Methodology
How we verify
Most “70% OFF” banners are measured against a price nobody was paying. Every discount on this site is checked against independent evidence before we call it anything — and when we can't verify it, we say exactly that instead of amplifying it.
The three tiers
Every offer carries exactly one of these, computed from evidence — never assigned by hand:
✓ VERIFIED two or more independent evidence sources corroborate the discount.
CORROBORATED exactly one independent source backs it.
CLAIMED the retailer's word only. Labeled amber, scored at half weight — a fat fake discount cannot outrank a modest real one.
The evidence sources
Five kinds of evidence can back a discount. Not all of them produce data yet — and every phrase a source powers stays locked until that source can actually back it:
- Discount-culture snapshots (live) — how deeply each retailer routinely marks down, measured across hundreds of its own sale SKUs. This is what backs today's “corroborated” labels, and the chip says so in as many words.
- Our own tracked history (accruing) — every price we observe is logged permanently. Long-horizon claims (“lowest we've tracked”) unlock only after 60 days of continuous coverage.
- Cross-store price checks (as matches accrue) — the same product, matched by style code, priced across every store we track. Most of the current catalog is single-store brands, so these labels appear product-by-product as matches do.
- Independent typical-price ranges (not yet wired in) — what the market actually charges, from Google Shopping's aggregated data.
- Resale floors (not yet wired in) — for sneakers, the licensed secondary-market floor as a sanity check on “retail value.”
Was-prices: the strikethrough rule
A struck-through former price ($148) appears only when independent evidence supports that the product actually sells around that price. When it doesn't, you see the amber chip — “retailer claims $148” — and the claim plays no part in the deal's rank beyond its half-weight tier. This rule is enforced in the rendering code itself: the layout is incapable of striking through an unverified claim.
Every phrase has an evidence bar
| Phrase | Evidence required before it can render |
|---|---|
| “−38% vs typical $128–152” | An independent typical-price range (Google Shopping) or a cross-store price check on the same style code across 3+ stores, no older than 7 days. |
| “cheapest of 6 stores” | At least 2 matched stores, every price checked within the last 24 hours. |
| “lowest we've tracked since [date]” | Our own price observations for that offer covering at least 60 days, and the current price actually being the minimum of the series. |
| “retailer claims $148” | Nothing — this IS the unverified state, and it says so. It renders in amber and is never shown as a strikethrough. |
| “deeper than 78% of Nike's own markdowns” | A discount-culture snapshot of 200+ sale SKUs for that retailer, no older than 30 days. |
| “your size: 3 left” | A size-level stock read from the store within its freshness window, with the read time shown on the card. |
The symmetry rule
Some links on this site may earn us a commission. Identical math and identical labels apply to affiliated and non-affiliated merchants. We don't suppress a bad verdict on a store that pays us, and we don't boost one for paying.
Freshness
Every card shows when its price was last checked. The most visible deals — the top of every feed and hub — are re-checked about every half hour, and the full catalog is re-snapshotted several times a day. An offer that goes unconfirmed past its re-check window automatically loses score until it's checked again: stale deals sink on their own. There are no countdown timers and no manufactured urgency. Stock counts are real size-level reads from the store, taken in the same check the card's timestamp reports, and that stamp is the only time pressure on a card.
The score, decomposed
The 0–100 deal score is six public components: real discount depth (up to 40, and only as verified as its evidence), price position across stores (10), rating confidence (15), size availability (15), retailer quality (10) and momentum (10). Every card's “why?” opens the exact per-component breakdown — no mystery numbers.
Corrections
If we've labeled your price wrong — merchant or shopper — we want to know. Email hello@priceandthread.com (or see the contact page); merchant disputes get a 72-hour response target, and receipts studies carry versioned changelogs.
This page mirrors the rules enforced in the site's rendering code. Back to the feed →