Hannah Meier·
Clustered 900 Search Console queries into 14 content jobs in one pass — intent groups, the page each cluster deserves, and which existing URLs cannibalize each other.
Groups exported GSC queries into intent clusters, maps each to a target page, and flags cannibalization.
Search Console Query Clustering
You are an SEO strategist. I will paste an export from Google Search Console: queries with clicks, impressions, CTR, and average position. Optionally also the page each query currently ranks with.
GSC export:
{{gsc_export}}
My site and what it sells: {{site_context}}
Do the following:
1. **Cluster by intent, not by string.** Group queries that a single page could satisfy — "price", "cost", "how much" belong together; "X vs Y" is a different job than "what is X". Name each cluster by its job-to-be-done. Target 8-20 clusters; put genuine one-offs into a labeled remainder bucket instead of forcing them.
2. **Cluster scorecard.** Per cluster: total impressions, total clicks, impression-weighted average position, and an opportunity rating. Opportunity logic: high impressions + position 5-15 = quick win; high impressions + position >15 = content gap; position 1-4 with low CTR = snippet/title problem — say which case applies.
3. **Page mapping.** For each cluster, state which existing URL (from the export) should own it. If two or more URLs rank interchangeably for the same cluster, flag it as cannibalization and recommend: consolidate (which survives, which redirects), differentiate (how), or intentional — with one line of reasoning.
4. **Content actions.** Convert the top clusters into a ranked action list: new page (with working title + the queries it targets), rewrite (which URL, what changes), or on-page fix (title/meta/heading change, quoted). Rank by expected clicks gained: impressions × plausible CTR uplift — show the arithmetic assumption, keep it honest.
5. **Do-not-touch list.** Clusters that look tempting but should be left alone (branded queries, position 1 with healthy CTR, irrelevant intent) — and why.
No generic SEO advice. Everything must reference actual queries and numbers from my export.
Ergebnisse
## Query Clustering — 903 queries → 14 clusters (excerpt)
### Cluster Scorecard (top 6)
| Cluster (job-to-be-done) | Impr. | Clicks | W.Pos | Diagnosis |
|---|---|---|---|---|
| "What does bookkeeping software cost in CH" | 48,200 | 610 | 8.4 | **Quick win** — pos 5-15, high demand |
| "Bexio vs [us]" comparisons | 21,400 | 240 | 12.1 | **Quick win** |
| "VAT return how-to" | 33,100 | 190 | 17.8 | **Content gap** — pos >15 |
| "Is [brand] secure/serious" | 6,900 | 85 | 6.2 | Trust content, quick win |
| "[brand] login" | 41,000 | 12,400 | 1.1 | Healthy — do not touch |
| "Accounting basics for founders" | 27,800 | 95 | 24.6 | Content gap, low commercial intent — deprioritize |
Remainder bucket: 41 one-off queries (4.6% of impressions) — ignored deliberately.
### Cannibalization Flags
- **Pricing cluster:** `/preise` (pos 7.2) and `/blog/was-kostet-buchhaltung` (pos 9.8) alternate for the same 22 queries. **Consolidate:** blog post 301s to `/preise`; move its cost-example table into the pricing page first — the table is what earns the rankings.
- **VAT cluster:** `/blog/mwst-guide` and `/features/mwst` both hover at 15-20. **Differentiate:** blog owns "how to file" informational queries, feature page owns "software for VAT" commercial ones; cross-link with exact-match anchors.
### Ranked Content Actions
1. **Rewrite `/preise`** — absorb cost-example table, add "cost" + "how much" phrasing to H2s. Assumption: pos 8→4 doubles CTR from 1.3% to ~4% on 48k impressions ≈ **+1,300 clicks/mo**.
2. **New page: "Bexio Alternative: honest comparison"** — targets 38 vs-queries currently scattered across 3 blog posts. Est. +400 clicks/mo at pos 5-6.
3. **New guide: VAT return step-by-step** — content gap, 33k impressions with no page above pos 15. Est. +350 clicks/mo, slower ramp.
4. **Title fix on trust cluster page:** current title "About us" → "Is [brand] secure? Audits, hosting & data protection" — CTR problem at pos 6.
### Do-Not-Touch
- `[brand] login` — navigational, pos 1.1, 30% CTR. Any change only risks it.
- "Accounting basics" cluster — 27.8k impressions but founder-student intent, historical conversion near zero. Not this quarter.
Modell: Gemini 2.5 Flash
36 Likes24 SavesScore: 25
1 Kommentar
David Park·
Showing the CTR-uplift arithmetic instead of just saying 'high opportunity' — this is how you get an SEO roadmap funded.