The $17.99 laptop: how parse-guard saves you from scam alerts
A real-world look at one of the noisiest failure modes in retail price monitoring, and the 5%-of-list floor that fixed it.
The first version of Priceglitch's retail scraper had a class of bug that kept showing up in user reports: a $17.99 ASUS laptop, a $10 LG TV, a $0.99 KAIGERR something. The price was technically what the scraper saw — but the scraper was looking at the wrong cell.
Where these come from
SerpAPI's google_shopping engine returns a carousel of related items for any query. Bigger ticket items get a sponsored sidebar of accessories. The carousel for a $1,199 laptop reliably has a $17.99 charger or a $10 sleeve as the second cell. Our parser took the first numeric value and shipped it. The detector saw a 98%-off drop and auto-fired. Users complained.
The 5% floor
Every scraped retail price is now gated by a parse-guard: if the target has a `list_price_usd` set and the scraped value is below 5% of it, we reject the scrape outright. We log it as a parser error on the target row (not a real failure — no auto-pause counter advance), then move on. Real penny glitches on Home Depot don't have a list_price set, so they're unaffected. Real deep-discount deals (60%+ off) clear the floor by a wide margin.
Why 5%? A genuine retail glitch is in the 70–99%-off range. A 60%-off floor is unusually deep but plausible. A scraper picking up an accessory cell is almost always at < 5% of the list price. The guard sits exactly where the histogram of legitimate deals ends and where the histogram of scraper errors begins.
What you'll see in your dashboard
When the guard trips, the affected target gets a `last_error: parse_guard: rejected $X.XX (< 5% of list $Y.YY)` annotation. The target stays enabled and continues to scrape. The next clean scrape clears the error. If a known-good deal doesn't fire and you see this annotation, the cause is almost always a stale or wrong list_price — adjust it, let the cron run, you're back in business.
What we don't gate
Targets without `list_price_usd` (mostly Home Depot SKUs and Best Buy Google-catalog finds) skip the guard. These are the penny-error happy hunting ground — HD's clearance pages occasionally surface $0.01–$0.20 prices that are real, and we don't want to filter those out by accident.