Start with a decision your buyer needs to make. Give each article one useful job, its own evidence and a clear next step instead of chasing an AI-detector score.
Collect questions before choosing keywords
A useful content brief can begin with a sentence from a sales call: how do we know these emails are usable, what will the whole list cost, or who should we target first? These are decisions with consequences. Begin by collecting questions your intended buyers actually ask, then separate observed questions from ideas your team merely assumes are popular.
If you do not yet have customer conversations, interview potential users and label the notes as exploratory research. Public competitor pages can help identify vocabulary and gaps, but they do not prove your audience's priorities or a keyword's search volume. Do not turn a search result you happened to see into a claim that a topic is the highest-converting opportunity.
Assign one decision to each page
Create a small question ledger with five fields: question, audience, decision, evidence needed and next step. For example, a founder asking what an ICP should contain needs to choose account rules. A sales-operations lead asking about waterfall cost needs a calculation they can audit. Those readers may use similar words while needing different answers.
Our suggested brief is a paragraph, not a keyword dump: this page helps a specific reader decide a specific thing by explaining a distinction, showing an example and naming the limitations. If you cannot finish that sentence, do more research before drafting. It is cheaper to reject an unclear brief than to publish a thousand words that lead nowhere.
- Decision: what should the reader be able to choose after reading?
- Evidence: which source, calculation or observed workflow supports the answer?
- Boundary: what does the page not establish?
- Next step: what action naturally follows, without forcing a sale?
Use an evidence ledger while you draft
Tag each factual claim as product behaviour, external rule, observation or example. Product behaviour should match the current release. External rules need an authoritative source and a check date. Observations need a stated method. Examples need an explicit hypothetical label when they are invented. This lightweight ledger makes the final edit much faster because you know which sentences require proof.
Google's Creating helpful, reliable, people-first content guidance prioritises useful, original material and clear authorship, and warns against fabricated creator identities. It also says Google has no preferred word count. Use those principles as an editorial boundary, not a formula for making every article identical. A concise worked answer is better than a long page padded to meet a supposed SEO target.
Replace generic advice with a usable artefact
Compare improve your targeting with a worksheet that separates mandatory account filters from preferences. Compare track your costs with a calculation that includes rejected records and review time. The second version gives the reader something to use. Aim for at least one such artefact: a decision rule, a worked example, a checklist or a small evaluation table.
Make the artefact specific enough to reveal trade-offs. A contact-quality checklist should explain what happens to an unresolved result, not simply tell readers to buy good data. A content plan should include pages you decide not to create because they duplicate an existing answer. Showing the boundary of a method often teaches more than another list of benefits.
Edit for a reader, not an AI detector
Read the draft aloud. Remove the opening paragraph if it only says the market is changing. Replace broad verbs such as optimise with the action you mean. Check that every statistic has a traceable source and that each example is clearly hypothetical or supported by records. Vary sentence length because the argument needs it, not to imitate a detector's supposed preferences.
Do not invent a personal anecdote, customer quote or specialist reviewer to make a page feel human. If AI assisted the work, describe that process honestly where it matters. Ask an accountable editor to challenge the claim that matters most. Originality comes from the choices, evidence and reasoning in the piece, not a promise that a classifier will label it human-written.
Publish dates and promises honestly
Google's publication-date guidance distinguishes when a page was published from when it was substantially updated. Keep visible dates and metadata truthful. If you organise a newly published series into weekly editorial slots, label those slots as editorial dates rather than implying the pages existed earlier. Changing a calendar label does not add experience or evidence.
Before release, check the product promises one more time. An educational article about direct dials must not imply an unavailable feature is live. Link to a relevant explanation or contact route, then record what you want to learn from readers. The next article should respond to a real unanswered question, not merely occupy another place on the calendar.
Sources and further reading
Prepared with AI assistance and checked against the linked sources. Examples are illustrative, not customer results. See our editorial standards or request a correction.