Diagnosing an organic traffic drop

Most traffic drop advice starts at the wrong end. Somebody notices sessions are down 30 percent, opens Twitter, finds a thread about an update, and the diagnosis is finished before it started. Then three weeks disappear into rewriting content that was never the problem.

The order below is deliberately unglamorous. Tracking first, because it is the most common cause and the cheapest to rule out. Algorithm updates fourth, because they are the most frequently blamed and the least frequently responsible.

Split the drop before you diagnose anything

Do this first, before any of the eight steps. A single sitewide sessions number is the reason most SaaS teams misdiagnose a drop, because it averages two completely different populations with different causes and different business consequences.

Open Search Console, go to Performance, Search results, then Queries, and add a filter of Query, Custom (regex). Run these two cuts over the same date comparison:

Informational: (^|\s)(how|what|why|when|guide|tutorial|template|examples?|ideas|tips)(\s|$)
Commercial: (vs|versus|alternative|alternatives|pricing|cost|review|reviews|best|software|tool|platform|integration)

Export both. Record impressions, clicks, average position and CTR for each cut, for the loss window and the equivalent preceding window. Those eight numbers determine which of the eight causes below is even plausible, and they take about ten minutes to produce.

Remember that Search Console holds 16 months of performance data. If your drop is a slow year-long decline rather than a step change, export monthly now, because the early months age out of the interface while you are still investigating.

The eight causes, in elimination order

Causes one and two: tracking, then indexation

These two account for more sudden double-digit drops than anything else, and together they take under thirty minutes to eliminate.

Tracking first. Compare Search Console clicks with GA4 organic sessions over the same window. If Search Console is flat and GA4 fell off a cliff, no traffic was lost. Something changed in measurement: a consent banner default flipped to opt-out, a tag fired on fewer pages after a template deploy, a bot filter started working, or somebody added a second GA4 container. Check the deploy log for the drop date before anything else.

Indexation second. Open Indexing, then Pages, and read the status counts as a time series rather than a snapshot. The three that matter are Excluded by noindex, Blocked by robots.txt, and Crawled currently not indexed. A step change in any of them on your drop date is your answer, and the cause is usually a staging configuration promoted to production.

If large chunks of a templated page set stopped indexing rather than the whole site, the diagnosis path is different and specific to scaled pages. The failure statuses and what each one means are broken down in our programmatic page indexation study.

Cause three: was there actually an update on your dates

Write your exact loss dates down before you look at any update list. Then compare. The order matters because reading the update list first makes everyone find a match.

An algorithmic cause has a recognisable shape: the decline starts within a few days of a confirmed update rollout, it affects a coherent group of pages rather than everything, and it does not recover on its own. A drop that begins on a Tuesday in a month with no confirmed update, affecting your whole site evenly, is almost certainly technical or measurement related.

If it is genuinely a core update, the honest timeline is uncomfortable. Meaningful recovery usually needs the next core update to reassess your site, which is typically two to six months out. The teams that recover prune and consolidate. The teams that publish forty more posts of the same kind mostly do not, for reasons set out in why SaaS SEO fails.

Causes four and five: SERP layout and AI Overviews

These two travel together and they are the defining diagnosis of 2026. The signature is specific: impressions flat or rising, average position stable or improving, CTR down sharply, clicks down with it.

Search Console does not currently break out AI Overview impressions or clicks as a separate filter. They are folded into the Web search type total. That absence is why the position-versus-CTR comparison is the practical method rather than a workaround, and it is worth checking the Search Console documentation each quarter in case that changes.

Pull the top 20 losing queries and look at each live SERP yourself, logged out. Count what sits above the first organic result: an AI Overview, a video carousel, four ads, a People Also Ask block, a product grid. Then compare against an archived screenshot if you kept one, or against a lower-volume sibling query that has not changed.

Two things follow, and they point in opposite directions. Informational click loss to AI answers is mostly unrecoverable through ranking work, because the click is not going anywhere else. Commercial click loss is a different matter, since buying queries still send people to vendor pages and a comparison or pricing page losing clicks is a real ranking or relevance problem worth fixing.

What changes is measurement. Your informational pages are now doing citation and brand work you cannot see in sessions, which means the KPI has to move too. The methods for instrumenting that, including prompt set construction and sampling, are in measuring AI search visibility. Re-pricing the remaining traffic honestly is worth doing at the same time, and the SEO traffic value calculator will give you a defensible number for what the sessions you kept are actually worth.

