We ran a control group on our own store

Two weeks ago we wrote that we’d re-comped eight aged listings on our own store and four of them sold — two inside 48 hours, after sitting more than a year.

It’s a good story. It’s also the kind of story every tool tells about itself, and it has an obvious hole: maybe they’d have sold anyway. Stock sells. Old stock sells sometimes. Four out of eight with no baseline is an anecdote wearing a number.

So we went back and measured the baseline.

The control group

On 3 August, the day we repriced those eight, the book also held 132 other listings that were in the same position: aged 90 days or more, under $200, and not repriced. We left them exactly as they were.

Then we watched both groups for the same 30 days. Figures as at 5 September.

Six times the conversion rate, on matched stock, over the same window.

That’s the number that turns “we repriced some things and they sold” into something you can actually reason about. The untouched aged stock kept selling at its own slow rate — eight per cent a month — and the repriced stock sold at six times that.

The bits we’re obliged to say

We’re running a store, not a lab, and there are three things a careful reader would push on.

Eight is a small number. It’s directionally overwhelming and statistically thin. We’d want thirty before we’d call it settled, and the only honest way to get there is to keep tagging every reprice and watching the untreated cohort alongside it. That’s now running.

One of the four fell over afterwards. The Freaked Collector’s Edition sold on the reprice and was then cancelled — because we shipped that unit to the wrong buyer. That was a warehouse error, not the market’s verdict on the price. The reprice did its job; we didn’t. We’ve counted it, because a cancelled sale is a normal thing that happens in a store.

One of the eight was dearer than the control. The Ryzen at $315 sits above the $200 cap we used to match the control group. Take it out and it’s three of seven — still 43%, still five times the baseline.

Why the baseline matters more than the result

Here’s what the control group actually tells you, and it’s not “repricing works.”

It’s that aged stock you leave alone converts at about 8% a month. That’s the cost of waiting, made visible. Every month a listing sits at a price the market has walked away from, there’s roughly a 92% chance it’s still sitting there next month. Waiting isn’t neutral. It’s a decision with a measurable failure rate.

And it means the honest version of the tool’s claim is modest: a re-comp doesn’t perform magic on stock. It moves an item from the 8% group into the 50% group, by making its price true again. Most of the book doesn’t need moving — the answer for most listings is hold. The value is in finding the handful that do, and not spending your evening on the rest.

We’ll publish the next thirty when we have them.