The Best SEO Copywriter Is the One Who Argues With Your Brief

The best SEO copywriting service is not the one that accepts your keyword, word count and deadline with the least friction. It is the one willing to challenge whether the requested page should exist, what it may honestly claim and which business decision it must help a reader make. That challenge is part of the product.

A portfolio can prove that somebody produces attractive sentences. It cannot show whether the writer stopped an unsupported promise, found an existing page that already owned the subject or noticed that the call to action led to a service the business could not deliver as described.

The writer should not become difficult for sport. They should create useful resistance at the points where a cheap yes becomes an expensive live page. If every brief becomes 1,500 words of agreeable copy, the service is selling output volume, not judgement.

Your brief is evidence, not an instruction from the gods

Clients know their businesses. They also arrive with internal language, inherited assumptions and solutions chosen before the problem has been defined. “We need a page targeting growth marketing agency Brighton” may mean several things. The company may need a proper service page. It may already have one under a different name. The real gap may be proof, pricing context or audience clarity. The phrase may describe no coherent offer at all.

I would begin by converting the request into a page decision: this page helps a defined reader in a recognisable situation decide a specific question using an answer the business can prove, then take an appropriate next step. Then I would ask what the page must not own.

That second question matters because copy expands easily. Company history, beginner education, every adjacent service, the complete growth strategy and an entire sales pitch can all be made “relevant.” The result feels comprehensive while weakening every decision. A good SEO copywriter does not merely fill the page. They establish its borders.

The first sign of quality is a better disagreement

Not all pushback is useful. A writer who argues about commas while accepting a fictional claim is not protecting the work. Nor is an agency that performs theatrical contrarianism and then delivers the same competitor-derived headings. Useful disagreement changes one of five things. I call these the Five Editorial Frictions:

They are not delays to remove from the process. They are the places where professional judgement earns its fee. If the writer never finds one meaningful friction, either the brief is unusually complete or the discovery process is too polite.

Ask for the decisions behind one portfolio sample

Most buyers inspect portfolios like galleries: scroll, admire, imagine their own brand receiving the same finish. I would treat a sample as an archaeological site. Ask the provider:

• What job did this URL need to perform?
• What was wrong or unresolved in the original request?
• Which evidence was available?
• Which proposed claims were narrowed, attributed or removed?
• How did search evidence alter the argument rather than just its vocabulary?
• Why did the page follow this sequence?
• What did the subject owner challenge?
• What changed when the copy entered the website design or CMS?
• What could be observed after publication, and what could not be attributed to the words?

Confidentiality may prevent the provider showing private documents or naming a client. That is reasonable. It does not prevent them explaining the kinds of decisions they made and the standards they applied. “The client loved it” is a reaction. “Traffic increased” is an observation needing scope and causal restraint. Neither answer reveals whether the copywriter controlled factual risk or improved the buyer’s decision. The sample matters. The reasoning behind it is what you are hiring.

Separate business authority from writing authority

A capable writer should challenge the brief without pretending to own the business’s facts. The cleanest commissioning model separates authority.

One provider may perform several roles. That does not grant them universal authority. An SEO copywriter should not invent a service territory because the keyword includes a town. A branding agency should not create a credential the business does not possess. A conversion rate optimisation specialist should not promise a result the evidence cannot support. Equally, the client should not dictate the narrative sentence by sentence and then blame the writer for weak persuasion. The productive relationship is not “client knows everything” or “writer knows best.” It is explicit decision ownership with permission to challenge across the boundary.

Demand a Copy Decision Trail

The finished document hides the dangerous moments because good editing removes visible uncertainty. I want a small record of what happened before the prose became smooth. A useful Copy Decision Trail contains:

1. the intended reader, situation and decision;
2. approved business facts, interviews and primary sources;
3. relevant search observations and their date;
4. claims retained, narrowed, attributed, held or rejected;
5. evidence matched to consequential propositions;
6. the reason for the opening and section order;
7. the intended relationship between title, main heading and opening;
8. unresolved review or implementation dependencies;
9. consequential revision choices and their owners;
10. conditions for approving the live page.

This is not a demand for bureaucratic cosplay. A small commission may fit on one sheet. The standard is reconstructability: can somebody understand why the page says this rather than something easier, louder or more convenient? If a provider can show only the final Word document, you can judge style. You cannot inspect judgement.

Watch what happens when the evidence runs out

Give the prospective writer a phrase your business likes to use: “trusted specialists,” “proven results,” “fast response,” “tailored service” or “industry-leading expertise.” Then ask what they would need before publishing it. A responsible process gives the claim one of five treatments:

• Retain it because appropriate evidence supports the likely reading.
• Attribute it to the source that actually makes it.
• Narrow it to the scope the evidence can carry.
• Hold it for verification.
• Remove it.

The wrong response is to make the unsupported sentence more persuasive. A review describes one customer’s reported experience; it does not prove a universal outcome. A qualification held by one person does not necessarily belong to the company. A case result shows what occurred under defined conditions, not what every future customer should expect. Google’s people-first content guidance asks whether material offers original information or analysis, uses clear sourcing and helps the reader achieve their goal. It also asks whether sourced material adds substantial value instead of merely rewriting what already exists (Google Search Central: creating helpful, reliable, people-first content). Those are useful editorial questions. They are not permission to invent the experience or evidence required to answer yes.

