Why Every Deal Carries Two Prices: Building a Verification Pipeline Shoppers Can Trust
Why Every Deal Carries Two Prices: Building a Verification Pipeline Shoppers Can Trust A price is only a deal if you can prove it — and a stale proof is worse than no proof at all Every product guide on Savl
Why Every Deal Carries Two Prices: Building a Verification Pipeline Shoppers Can Trust
A price is only a deal if you can prove it — and a stale proof is worse than no proof at all
Every product guide on Savlume lists two prices for the same item: the price we saw today, and the price we saw before. That is not decoration. It is the visible half of a verification pipeline that exists because deal prices are the least trustworthy numbers on the internet — and because a price that was real last week is the most convincing way to sell a fake discount today.
The problem: one price, three different truths
A retailer listing can show three prices at once: a list price, a sale price, and a price with a coupon applied. Add a price-history feed, a press release, and a social post quoting last month's deal, and the same product can plausibly be "$129.97", "$89.97", or "$24.99" — all with a source attached. Our portable tire inflator guide is a live example: the Fanttik X8 Apex is $129.97 today, but $89.97 on November 6. Both numbers are real. Only one is the price you would pay right now.
So we stopped publishing a single price. Every offer carries a status — retailer-verified or source-reported — and a date. Retailer-verified means a check confirmed the price on the retailer's own page. Source-reported means the price came from a feed, a listing snapshot, or a third-party report and has not been confirmed at the retailer today. The status is not a judgment about the retailer; it is a statement about our evidence.
How the pipeline works
1. Ingest with provenance. Every price enters the system with its source, its timestamp, and the exact conditions attached to it — promo code, first-order eligibility, membership requirement. A price without conditions is a bug: the $11.00 Woot inflator in our guide is only $11 with a first-time account and code INFLATE, and storing it without those conditions would manufacture a deal that does not exist for most shoppers.
2. Verify against the retailer, not against the feed. Verification means fetching the retailer's own page — or a retailer API response — and reading the price a shopper would actually see. A feed that agrees with last week's feed proves nothing. This is the most expensive step in the pipeline, so it runs on a schedule weighted by volatility: prices that move often get checked often, stable prices less so.
3. Keep history, don't overwrite it. When a new price arrives, the old one becomes history, not garbage. That is what lets a guide say "it was $89.97 on November 6" instead of silently pretending $129.97 is a discount. History is also how we catch inflated reference prices: if a product was never actually sold at its "original" price, the discount percentage is fiction, and the history proves it.
4. Separate the summary from the source of truth. Our guides render a comparison table, a current-offers section, and a Q&A — and they can disagree. The table is a summary; the current-offers section is the source of truth. The pipeline enforces one rule: the current-offers section is generated from the latest verified or reported record, and every other section is derived from it. When a table still shows a November 6 price as current, that is a rendering bug we have to fix, not a pricing decision. We found exactly this in the tire inflator guide this week and flagged it in the article itself rather than quietly editing the number.
5. Expiry is a feature. Prices carry a recorded date, and anything older than its freshness window is labelled as past, not current. A deal site that never expires a price will eventually show a bargain from three months ago as if it were live. Our feeds — the RSS feed at /feed/all.xml and the API at /api/v1/items — include the verification timestamp so downstream readers and AI agents can apply the same rule.
What this costs, and why it is worth it
The pipeline costs us crawl budget, storage for every price snapshot, and editorial time spent writing "this was the price on November 6" instead of "now only $89.97". What it buys is the one thing a deal site cannot borrow: trust. A reader who catches us showing a stale price as current has no reason to believe the next price either.
The deeper lesson is that price data is not a fact; it is a claim with a timestamp and a source. Any system that treats it as a single number — a scraper, a feed, an AI shopping assistant — will eventually publish a deal that does not exist. Build the pipeline so that the evidence travels with the number, and the number stays honest even when it goes up.
More on how Savlume works: https://savlume.com/
Follow Savlume Deals on WhatsApp: https://whatsapp.com/channel/0029VbE9P7J7DAX7rD8TAC2U
Originally published by Dev.to AI. Aggregated on AIWithGhost for educational purposes — full credit and traffic to the original publisher.