that sells premium bags through their website and app. Their business has grown considerably over the past few years, boosted by the COVID-19 backed shift to online shopping. However, their new customer growth has stalled recently due to multiple headwinds that most growth brands face today: rising customer acquisition costs (CACs) due to diminishing returns on additional advertising spend, coupled with the profound sea change taking place in online advertising around privacy and tracking, which is limiting the efficacy of advertising and traditional attribution, and finally, unfavorable macroeconomic conditions. As a way to fill this gap, the business is shifting focus to increasing revenue from existing customers.

The company decides to try out a promotion to drive repurchases. Choosing to be rigorous about it, they run the promotion within a randomized A/B test¹, targeting some sub-population of their customers. Unfortunately, the experiment ends up being not very informative, from a net profit perspective, due to the revenue lift being somewhat canceled out by the added promotion costs. 

They decide to repeat the test with different promotions and customer segments, hoping to eventually find the validation they're looking for. In the meantime, the business' leadership still has no idea whether promotions are doing the business any good.

Does this story sound familiar to you? 

An alternative to offering promotions to the entire user-base, or to large segments of it, is to try and personalize them by targeting customers based on propensity modeling. In propensity modeling we take past customer data, such as purchase history, website usage, etc. and try to predict the likelihood of the customer to take some action within some time horizon, e.g. repurchase within a month. It might seem reasonable to select the customers who are most likely to take the action of interest and give them a promotion. However, this approach can actually be wasteful and cost the business money. The reason for that lies in the way different people react to promotions. See the below table for the 4 main types of response patterns.

‍

‍

‍

As can be seen in the table, always-takers would have taken the action anyway and thus promoting to them squanders precious dollars which could be used to engage other customers. They are usually the main reason why both promoting to the entire user-base, or targeting customers using propensity modeling, can yield inconclusive results. Another group, the defiers, embody the idea of letting sleeping dogs lie - if you promote to these people you will get worse outcomes (think email unsubscribes). Obviously, you wouldn’t want to target defiers either. Finally, never-takers are indifferent to your promotion efforts.

As should be evident by now, compliers are the only type of customers that benefit from a promotion and we should strive to only target them.

Keep reading here to learn how we can do that.

‍

¹ If no A/B test is done, and the roll out of the promotion is to everyone, it will be really hard to actually gain insight into the impact of the promotion on the business