I trust the writer who makes missing proof visible. They are protecting the brand from a sentence that becomes harder to question after everyone falls in love with it.

Search research should sharpen the argument, not colonise it

SEO copywriting earns its name when search evidence improves the page’s usefulness. It becomes keyword theatre when the result set is treated as an outline generator. Ask how the writer decides which query or need belongs to the page; whether another URL already owns it; what the searcher is trying to decide; which related questions strengthen that same decision; which tempting material would make the page drift; where first-party evidence should overturn the generic market answer; and what deserves one deeper page rather than another shallow URL.

Google’s Search Essentials recommends using language people use to find the content in prominent page locations while centring helpful, reliable, people-first material (Google Search Essentials). That does not prescribe a density score, a heading for every variation or a remix of the current top ten. The best search insight is often subtractive. It tells the writer which apparent opportunity belongs elsewhere, which audience is wrong for the offer and which broad phrase would attract attention the business cannot convert.

For AI search optimisation, the same principle applies. AI SEO does not excuse vague authority language or synthetic scale. A page becomes easier to use and cite when its subject, answer, evidence and limits are clear, not when it accumulates fashionable terminology.

Test whether the promise survives the website

Copy is not finished when a document is approved. It is finished when the live page preserves the argument. Ask to see the proposed relationship between the search-title candidate, the visible page heading, the first meaningful passage, the principal claim and its nearest proof, and the primary next action. These should behave like one commercial sequence. A title promising pricing insight should not open with a company biography and hide the price factors below three calls to action. A narrow WordPress service should not use a broad web-design title to attract buyers whose problem it cannot solve.

Google explains that title links may be generated from the title element, visible title, headings and other page text, and recommends descriptive, concise and accurate title text (Google Search Central: title links). Google also explains that snippets are primarily generated from page content and may vary by query, although a meta description may be used when it better describes the page (Google Search Central: snippets). A provider can improve those inputs. They cannot guarantee the exact display.

The implementation check should also cover design hierarchy, proof placement, mobile behaviour, forms and any state that changes the offer. Great copy trapped in an accordion nobody opens is not a successful handover. Neither is a strong argument broken into interchangeable blocks because the template wanted symmetry.

Ask what would make the writer refuse the job

This is my favourite commissioning question because it exposes the provider’s commercial incentives. A serious answer should include some combination of: an offer that is still changing; no agreement about the intended audience; an existing page that already owns the job; material claims with no available evidence; contradictory business facts; a location or service promise operations cannot confirm; a design or development dependency that prevents a responsible release; a conversion route that fails after the click; or a request to manufacture volume without distinct value.

Sometimes the correct decision is not to hire an SEO copywriter yet. The business may need to settle the service, gather proof, redesign the page system or fix the acquisition route first. That willingness to pause costs the provider immediate revenue. It is also a costly signal that their growth strategy is not built entirely around selling more words. I would be wary of any production service that can convert every possible brief into an invoice without discovering a reason to narrow one.

Price the unresolved decisions, not just the word count

Length estimates effort. It does not describe the risk or thinking required. Compare proposals by the decisions included: discovery and stakeholder access; page permission and search ownership; source and evidence review; claim control; copywriting and editing; title and snippet recommendations; design and CMS handover; specialist review; revision and approval ownership; live-page verification; and post-release observation and its limits.

A short service page can require more judgement than a long informational article. A low-cost rewrite can be sensible when the offer, proof and architecture are already settled. An expensive retainer can still produce generic content. Ask what uncertainty the fee pays somebody competent to resolve. That question connects the copywriting purchase to business growth far better than price per word.

Judge AI use by accountability

AI assistance is not automatic evidence of poor writing. Invisible accountability is. Ask what source material enters the process, how confidential information is handled, who checks material claims, how generic output is rejected and which named person approves the article’s judgement. Ask what working record survives after the chat window closes. Google says generative AI can support research and structure, while generating many pages without adding value can violate its scaled-content abuse policy when the purpose is manipulation (Google Search Central: generative AI content; Google Search spam policies). The meaningful distinction is purpose, value and accountability, not whether every key was pressed by a human.

An AI-assisted writer who exposes sources, challenges claims and authors a defensible position can be more useful than a human typist who politely converts a keyword list into prose. The tool does not absolve either provider of judgement.

The buying test takes one sample and one difficult claim

You do not need a 40-question procurement theatre. Use this compact test:

The strongest answer may create a smaller page, a longer discovery phase or no immediate writing job. That is not poor service. It is evidence that the provider can distinguish customer acquisition content from content-shaped inventory. There is a boundary. Small, low-risk assignments do not need a governance cathedral. Experienced commissioners may have already settled the decisions and need clean execution. The process should scale with consequence.

But the principle survives: the best SEO copywriter will not make every request feel correct. They will expose the moment where your preferred wording, page or promise stops serving the reader and the business. Do not hire the portfolio alone. Hire the quality of the disagreement behind it.