Your search history, kept.
Google serves a rolling 16 months. Prism syncs your Search Console data once a day into its own store and keeps every day of it.
- pulled back on connect, then kept for good
- 16mo
- pulled back on connect, then kept for good
- days re-pulled every sync — Google revises recent numbers
- 5
- days re-pulled every sync — Google revises recent numbers
- sync a day, running on its own once a property is connected
- 1
- sync a day, running on its own once a property is connected
- credits — the warehouse never touches them
- 0
- credits — the warehouse never touches them
months is all Google serves. The window rolls: month 17 is dropped, not archived.
Google Search Console
Prism’s copy
today
Questions only history can answer.
These are the questions a rolling window quietly takes away.
- Year on year, in year three
Compare this March with the March two years ago — the comparison is still sitting there.
- A slide that took six months
A query losing two positions a quarter never looks like a drop in a 28-day view. Plot the whole thing and it does.
- Seasonality you can trust
Two Christmases is a comparison. One Christmas is an anecdote.
Keep your own copy, and Google’s limits stop being yours.
A tool that calls Google's API when you open a report inherits Google's limits — the quota, the row cap, and the one that throws your history away. Prism takes a copy once a day and answers from that.
- Stored in full detail
- Every query, on every page, split by country and device, day by day. True per-date totals are kept separately, because the sum of a row-capped list is not the total.
- Yesterday corrects itself
- Google finalises its numbers two to three days late, so every sync re-pulls the last five days.
- Backfilled on connect
- Connecting a property queues the whole 16-month window, one month per pass, and carries on until it is done.
How the copy is kept.
Your history starts at Google’s full 16 months and grows one daily sync at a time. A year in, you hold data Google no longer has anywhere.
Google budgets result rows per request, so each day is fetched as its own request. A day that fails is recorded and skipped, and the next sync picks it up.
Queries compared on what they mean, not the words in them — Keyword Clusters, Cannibalization and the AI ↔ SEO Bridge all run off it.
Rows are stored and queried per property and access is granted per user, so an agency and a client's own team can read the same roster without seeing each other's numbers.
Frequently asked questions
How detailed is the stored data?
Every query, on every page, split by country and by device, day by day. Prism also stores true per-date totals separately, because Google caps the rows it returns on a detailed query, and the sum of a truncated list is not the real total.
Why are yesterday's numbers more accurate here?
Search Console finalises its numbers two to three days late. Every sync re-pulls the trailing five days, so yesterday corrects itself as soon as Google catches up.
How far back does the backfill go?
Sixteen months, which is everything Google will serve. Connecting a property queues the whole window, one month per pass, with a cursor that survives a restart, so a large property backfills as a chain of small resumable jobs rather than one run that can lose everything.
Where does the data actually live?
On Prism's infrastructure, in a store Prism runs and syncs into once a day. Prism keeps the copy for you, per property, so your Search Console history stays readable after Google has dropped its own copy, and goes on growing for as long as the property stays connected.
How does it group queries that mean the same thing?
Every distinct query gets turned into a numeric representation of its meaning, and queries are compared on that rather than on the words in them. Keyword Clusters, Cannibalization and the AI to SEO Bridge all run off it.
Elsewhere in Prism
Built on the warehouse
- Search Performance 16 reports over it, from Opportunities to Cannibalization to Click Gap. Open a single query and the page-swap history reads the whole stored span, not the last 28 days.
- Reports The scheduled summary reads the warehouse, so the email and the screen cannot disagree.
- Actions Alerts are evaluated on every sync.
- AI answer monitoring The AI ↔ SEO Bridge joins your prompts to your real queries by meaning, and both live here.
The other four features
- Site crawling Prism crawls your whole site, scores each page with Lighthouse, and reads the X-Robots-Tag response header as well as the robots meta tag. Compare any two runs to see exactly what a release changed.
- AI answer monitoring Your prompts put to ChatGPT, Claude, Gemini and Perplexity on a schedule. Prism stores the complete answer, so every number on every report traces back to the sentence it came from.
- Smart Campaigns Define a campaign by URL and query patterns, give it a start and end date, and Prism tracks that whole group as one thing. The indexed view rebases every campaign to 100 so a big one and a small one are finally comparable.
- Segments A segment is a saved query group or URL group built from plain rules. Pick one in the toolbar and every Search Performance report stays scoped to it, so brand against non-brand is one click rather than a rebuild.
Nothing you do later can recover month 17.
Connect a property and the backfill starts on its own, pulling back everything Google still holds and keeping every day that follows.