A sporting goods retailer opens this file every Monday at 7am. 2,000 SKUs across 6 competitors. They filter by price gap, reprice 340 SKUs before lunch, and close the file. Here is what that file looks like.
SKUs tracked
competitor sites
completeness
accuracy
↑ Row 7: Limited stock flagged — not just "In Stock" vs "Out." Row 16: Backordered — different from out-of-stock. Row 17: Product not found — flagged explicitly, never blank. Row 18: $0.00 anomaly — flagged as probable scraper error, excluded from analysis, queued for same-day fix.
(scraper error — excluded from file, queued for fix)
4 products collected on retry, no data loss.
The $0 anomaly was caught before it reached your BI dashboard. The 503 error was retried automatically — no data loss, no manual intervention. In a self-service tool, that row corrupts your average and the site outage means missing data until someone notices.
Real data from a NZ pet retailer. Same products, but each site offers different pricing tiers:
PetStock shows $65.90 on-page. The subscription price is $56.02. If you track headline prices only, you are comparing against the wrong number.
When TennisWarehouse delists a product mid-cycle, the row does not disappear or show blank fields. It is flagged: "Not found on site." We distinguish "delisted," "out of stock," and "scraper error" — three different situations requiring three different responses from your pricing team.
A rug manufacturer needed to know which retailers were undercutting RRP — and which were compliant. We tracked 4,729 SKUs across 8 UK retailers for 8 months. This is from their actual January 2026 enforcement report.
largest gap found
price changes
retailers tracked
SKUs monitored
↑ Highlighted: £754 gap — Form Wool Green at BM (£644.76) vs RRP (£1,399). Same product, same month, same report. Compliant retailers like John Lewis and Heal's are tracked alongside violators — the report shows the full picture, not just problems.
Every violation includes a timestamped screenshot showing the advertised price, URL bar, and capture time:
beddingmill.co.uk/asiatic-form-wool-green-200x290
therugshopuk.co.uk/asiatic-gatsby-blue-200x290
marksandspencer.com/asiatic-tate-grey-200x290
Retailers cannot claim "we never charged that." This evidence has been used in actual enforcement actions.
Name + storefront ID
Code + name + size
Manufacturer price
After all discounts
Exact £ below RRP
Live link
ISO 8601
S3 link, timestamped
First-time vs chronic
8 consecutive months below RRP. The price fluctuates — £560 one month, £829 the next — but never reaches RRP. This pattern turns "we will look into it" into a cease-and-desist.
One workwear brand found 700 unauthorized sellers over 4 years. They were hiding in the 40% coverage gap their previous vendor could not reach.
A luxury marketplace's account team uses this before every seller meeting. Instead of "please add more products" they say: "You have 29 Gucci products. Average seller has 258. Here's which categories you're missing."
sellers
brands
products/wk
largest gap
↑ Highlighted: Boutique Tricot — 29 vs 258 average. That gap is the conversation your account team now has with exact numbers.
876 clothing items, zero bags. Not a strategy decision — a category expansion opportunity. One marketplace used this data in seller meetings: assortment went from 50% to 98% completion.
Without monitoring, nobody notices TOPS! Lithuania vanished for weeks. Your account team knows Monday morning.
A luxury fashion marketplace needed to match 2,068 Farfetch products against the Isabel Marant US catalog. 6.8 million candidate pairs were scanned. 3,795 were evaluated with text + image verification. 632 exact matches were confirmed. Here is what that process looks like.
pairs scanned
candidates evaluated
exact matches
accuracy after QA
↑ Highlighted: Oskan Moon at 90% — same product line, different color (camel vs taupe). Dalby Boots at 95% — same silhouette, different shade. Both classified as color variants, not exact matches. Row 8: Duerto boots at 60% — same model name but different material and design details. Automated systems accept 60%+ matches. Human QA catches the difference.
Product names, brand, model. Fast but cannot distinguish accessories from base products or color variants.
Visual product photo comparison. Catches color and model variants that text matching misses.
Size, weight, specs, material. Catches near-misses like suede vs leather versions of the same bag.
Every match 60–98% goes to human review. The 90% Oskan Moon was correctly classified as a variant here — not a false exact match.
50.4% had no match — products on Farfetch that Isabel Marant US doesn't carry. That gap is itself a competitive insight: which products are available through resellers but not the official US store.
"Oskan Moon suede shoulder bag" and "OSKAN MOON shoulder bag" score 90% on text similarity. Automated systems call that a match. But the Farfetch version is camel brown suede. The Isabel Marant US version is taupe leather. Different color, different material — a variant, not an exact match. If your pricing decisions depend on correct matches, automated matching silently corrupts your price comparisons.
A workwear brand tracked Buy Box ownership across Amazon US. The question is not just "who is winning" — it is "why are they winning, and what is the price gap I need to close?"
products tracked
seller listings
smallest winning gap
unauthorized sellers
↑ Highlighted: WorkwearDirect is $0.84 cheaper but does not hold the Buy Box. Price is not the only factor — seller rating, fulfillment method, and stock history matter too.
A single snapshot tells you who is winning. Weekly tracking tells you patterns: which sellers consistently undercut, when ownership shifts, and whether unauthorized sellers are gaining ground. One brand found that 3 unauthorized sellers were rotating Buy Box wins by cycling prices — a pattern only visible with historical data.
A wholesale apparel distributor tracks size-level pricing across 11 competitor sites. Same product, same color, different sizes — and prices change at the size break. If you track at product level, you miss the variant-level competition.
competitor sites
size variants
largest size premium
delivery cycle
↑ Highlighted: BlankStyle is cheapest at every size — $23.02 (base) vs JiffyShirts $26.25. That is a $3.23 gap at S–XL that widens to $5.20 at 3XL.
JiffyShirts and ShirtMax do not carry 3XL in this style. BlankStyle and ShirtSpace do. If you sell 3XL, only 2 of 4 competitors offer it — that is pricing power.
Product-level pricing shows "Pullover Hoodie: $26.25." But the real competitive picture is: $26.25 at S–XL, $30.29 at 2XL, $33.32 at 3XL. A competitor who is cheaper at S–M but expensive at 2XL+ has a different strategy than one who is cheaper across all sizes. Variant tracking surfaces this — product-level tracking hides it.
Some wholesalers offer quantity discounts per size. Real data from Needen:
XL premium is $1.16 at qty 1 but only $0.99 at qty 576. We capture every tier the site offers.
Every deliverable includes
You define the fields. Column names do not change without sign-off. 3+ business days notice. Version tracking on every change.
CSV · Excel · JSON · Google Sheets · S3 · BigQuery · Snowflake · Redshift · SFTP · API
