Guides / Guide
Prompt templates for a content brief that targets what someone's actually searching for
Background
A content brief built by asking "write a brief for a page about [topic]" tends to produce a competent outline of the topic - headings, a rough word count, a list of subtopics to cover. What it doesn't do is answer the question that actually determines whether the page performs: what is someone typing that search actually trying to accomplish, and does this page's structure match that?
Search intent splits, roughly, into wanting information, wanting to compare options, or being ready to act - and a page built for the wrong one of those three loses regardless of how well it's written. A comparison-intent search met with a single-product sales page bounces. An action-ready search met with a long informational essay bounces just as fast, for the opposite reason. Getting the brief to name the intent before it names the headings is what stops the mismatch happening in the first place, rather than being found in analytics three months later.
Setup
The shape every template below follows:
- Role - an editor briefing a writer, not a writer drafting the page itself. The brief is a spec, not the content.
- Context - the search term or topic, and what should come back (a brief, not a draft).
- Inputs - the actual target term, who's likely searching it, and what you already rank for or compete against on the topic.
- Task - name the intent first, then build the structure around it.
- Constraints - no invented statistics, no keyword-stuffing instructions, a brief a writer could actually follow.
Standard content brief
For a single target page or blog post:
You're an editor briefing a writer on a page targeting a specific search
term.
Context: return a content brief, not a draft - headings, what each
section needs to cover, and the intent it's written for. I'll hand this to
a writer.
Inputs:
- Target search term: [e.g. "HOW TO CHOOSE A PAYROLL PROVIDER"]
- Who's likely searching it and why: [YOUR BEST GUESS - ROLE, SITUATION]
- What we already have on this topic, if anything: [DESCRIBE OR "NOTHING"]
- Anything we know ranks well for this term currently: [DESCRIBE OR
"UNKNOWN"]
Task: state the intent in one line first (informational, comparison, or
action-ready) - everything else in the brief has to match that intent.
Build a heading structure that answers the search directly in the first
section, not after three paragraphs of preamble. Note what a genuinely
useful page on this topic includes that a thin one wouldn't.
Constraints: no invented statistics or claims - mark anywhere real data
would strengthen a point as "[NEEDS A REAL SOURCE]" rather than inventing
one. No instruction to repeat the keyword a set number of times - that's
not how this ranks and reads as dated advice if it shows up in the brief.
Comparison / "best of" page brief
For pages where the searcher is actively choosing between options - these fail hardest when built with the wrong structure, because a reader who's comparing wants the comparison up front, not buried under an introduction:
You're briefing a comparison-style page, where the reader is actively
choosing between options.
Context: return a brief for a page that puts the comparison first, not an
introduction to the category first.
Inputs:
- What's being compared: [e.g. "OPTIONS X, Y, Z FOR A SPECIFIC USE CASE"]
- The actual differences that matter to the decision: [LIST WHAT YOU KNOW]
- Anything genuinely disqualifying about an option, if there is one:
[DESCRIBE OR "NONE KNOWN"]
Task: brief a structure that gets to a comparison table or clear
differentiator within the first section, with supporting detail after -
not a long preamble before the comparison the reader actually came for.
Flag if any of the differences I've given you are too vague to write a
genuine comparison from ("it's more flexible" isn't a comparison point
until it says flexible in what way).
Constraints: never invent a feature, price or limitation for any option -
mark anything unverified as "[VERIFY]" rather than guessing at it, because
a wrong claim about a real product is a real, checkable mistake, not a
stylistic one.
Existing-page refresh brief
For a page that already exists and ranks, but has drifted or underperforms
- the brief here is a diagnosis first, an update second:
You're briefing a refresh of an existing page, not a rewrite from
scratch.
Context: return a brief that names what's actually likely wrong with the
current page relative to intent, then what to change - not a generic
"make it better" list.
Inputs:
- The page's current content: [PASTE OR SUMMARISE]
- The term it's meant to rank for: [TERM]
- What's changed since it was written, if you know (new competitors, a
policy change, a product update): [DESCRIBE OR "UNKNOWN"]
Task: identify the specific gap - wrong intent match, outdated information,
missing sections a genuinely useful page on this topic would have - before
suggesting rewrites. Keep anything on the page that's still accurate and
well-targeted; don't recommend rewriting a section that isn't broken.
Constraints: distinguish clearly between "this is factually outdated" and
"this could be written better" - they need different fixes, and conflating
them produces a brief that rewrites accurate content unnecessarily.Edge Cases & BCP
- A model will confidently assign an intent it hasn't actually verified. It's reasoning from the search term's wording, not real search data - if you have access to actual search console data, ranking competitor pages, or "people also ask" results for the term, feed those in rather than letting it guess intent from the phrase alone.
- "Comprehensive" is not the same as "matches intent." A brief that tries to cover every possible angle on a topic often serves nobody - the action-ready searcher wants the shortest path to deciding, not the most thorough essay. More sections isn't automatically a better brief.
- Watch for it recommending keyword density instructions or exact-match heading text as if that still moves rankings on its own - that's outdated SEO folklore the model can pick up from older training data. A brief built around genuinely answering the intent needs no keyword quota bolted onto it.
- A brief is not a substitute for checking what's actually ranking right now. Search results for the same term shift, and a competitor page the model doesn't know about might already own the exact angle you're about to brief - check the live results before commissioning the page, not after it's written.