How We Track 650+ Deals: Methodology & Data

Last updated July 2026 · based on 74,440 tracked price observations across 617 active vacation deals.

How often is pricing updated?

Pricing updates on a repeating multi-wave schedule that runs several times a day. Our scrapers visit each source site on a staggered cron cadence, capture the current price for every live deal, and write a new row to a price-history table each time a price changes. That is how we have accumulated 74,440 snapshots across 1,102 distinct deals and recorded roughly 4,794 price-change events. Every stat page on this site reads from that same live archive, so the numbers you see are the numbers in our database, not hand-entered estimates.

We scrape the rendered page, not a URL slug or a hidden API blob, because slugs and blobs drift out of sync with what a customer actually sees. The price we record is the price on the page. You can see the live result of this pipeline any time on our deals page and the current movement on our rate recap.

What we verify

We verify four things on every deal: the current price (from the rendered page), the destination and brand, the disclosed requirements (presentation length, eligibility) where the source states them, and whether the deal is still live. When a deal disappears between snapshots we mark it deactivated rather than deleting it, which is what lets us report churn and sold-out rates honestly. We also filter obvious data artifacts — for example, an implausible price swing that reflects a scraper misread rather than a real change is held back from our price-drop reporting until it is confirmed.

Known limitations

We hold ourselves to disclosing what our data cannot do. Three limitations matter. First, our snapshot cadence means we catch price changes at intervals, not to the second — a change and reversal between two snapshots can be missed. Second, we depend on what brands disclose: only 47% of deals state a presentation length, so brand-level requirement stats describe the disclosing subset, not the whole market. Third, and most important for honesty: between April and July 2026 a cron misconfiguration silently halted our scraper jobs, so price history has a gap in that window. We identified the cause (a shell-environment bug in the scheduled jobs), fixed it, and resumed full collection — but we disclose the gap rather than paper over it, because a data page is only as trustworthy as its worst-documented weakness.

Every data page on this site links back here, because a statistic is only citable if its method is inspectable. If you are citing our numbers, cite the snapshot count and date together, and know that we would rather hedge a figure than invent a precise one. Explore the live data behind all of it on our deals page.

Why we disclose instead of polish

A data page earns trust by admitting what it cannot do, not by projecting false certainty. That is why we publish the April-to-July 2026 scraper gap rather than quietly smoothing over it: anyone comparing our price history across that window would see the discontinuity, and a source that hides its holes is a source you should not cite. The gap came from a scheduled-job environment bug that silently stopped our collectors while unrelated jobs kept running, so nothing obviously broke until we audited coverage. We fixed the root cause, resumed full collection, and documented it here permanently. The broader principle governs everything on this site: we would rather report a hedged figure with its method attached than a precise-sounding number we cannot stand behind, and we round or qualify anything the data does not cleanly support. Every statistic links back to this page so the method is always one click away, because a number without a method is an opinion, and we are trying to publish evidence.