How Wish uses BoostKPI to improve their marketing ROI

Background 

Wish is an e-commerce marketplace active in over 50 countries. It has over a million registered merchants selling millions of products easily crossing ARR of USD  2.5 billion in 2021 alone.  

Wish has been one of the largest advertisers on Facebook, spending over a billion dollars in a year. A key ad format it uses is dynamic (product) ads. Wish uploads a subset of its product catalog to Facebook. Facebook then decides which products to display to its users in a carousel format.

Problem

There are millions of possible combinations of products, countries and ad_set. Wish wanted to track any sudden changes in spend, impressions, clicks, or installs for any of these combinations. 

The marketing team at Wish initially built their own alerting system internally to track these changes. However, the system was not comprehensive and difficult to maintain.

Wish wanted to:

• Find out root causes of any large spend changes easily
• Receive ML-guided alerts which they could customize without deploying any new code
• Monitor custom use cases. For example, ensure that there was no spend in a country (that Wish was no longer active in). Or, ensure that cpcs never exceed a threshold

Solution

The team at Wish turned to BoostKPI to detect anomalies and root causes in their marketing data. With BoostKPI, Wish was able to:

• Use the BoostKPI dashboard to figure out root causes of any spend changes. 
• Receive ML-guided alerts. The alerts worked even as the products, ad_sets, or the country list changed. Wish was able to customize its alerts. 
• Wish was also able to set up custom alerts. For example, a single alert configuration (added on the BoostKPI dashboard)  ensured that Wish was not wasting marketing spend in countries they were no longer active in.

In addition, Wish chose to receive the alerts both over email and slack. 

An example of the slack alert is shown below. The alert is triggered when spend on any ad_set changes over a certain threshold across a number of whitelisted countries. 

‍

‍

Tianhang, an engineer managing the marketing spend at Wish, says: