Traffic is down, and Google announced a core update last week. The tempting move is to close the investigation right there. But "the update hit us" is a conclusion that should first survive three competing explanations — seasonality, a technical regression, and a change in the SERP itself — because each has a different fix, and two of them are fixable this quarter while a core update hit usually is not.
Search Console can separate these cases. It just has to be read with bounded dates and a control group, not eyeballed as one wiggly line.
Pin down the rollout dates first
Google publishes the start and completion of each core update rollout on its Search Status Dashboard. Rollouts run from a few days to a couple of weeks, and ranking movement can land anywhere inside that window.
That gives you the first test: alignment. Chart daily clicks and impressions and overlay the rollout window. A genuine core update hit begins inside the window — sometimes days after the announced start — and settles at a new, lower plateau once the rollout completes. A decline that began before the window opened, or a slow slide spread across months, is a different problem that happens to have an update nearby. And with updates announced several times a year, "nearby" is cheap evidence on its own.
Build a control set of queries
The most useful diagnostic move is splitting your queries into a treatment group and a control group before comparing any periods.
Brand and navigational queries are the control. Someone searching your company name will click your result regardless of how a core update reweights quality signals; those queries are close to immune to algorithmic reranking. Non-brand informational and commercial queries are the treatment group — they are where core updates act. A saved regex filter for brand terms, and its inverse, makes the split take minutes.
Now compare the two lines across the window:
- Non-brand down, brand flat. Consistent with an algorithmic hit. Ranking moved where ranking was contestable and stayed put where it was not.
- Both down together. Almost certainly not the update. Brand demand does not fall because a core update ran — look for a technical problem or a market-wide demand drop.
- Search Console flat, but sessions or revenue down. Your search traffic did not change. Your analytics, tagging, or conversion path did.
Use year-over-year, not month-over-month
Seasonality wrecks adjacent-period comparisons. Search Console retains sixteen months of data — enough to compare the decline window against the same weeks last year.
The tell is shape. If this year's curve tracks last year's curve — same slope, similar timing — you are looking at demand, not ranking. The confirming signal is impressions falling while average position holds roughly steady on a fixed query set: fewer people searching, not a worse position when they do. An algorithmic decline shows the opposite — position degrading while underlying demand looks normal against last year.
Read position, impressions, and clicks as separate instruments
Each failure mode leaves a different signature across the three metrics. Compare fixed date ranges on a fixed query or page set, and remember that sitewide average position is a mix-weighted number that shifts when composition shifts — filter first.
- Algorithmic reranking: position worsens on the affected queries; impressions then fall as you slide off page one; nothing changed about the pages themselves.
- Technical regression: impressions for affected URLs collapse toward zero rather than degrading, and the indexing report shows pages leaving the index. Average position can even improve, because only your strongest pages remain in the sample.
- SERP change or competitor move: position roughly stable, impressions stable, clicks down. Something above or beside you — a new feature, an AI answer, a stronger competing snippet — is absorbing clicks you used to get.
- Seasonality: impressions down, position stable, click-through rate stable.
Rule out the technical explanation directly
Do not settle this by inference when you can check it. Open the indexing report and look for a drop in indexed pages around the decline date. Put your deploy log next to the traffic chart — releases that touched templates, canonicals, robots directives, or rendering are prime suspects. Run a handful of affected URLs through the URL Inspection tool and confirm they are indexed, canonical to themselves, and rendering their content.
A technical break that coincides with an update window is common enough to check for every time, and it is the one diagnosis with a fast, complete fix.
Look at the SERP, not just your line
If position held but clicks fell, go look at the actual results page for your biggest affected queries. Note what sits above the fold now versus what your historical CTR implies used to be there. This is the one step Search Console cannot do for you, and it regularly explains "losses" that no amount of on-site work will reverse.
What to do with each diagnosis
- Confirmed core update hit: Google's documented guidance is that there is no single fix — updates reassess content broadly, and recovery, when it happens, tends to surface at later updates. Pick your most valuable affected pages and improve them substantively. Do not mass-rewrite in a panic; you will destroy your ability to evaluate what worked.
- Seasonality: annotate the date, stop investigating, and plan production around the peak instead.
- Technical regression: fix it, validate with URL Inspection, request recrawling, and expect recovery on a crawl timescale — days to weeks, not months.
- SERP change: you are now competing for the click, not the rank. Work on titles, snippet-worthiness, and the queries where the new SERP still leaves room.
The diagnosis is the deliverable. A team that knows which problem it has spends the next quarter very differently from one that only knows traffic is down.