Cause six: cannibalisation from your own publishing

Check what you shipped before you blame anything external. Cannibalisation is self-inflicted, it usually starts within 60 days of a publishing push, and it is the cause teams find last because nobody wants to look there.

The diagnostic is quick. In the Search Console Performance report, filter to a query that lost clicks, then open the Pages tab within that filtered view. If two or three URLs alternate as the ranking page across the period, you have it. A single stable URL with a falling position is a different problem.

SaaS has a second, structural version of this that most sites never look for: your help centre, docs site, community forum and changelog competing with your marketing pages for commercial queries. A support article titled how to set up SSO outranking your security page for an evaluation query is a conversion problem disguised as a ranking win. The intent split and the indexation policy that fixes it are covered in documentation SEO for SaaS.

The fix for true cannibalisation is consolidation, not deletion. Pick the URL with the strongest link profile and conversion history, merge the unique substance from the others into it, then 301 the weaker URLs. Expect four to eight weeks before the merged page stabilises, and expect a short dip first.

Causes seven and eight: links, redirects, seasonality and contraction

Link loss is slow and undramatic, which is why it escapes detection. Pull referring domains month over month in Ahrefs or Semrush and look for a step change rather than a trend. The usual culprits are a partner redesigning their site and dropping a resources page, a directory going paid, or a guest post being unpublished.

Redirects deserve a separate pass after any migration, rebrand or CMS change. Crawl your top 200 URLs by historic clicks and look for 404s, redirect chains of three or more hops, and redirects pointing at a category page rather than the equivalent page. That last one quietly destroys rankings while every link technically still resolves.

Seasonality is the cause everyone dismisses and it is real in B2B SaaS. Compare the same calendar window against two prior years, not against last quarter. Late December, mid August and the weeks of the big category conferences produce reliable troughs. If the same shaped dip appears in both prior years on the same dates, write one line in the report and move on.

Market contraction is the hardest conversation and the one worth having. If your category’s overall search demand is down 20 percent year over year, your organic sessions will fall no matter how well the programme runs. Check category head term trend over three years before you accept an internal diagnosis that the content got worse. Where your numbers sit against comparable companies is the other half of that answer, and the segmented figures in our SaaS SEO benchmarks are built for exactly this comparison.

A worked case: 38 percent fewer impressions, flat demos

A B2B SaaS company at roughly 28,000 dollars ACV, in a workflow automation category, came to us with organic impressions down 38 percent and clicks down 24 percent year over year. The board deck had a slide titled SEO recovery plan. The plan was wrong.

The two-cut split took fifteen minutes and changed the conclusion completely.

Average position on informational queries improved slightly while CTR fell by nearly 40 percent. That is not ranking loss. That is an AI Overview sitting above results the company still ranked well for. Commercial queries, where the pipeline comes from, were essentially unchanged. Demo requests were flat to slightly up.

Three changes followed. The dashboard was rebaselined around non-brand commercial clicks, trials and organic-sourced pipeline, with total sessions demoted to a secondary chart. The informational set moved to a citation-share measurement instead of a sessions target, sampled monthly across a fixed prompt list. And roughly 30 percent of the editorial budget shifted from new informational posts to integration and comparison pages, since that was the only segment where clicks still converted at a defensible rate.

What to report, and what to change this month

Report the split, always. One slide with three lines: commercial clicks, brand clicks, informational clicks, each indexed to the same starting month. A single organic sessions line will get you asked the wrong question by somebody senior, and you will spend a quarter answering it.

Attach money to the commercial line. Recomputing content-sourced CAC and payback with the current numbers, using the SaaS SEO ROI calculator, turns the conversation from a traffic chart into a financial one. Then rebuild the forward projection from where you actually are, since your old forecast assumed a CTR that no longer exists. The organic traffic forecast calculator takes the revised CTR as an input for exactly this reason.

One last note on proportion. A programme built on pages tied closely to the product tends to lose far less in this environment than one built on general advice content, which is the pattern running through both our SaaS SEO work and the Ahrefs product led SEO teardown. If your diagnosis lands on AI Overview compression of informational traffic, the durable response is a shift in what you publish, not a rescue mission on what you already published.