Pricing Guide

What this is

A pricing guide built entirely from your own selling history — every sale across eBay, Poshmark, Depop and Etsy, plus everything currently listed. Paste a generated title into the box on Price a draft and it finds what comparable garments actually fetched and suggests a list price.

It is not a market research tool. It never tells you what a garment is worth in general, only what your stock has sold for. That is the point: it is experience written down, not a guess.

Pricing a draft

Paste the whole nifty title. Do not trim it — the grammar is what the app reads:

[Era] Brand [Model] Type Gender Size Colour [features]

The gender marker anchors everything: brand and item type are read to its left, size immediately to its right. Press / from anywhere to jump to the box, Esc to clear it.

Reading the answer

• List at is the suggested price. It is the median of what comparable items sold for, divided by 0.80 — because you list high and discount. Taking that percentage off it lands back on what things really fetched.
• Typically sells is the raw median of the sold comps. This is history, not a target.
• Most sell between is the middle half of them. A wide band means treat the number loosely; a tight one means trust it.
• Matched on tells you how hard the app had to work. brand+type is the tightest and best. brand or type alone means it widened to find enough history. text means it found nothing structured and fell back to word matching — treat those numbers with suspicion.

What actually moves the number

Measured by stripping parts out of 300 real titles and watching the suggestion move:

So the four structured facets do nearly all the work. If a number looks wrong, check that the brand and type were detected correctly before anything else — the detected values are printed under the box.

When to distrust it

• Only one or two sold comps. The count is shown next to the median. One sale is an anecdote.
• Band tees and anything collectible. A rare band shirt comps against generic ones and will be badly underpriced. The app cannot see this yet — band names are not treated as brands. Price those by hand.
• Matched on text. Nothing structured was recognised.
• Comps mostly from Poshmark or Depop. Said under the price when it happens. Those are adjusted to an eBay-equivalent figure, but the adjustment is an average, not a fact about that garment.

Listing age

Everything live, oldest first. Age is measured from the first time the SKU was listed, not the current listing — the relist tool recreates listings nearing 90 days, which resets eBay's own clock. A garment sitting eight months would otherwise read as three weeks old.

The tiles double as filters. Read shows views, watchers and impressions where eBay reported them: live 60+ days with no watchers is the strongest "something is wrong" signal available — usually price, photos or title.

Sales

Daily revenue, net of platform fees. Pick a range, pick what to compare against, pick a platform, and click any day to list that day's orders.

• Platform scopes the whole view, not just the stacked chart — KPIs, chart, comparison and the day drill-down all follow it. Pick All for everything, or one platform to see it alone. Clicking a channel in the legend does the same thing; clicking it again goes back to all.
• Net is after platform fees; Gross includes the shipping the buyer paid. Net is the honest one and the default.
• Comparison periods shift by whole weeks so Mondays compare to Mondays. A 30-day range compares against 35 days earlier, not 30, because shifting by 30 slides the calendar by two days.
• If a comparison window reaches back before your history starts, no comparison is shown rather than a misleading one.
• Units equals orders — the export is one row per item and carries no quantity, so the two always match.

Issues

Things costing money right now. The red banner only appears for the ones where a buyer can order something you cannot send.

• Sold on another platform, still live on eBay. The urgent one. eBay has no way to know, so it will never end that listing on its own.
• Sold but still listed. The same problem from an eBay sale.
• Listed on eBay, no longer in stock. Gone from your nifty inventory with no sale on record anywhere.
• Sold, but still in stock and in stock, not listed — bookkeeping rather than danger, so no banner.
• One SKU, two garments. The most damaging even though nothing is mispriced today: age tracking, relist chains and every sold-but-listed check assume a SKU means one garment.

Mark fixed hides a row once you have dealt with it. That is saved for everyone, so what one of you clears the other sees, on any device. Nothing is deleted — the count stays and one click brings them all back. Most clear themselves anyway once the listing is ended and you next sync. The same applies to the specifics audit and the stale worklist. If the page ever cannot reach the server it says so and keeps the mark on that device only, so a click is never lost.

Where the numbers come from

Keeping it current

Export from nifty, then run one command. Neither import needs a token or the internet.

npm run daily

That picks up the newest export in your Downloads folder, merges sales and inventory, and rebuilds this page. Overlapping exports are safe — re-importing the same year merges nothing, so export whatever range is easiest.

Then weekly, with a fresh eBay token pasted into .env.local:

npm run refresh

That is the only part that needs a token. It refreshes live listings, watchers and cancellations — and it is what keeps the Issues checks accurate, because they compare against the last listings snapshot. Skip it too long and those checks go quiet rather than wrong.

What it cannot tell you

• Anything about a garment you have never sold something like. It says so rather than guessing.
• Condition. Neither feed carries it.
• Seasonality — not yet. The framework is running and reports its own progress at the foot of the Sales view, but a month needs to be seen in two different years before it can move a price. Roughly another year.
• Whether a price is right. It reports what happened. A garment that sold in two days was probably underpriced, and the app cannot tell.

If something looks wrong

• The suggested price looks mad. Check the detected brand and type first, then the sold-comp count, then whether it says text.
• An alert is about an item you never sold. Likely a reused SKU — check the "One SKU, two garments" section.
• Numbers look stale. The header carries two dates. Built is when this page was generated; eBay data appears beside it when the listings are older than that. npm run daily refreshes sales from the nifty export and rebuilds, but it does not talk to eBay — only npm run refresh does, and that needs a fresh token.
• A sync fails with 401. The token expired. Fetch a fresh one; it does not mean anything is broken.