Retention & Loyalty Early-Warning · Brief 01
A seller cohort is spending guest goodwill faster than it earns GMV
New Q3 sellers · week of 2026-01-19 · returns → shipping → ratings → the basket next door
The one line you configure
Which guest cohorts are showing the early tells of leaving — and where a cohort is deteriorating, is it the guest changing or somebody serving them badly?
In a nutshell
Somebody serving them badly, and it is one onboarding cohort. Sellers added last quarter are running a 31% marketplace return rate against a ~12% seasonal baseline, shipping on time 78% of the time against a 92% SLA, and 61% of the move sits with sellers now rated under 3.5. The cohort's GMV growth is flat year on year — the volume it was onboarded to add is being cancelled by what comes back. The tell that this is a retention problem and not a marketplace one: owned-brand small appliances on the same shelf sold 12% fewer units in the same week. Guests who had a bad experience paused the whole basket.
At a glance
What moved
What's happening · the return rate left its post-holiday band and did not come back on its own
Post-holiday weeks always return more; the baseline already carries that. What the baseline does not carry is a cohort in its first post-holiday window running at nearly three times it.
Why · it is an onboarding pipeline, not a category
Contribution analysis of the return rate, one value per dimension. Small Appliances is where it shows; the Shopify onboarding pipeline is where it came from — the same category onboarded direct is not in the cohort.
What it is costing, in three other places
The cost is not confined to the cohort. Guests who bought from it are buying less owned-brand on the shelf beside it, and the advertisers whose campaigns were credited with those orders are having the credit taken back.
So what
The retention damage runs ahead of the GMV damage and it is being done by seven weeks of onboarding decisions. Pausing new SKU additions from the cohort stops the bleeding today; re-reviewing the Shopify pipeline's acceptance bar is what stops the next cohort. Both are cheaper than the goodwill, which is already showing up as a paused basket on the owned-brand shelf.
The points that matter
The volume was never real
Cohort GMV growth is flat year on year despite the onboarding pace — the orders arrive and then come back. The cohort has been adding activity, not revenue.
flatGMV growthIt is a pipeline, not a category
52% of the move traces to how the sellers were onboarded rather than to what they sell. The same sub-category onboarded direct is inside its band.
52%one pipelineThe goodwill is already spent next door
Owned-brand small appliances on the same shelf sold 12% fewer units in the same week, at 1.25× the markdown. A bad third-party experience is pausing first-party baskets.
−12%own-brand unitsRecommendation
Pause new SKU additions from the New Q3 cohort, put the Shopify onboarding pipeline back through review, and tell the affected mid-tier Home advertisers their attribution will be restated before they find the reversal themselves.
Questions it already answers
Is this seasonal? Post-holiday weeks always return more.
The baseline already carries the season — ~12% for new-seller cohorts in their first post-holiday window. This cohort is at 31%, and its on-time ship rate is 14 points under SLA, which the season does not explain.
Is it what they sell, or how we onboarded them?
How they were onboarded. The Shopify pipeline carries 52% of the move and sellers now rated under 3.5 carry 61%; Small Appliances sold by direct-onboarded sellers sits inside its band.
Who else is affected?
Owned-brand small appliances, down 12% in units on the same shelf, and the mid-tier Home advertisers whose attributed orders are being reversed by the returns — their attribution reads 18% lower for reasons that have nothing to do with their campaigns.