How to Personalize Offers & Prices

Olcan Sercinoglu

Oct 19, 2018

In previous posts, we began exploring how to maximize revenue growth. In this post, we look at how to do so using personalized offers and prices.

Introduction

Offers are a great way to dynamically impact price or perceived value, the maximum price that a customer is willing to pay during a session.

Below is an example offer from Eight Sleep. The offer is for free pillows if a mattress is purchased within a limited time. It increases perceived value, including indirectly by creating urgency to pay during the current session.

offers-eightsleep.png#asset:791

The Revenue Growth Opportunity

Revenue is maximized when perceived value equals price. Since value perception depends on the customer, a single value proposition (and price point) leaves open an enormous opportunity for growth in two key segments, illustrated below.

offers-opportunity-segments.png#asset:795

How to Personalize for Revenue Growth

We can realize this opportunity by designing variants for these segments, and then using Amp.ai to utilize each variant in the right segment. This is commonly known as personalization and emerges naturally out of an Autonomous Optimization (AO) project designed to maximize revenue.

A well-designed AO project can achieve significant revenue growth within 2-4 weeks, turning the initial cost of learning into a great investment for near-future gains. In contrast, A/B Testing is not useful here for two reasons:

Designing Variants to Maximize Revenue Growth

We follow a simple framework to design variants that maximize the revenue growth opportunity. For any given design, we only need to estimate three parameters (highlighted in bold below) to calculate the precise opportunity. 

offers-framework.png#asset:793

We should run through several design scenarios with different estimates to get a sense of what is possible and how it is affected by our design choices. Doing this can significantly improve our design.

Below is an example where we estimate the growth opportunity as up to +280%. Our estimated parameters are in bold, and we assume a baseline conversion rate (where perceived value >= price) of 40%.

offers-framework-example.png#asset:792

Realizing (More) Revenue Growth

How can we realize (more of) this opportunity? The key assumption we make in our opportunity calculation is that we can apply variants precisely on their target segments based on perceived value. In reality, we can only observe other relevant context variables, which limits us to segments that can be discerned from those variables. Thus, to realize more revenue growth, we need to make more relevant context variables available at the decision point.

For example, consider a 50% discount. Although we cannot directly discern the target segment of people who need this discount, we CAN discern the segment where the customer is using a lower-end device, connecting from a lower-income country, visiting through ads/searches for an affordable product, and so on, as long as these variables are available at the decision point.

offers-learning-segment.png#asset:815


Initial Metrics and Allocation

Besides ensuring good variant design, we should also monitor the initial metrics and consider optimizing a smaller percentage of sessions until we are comfortable with the initial cost of learning.

Fairness in Personalized Pricing

Fairness can be a concern for personalized pricing. Below are a few solutions:

In future posts  we will look at in-app purchase optimization, push-button integration with major ecommerce platforms such as Shopify, and several other ecommerce related topics. We will also continue to explore many other use cases across the customer funnel and software stack.