How much time does a buyer need after a price drops? If approval takes 30 minutes, spotting a discount at 09:00 that ends at 09:10 gets it into the price history but leaves the buyer no time to act. Before paying for more frequent checks, count the promotions an extra check would have caught in time.
Use the bilingual replay worksheet to compare daily checks, six-hour checks and merchant notices. The four promotions, prices, timings and costs below are fictional. Replace them with records for a source you are permitted to monitor.
Fix the offer and the buyer’s deadline
Follow one blue AX14 SKU from merchant-7 in California, in USD, for a returning customer with standard delivery. For this replay, the page reflects the merchant’s declared price, all scheduled checks succeed, and each result reaches the buyer immediately. In a real pilot, record when a sale takes effect, when the merchant notice arrives, when the page is successfully observed, and when the buyer receives the alert. Use that last timestamp to calculate how much time the buyer has left.
All times are UTC. The 48-hour review begins at 00:00 on November 1, 2026, and stops just before 00:00 on November 3. In this example the price applies from the start up to, but excluding, the end of each interval. Google Merchant Center’s sale price effective date guidance shows how a merchant can supply start and end times with a time zone. A scheduled sale in a feed and a price actually shown to a particular buyer answer different questions.
Replay four promotions against the same clock
| Episode | Merchant-declared interval | Feed received_at | What it demonstrates |
|---|---|---|---|
| P1 | USD 80 [Nov 1 08:30, 11:00) | 08:40 | Timely feed delivery; both schedules see it at 09:00, but it counts once. |
| P2 | USD 75 [Nov 1 14:40, 16:10) | 15:55 | Feed is received with only 15 minutes left, too late for this 30-minute action rule. |
| P3 | USD 70 [Nov 1 19:00, 21:00) | 19:05 | At exactly 21:00 the end is excluded: base price applies. |
| P4 | USD 85 [Nov 2 05:30, 09:10) | 05:50 | Both schedules see it at 09:00, but only 10 minutes remain, so it is late. |
Consider P2. Its notice says the sale started at 14:40, but the team did not receive it until 15:55. Using the start time as the notification time would give the buyer credit for information they did not have. P3 illustrates the opposite trap: at 21:00 the regular price has returned, so a successful check at that instant misses the entire sale.
Did six-hour checks give the buyer more time?
List each check inside the review window. For each promotion, find the first successful matching observation at or after the start and strictly before the end. Count it as timely if the alert arrives with at least 30 minutes remaining. Multiple sightings count as one promotion. Assume a variable cost of $0.05 per page check.
| Plan | Observed_at values | Checks / assumed cost | P1 | P2 | P3 | P4 |
|---|---|---|---|---|---|---|
| Once daily | Nov 1 09:00; Nov 2 09:00 | 2 / USD 0.10 | TIMELY | MISS | MISS | LATE: 10 minutes remain |
| Every 6 hours | Nov 1 03:00, 09:00, 15:00, 21:00; Nov 2 03:00, 09:00, 15:00, 21:00 | 8 / USD 0.40 | TIMELY | TIMELY at 15:00 | MISS: 21:00 is excluded end | LATE: 10 minutes remain |
Daily checks see two promotions, but only P1 in time. Six-hour checks see three and give the buyer time to act on P1 and P2. The extra six checks cost $0.30 and add one timely discovery in this replay. Both plans miss P3. If the six-hour plan’s P2 alert takes until 15:45 to reach the buyer, only 25 minutes remain and that incremental gain disappears.
Decide what to fund next
- For the merchant’s announced offers, test the feed first. It records all four intervals here and delivers P1, P3 and P4 in time. Evaluate its delivery history against the actual decision deadline. A permitted feed that meets the need can be the whole solution.
- For the price displayed in a particular market, check the page. Match the variant, seller, location, customer eligibility and delivery terms. Record an inaccessible page, invalid result or mismatched offer as unknown. Without an independent event history, a collection of daily snapshots cannot tell you how many promotions fell between them.
- For a budget increase, replay another representative period. Compare incremental check and review costs with extra timely discoveries. With no additional timely discoveries, cost per additional discovery is undefined. Keep frequency within the merchant’s permitted request rate.
Use actual observation timestamps when testing a real schedule. Keepa’s update interval documentation, for example, describes why a requested update interval need not produce perfectly spaced observations. If the reference history is incomplete, report observed promotions and leave total coverage unknown.
If you need to confirm what a local buyer sees and the merchant permits that check, review residential proxy options. Bring the market, matching offer conditions, permitted check rate and required lead time when you ask IPHTML about fit.
For the underlying price comparison, see how to match comparable baskets. For a wider monitoring pilot, use the cost and benefit model.