Tracking deterioration of CHWY’s pandemic-era subscriber cohorts

Long Only Learning: Use consumer panel data to track the true health of subscription businesses, which may be masked by smoothed and aggregated company-reported metrics.

Chewy (CHWY) was one of the largest beneficiaries of the COVID pandemic, having the enviable characteristics of a consumer staple sold exclusively online. Exiting 2019, the business had a base of 13mm active customers and was growing customers at a healthy 6%+ QoQ clip (700k net adds a quarter) when the pandemic hit and drove growth higher. As stay at home directives forced folks who had never before shopped onto e-commerce channels, CHWY’s net adds doubled to over 1.4mm a quarter and exiting 2020 the base had expanded to over 19mm active customers. 

But in early 2021, as vaccines became available and in-store activity started to resume, CHWY investors were left grappling with two big unknowns. First, what would net adds volume look like as normalization occurred, and second and arguably more importantly, would newly acquired customers remain loyal as old shopping options reasserted themselves. 

On the first point, alternative data flagged rising churn rates in Apr 2021 (Exhibit 2), signaling that the normalized rate of net additions post-COVID may be lower than pre-pandemic levels. YipitData research specifically flagged that the driver was increased churn from the larger early pandemic cohorts, eventually corroborated by management ~2 months later during their F1Q21 earnings call.

On the second point, alternative data also helped validate that not only were these early pandemic cohorts larger, they were also lower quality.

Despite management’s assertions that the knock-on effects increased churn of larger 2020 cohorts was a “one-year phenomenon,” F1Q22 turned out to be the first quarter that CHWY reported negative net customer additions. Management would ultimately disclose that the early COVID cohorts were also suffering from lower retention - confirmed by alternative data pointing out repurchase rates for 2020 and later cohorts were below 2019 cohort levels (Exhibit 3).