SEO Keyword Clustering and Content Brief for ChatGPT
Paste a raw keyword list; get it grouped into topic clusters by intent, one target page per cluster, and a full content brief for the cluster you pick, ready to hand to a writer.
Fill in the variables
Prompt
<role>
You are an SEO strategist who plans content for sites that compete on organic search. You group keywords by what the searcher wants, not by shared words, and you write briefs that a writer can execute without asking questions.
</role>
<context>
Site and what it does: {{site}}
Raw keyword list (one per line, optionally with monthly volume after a comma): {{keywords}}
Pages we already have on this topic (URLs or titles): {{existing_pages}}
Cluster to brief in detail (leave as "pick the best" if unsure): {{cluster_to_brief}}
</context>
<constraints>
- Cluster by search intent and the page that would satisfy it. Two keywords belong together only if one page can rank for both.
- Label each cluster's intent: informational, commercial investigation, transactional or navigational.
- Flag keywords that an existing page already targets, so we do not cannibalize.
- Do not invent search volumes or difficulty scores. Use the volumes I gave; if none, rank clusters by apparent intent value and say that volumes are missing.
- The brief must specify what the page needs to beat the current results, not generic "write quality content" advice. If you cannot see the SERP, say which SERP features to check and what to do for each.
</constraints>
<goals>
- A cluster map I can turn into a content calendar.
- Zero overlap between planned pages.
- One brief detailed enough that a freelance writer could draft from it.
</goals>
<instructions>
1. Cluster map: a table with cluster name, intent, keywords (primary first), suggested page type (guide, comparison, listicle, tool page, FAQ), and "existing page?" (yes with the page, or no).
2. Priority order: rank the clusters with one line of reasoning each.
3. Content brief for the chosen cluster: working title, URL slug, primary and secondary keywords, searcher's question in one sentence, required H2 sections with what each must contain, entities and terms to include, questions to answer (for FAQ and featured snippets), internal links to add (from and to), word count range with reasoning, and what would make this page better than the top results.
</instructions>
<output_format>
Three sections with headings. Use a simple table for the cluster map (pipe-separated is fine). No preamble.
</output_format>
Example input
Site: An AI tool directory with reviews, pricing and comparisons Keywords: ai background remover, 9900 best ai background remover, 1600 remove background from image ai free, 2400 ai background remover for product photos, 480 background remover vs photoshop, 210 how does ai background removal work, 170 bulk background remover, 390 background remover api, 320 remove.bg alternative, 880 canva background remover review, 260 Existing pages: /category/image-editing/ai-background-remover (category listing with 58 tools) Cluster to brief: pick the best
Example output generated with Claude Sonnet 4.6 from the example input above
CLUSTER MAP
Cluster | Intent | Keywords (primary first) | Page type | Existing page?
Best background removers | commercial investigation | best ai background remover (1,600); ai background remover (9,900); remove background from image ai free (2,400) | ranked comparison with pricing and test results | partly: the category page lists tools but has no comparison or scoring
Product-photo use case | commercial investigation | ai background remover for product photos (480); bulk background remover (390) | use-case guide + shortlist | no
remove.bg alternatives | commercial investigation | remove.bg alternative (880) | alternatives page | no (check whether a tool alternatives page exists for remove.bg)
Canva review | navigational / commercial | canva background remover review (260) | single-tool review | no (tool page exists; review section may cover it)
How it works | informational | how does ai background removal work (170); background remover vs photoshop (210) | explainer article | no
Developer | transactional | background remover api (320) | developer-focused listicle | no
Note: "ai background remover" (9,900) is the head term; its SERP is usually a mix of tools and listicles, so the comparison page targets it together with "best…". Volumes are the ones you supplied; difficulty is not estimated.
PRIORITY
1. Best background removers — highest combined volume, commercial intent, and the existing category page is close enough to upgrade rather than build from scratch.
2. remove.bg alternatives — single strong keyword with buyer intent and low content requirements if you already have alternatives data.
3. Product-photo use case — smaller but highly specific; converts well for a directory because the reader is choosing a tool.
4. Developer (API) — small volume, but a distinct audience your tool database can serve with a filter.
5. Canva review — only worth a page if the tool page cannot be expanded.
6. How it works — informational, lowest conversion; good for topical authority later.
CONTENT BRIEF — Best background removers
Working title: The Best AI Background Removers in 2026, Tested on 300 Product Photos
URL slug: /best-ai-background-remover (or upgrade the category page and set its title accordingly; do not create both)
Primary keyword: best ai background remover. Secondary: ai background remover; remove background from image ai free; free ai background remover.
Searcher's question: "Which background remover should I use, and is the free one good enough?"
Required H2 sections:
- How we tested — the photo set (count, categories: fur, reflective, transparent, batch), scoring criteria, date. Without this the page is just another list.
- The winner and runners-up — a results table: tool, free tier, price, edge score, batch support, API. Top 3 get a short paragraph each with one strong and one weak result.
- Best free option — answers the "free" secondary keyword directly; state the free tier's real limits (resolution, watermark, monthly cap).
- Best for product photos / for bulk / for developers — one short subsection each, linking to the deeper pages in clusters 2 and 4 when they exist.
- Where every tool failed — the honest section competitors lack; shows reflective and transparent cases.
- FAQ.
Entities and terms to include: edge refinement, hair/fur detection, transparent PNG, batch processing, API, resolution limits, commercial license, remove.bg, Canva, Photoshop's Remove Background, watermark.
Questions to answer: Is there a free AI background remover without a watermark? Which tool is best for hair? Can I remove backgrounds in bulk? Is AI background removal better than Photoshop? What resolution do free tiers export?
Internal links: from the category page and each of the top 10 tool pages to this article; from this article to the product-photography use-case page, the remove.bg alternatives page and the developer list; to each tool's pricing section.
Word count: 1,800–2,400. The ranking SERP for this term rewards thorough comparisons with tables; shorter pages lose to listicles, longer ones bury the table.
What beats the top results: real test images with before/after crops, a dated methodology, a cost-per-usable-image comparison instead of list prices, and a visible "last tested" date with a changelog. Most ranking pages are untested roundups.
Why this works & how to customize
Why it works: clustering by intent instead of by shared words is the step most keyword tools get wrong, and the "can one page rank for both" test makes the model do it properly. Asking for the existing-page check prevents the most common SEO mistake on growing sites: two pages fighting for the same term. The brief template is the one agencies use internally, so a writer can start without a call.
How to customize: paste up to about 80 keywords per run; beyond that, split by topic first. If you have difficulty scores from a tool, add them after the volume and the priority order will use them. For an e-commerce site, add "page type can be category, product or guide" so the model considers commerce pages. Run step 3 again with a different cluster name to brief the next page without redoing the map.
How to customize: paste up to about 80 keywords per run; beyond that, split by topic first. If you have difficulty scores from a tool, add them after the volume and the priority order will use them. For an e-commerce site, add "page type can be category, product or guide" so the model considers commerce pages. Run step 3 again with a different cluster name to brief the next page without redoing the map.
Comments (0)
Sign in to comment.
No comments yet.