Which pages and queries explain most of my organic decline?
Short answer
Sort by absolute clicks lost rather than percentage lost, then take the cumulative total until you reach 80% of the decline. On most sites that list is under twenty pages. Percentage sorting buries the answer, because a page falling from 12,000 clicks to 9,000 lost 3,000 clicks at a 25% decline, while a page falling from 40 to 4 lost 36 clicks at 90% and will sit at the top of your report.
Sort by contribution, not by severity
The instinct is to sort by percentage change, because a 90% drop looks alarming. It's the wrong axis, and it sends teams to investigate pages that never mattered.
Build the table this way instead:
| Column | Why |
|---|---|
| Clicks, current period | Where you are |
| Clicks, comparison period | Where you were |
| Clicks lost (absolute) | Sort on this |
| Cumulative clicks lost | Run the running total |
| Cumulative % of total decline | Stop reading at 80% |
Add a running total and a cumulative percentage, then draw a line at 80%. Everything above the line is the investigation. Everything below is noise you can safely ignore this quarter.
Most sites find the 80% line falls somewhere between five and twenty pages. That's a tractable list, and it's the entire value of doing the analysis this way.
Then decompose each page into queries
For each page above the line, pull its query-level detail for both periods and compute clicks lost per query.
Three patterns emerge, and each points somewhere different.
One or two queries collapsed. A specific ranking loss. Investigate the SERP for those queries and check who now holds the position.
Every query fell a little. Usually a CTR problem rather than a ranking one. Check whether position held while CTR dropped, which points at SERP composition. See Why is my organic traffic dropping while my rankings stay the same?.
Queries disappeared entirely. The page stopped ranking for them at all, which suggests indexation, cannibalisation, or a relevance shift rather than a competitive loss.
Separate the four causes before you act
Once you have the page-and-query list, sort each entry into one of four buckets. They need different work and mixing them produces a remediation plan nobody can execute.
Demand decline. Impressions fell with position and CTR stable. Nothing broke. Check seasonality before treating it as a problem, using How do I tell seasonality from a real SEO decline?.
Position loss. Average position worsened materially. Competitive or technical.
CTR loss at held position. The results page changed around you.
Indexation loss. Impressions went to near zero. Technical, and usually the fastest to fix.
The one that surprises teams most often is the first. A meaningful share of "declines" are demand contractions in the category, and no amount of content work reverses those.
Check whether the loss went to an AI answer
Before assigning a competitive cause, check whether the click was absorbed rather than taken.
Pew Research found that when an AI summary appears, users click a source link in only about 1% of visits. Our own state of AI search research, covering more than 22,000 answers, found citation presence reaches 97.4% on Google AI Mode.
That distinction changes the remediation entirely. A page still cited in the answer that replaced its click needs a measurement change, not a rewrite. A page whose query now returns an answer citing a competitor needs content work, and it's the more urgent case.
Build the decomposition as a repeatable report
Doing this by hand means two Search Console exports, a pivot table, and an afternoon. The value is in running it monthly, which nobody does manually.
GSC Top Pages for Site and Page-Keyword Breakdown supply both levels of the decomposition from one run, and the Code node holds the contribution arithmetic so the 80% cut-off is applied identically each time rather than eyeballed.
The Google Search node then classifies the SERP for the top declining queries, and Citation Watched Pages answers whether you're still the cited source.
A monthly agent:
Start (schedule, monthly) → GSC Top Pages for Site (current and comparison period) → Code (compute clicks lost per page, sort by absolute loss, cumulate to the 80% line) → Loop / For Each over the pages above the line → GSC Page-Keyword Breakdown per page → Google Search node on the top declining query per page → Citation Watched Pages → Code (classify each entry as demand, position, CTR, or indexation loss) → Prompt LLM (produce a remediation list grouped by cause) → Excel export → Send Email to the SEO lead.
The output worth having is the grouped remediation list. Ten pages with position loss go to one workstream, four with CTR loss go to another, and the demand-decline group gets closed with an explanation rather than a project. That grouping is what stops a decline report becoming a backlog nobody works through.
FAQ
Related answers
- How do I find the most likely cause of a traffic anomaly?
- Why is my organic traffic dropping while my rankings stay the same?
- Did a competitor take my traffic by publishing a better answer?
- How do I tell seasonality from a real SEO decline?
Want the 80% list produced automatically each month? Start a free Analyze AI trial and connect Search Console.