Correction — 12 September 2026: The original article counted a cancelled order among four sales and attributed outcomes to repricing. The corrected measure is three retained-sale listings out of eight. We have removed the unsupported revenue and causal claims, and the suggestion that traffic alone identifies a price or photography fault. The original URL and publication date are retained.
As a store grows, remembering which listings need another look becomes harder. An older listing can be a useful starting point, but its age does not tell you why it has not sold or whether a discount would help.
For our own Still In The Box AU store, the useful question is: what evidence would justify changing this listing? We are building TurnGoods around recording that decision and checking what happens afterwards.
What our own comparison recorded
Eight listings were repriced on 3 August 2026. In the 30-day window ending on 2 September, with that end date excluded, the saved 11 September calculation recorded:
| Group | Listings | Listings with a qualifying retained sale |
|---|---|---|
| Repriced on 3 August | 8 | 3 |
| Observational comparison | 103 | 11 |
One order counted in our first article was cancelled. It does not qualify as a retained sale, regardless of the reason for cancellation. A qualifying outcome here is a listing with at least one transaction retaining positive merchandise revenue after recorded reversals; it is not seller profit.
The groups were not randomised or fully matched. The comparison uses recorded price, age and markdown fields, and historic availability and other interventions are incomplete. One repriced item was above the comparison group’s price cap. With eight repriced listings, one outcome changes the rate by 12.5 percentage points.
The counts describe what was recorded. They do not show how many sales a price change caused, or promise a result for another store. The corrected comparison explains the selection rules and changed counts.
Traffic is a reason to investigate
The saved 11 September inventory contained 419 live listings. Traffic matched 167 of them, or 39.9%. The remaining 60.1% was unknown to this analysis, not zero traffic.
Using chosen review thresholds, 62 matched listings had at least eight views and zero reported conversion. A further 18 had at least 300 impressions and no more than three views. These are investigation queues drawn from one returned traffic payload, not a representative diagnosis of the whole store.
Views without a sale can reflect price, shipping, condition, variant, trust, timing or competition. Low click-through can prompt a review of the title, main image and offer, but it does not isolate a fault in any of them. Price may also affect whether a buyer clicks.
The traffic request covered 14 days. That differs from the historical comparison’s 30-day outcome window and from each listing’s age. Keep those periods separate when interpreting a report.
A review a seller can act on
- Check the record. Confirm the exact item, variant, condition, quantity, availability and current asking price. Label missing or uncertain dates.
- Gather relevant evidence. Compare genuinely comparable sold items where available, noting condition, shipping and how recent the evidence is. An asking price is not a completed sale.
- Consider more than price. Review the listing’s title, photographs, shipping terms and competition alongside traffic coverage. Record what is unknown.
- Choose and record the decision. A price change, a listing correction, a bundle or no change may be appropriate. Keep the reason, original state and seller approval attached.
- Check the outcome after a defined period. Include non-sales, cancellations, refunds and concurrent changes. Retained merchandise, seller profit and time saved are different measures.
A review does not need to result in a discount. Equally, we cannot claim that most listings should be held unchanged without evidence from the reviewed pool.
What TurnGoods can currently demonstrate
We have recorded inventory and historical outcomes from our own store, and built tooling to preserve evidence and listing decisions for a prospective review. We have not yet measured a prospective uplift or an external seller outcome. A consenting second seller’s isolated audit is the next test of whether this workflow is useful beyond our own stock.
Older stock remains worth reviewing, but it is one part of the book. In the same saved 11 September snapshot, listings aged 90 days or more accounted for 4% of recorded asking value. The inventory-age analysis explains what that means and what it cannot tell us.
The aim is to make each recommendation inspectable: what we observed, what we inferred, what the seller chose and what happened next.