How to structure a content team in 2026: roles, standards and a weekly rhythm

How to structure a content team in 2026: roles, standards and a weekly rhythm

On 25 September 2026 we counted the profile text on every live listing on TrustList. About 24,350 of them, roughly a quarter, carry fewer than 50 words describing the business or product, and 12,178 companies have no description at all. On the same day Google declined our application to its AdSense advertising programme for "low-value content". We did not argue with the decision. The count had already told us the same thing.

The traffic told us too. In the 92 days to 18 September 2026, Google Search Console recorded 652,463 impressions of our pages and 657 clicks. Pages about one business or product, where the profile is specific and documented, earned 528 of those clicks from about 33,000 impressions, a click rate of 1.6%. Our ranking pages, which list many firms with little said about each, earned 67 clicks from about 255,000 impressions. People clicked on the pages that answered a specific question with specific facts, and mostly ignored the ones that did not. This guide is about how a chief executive should set up a content team so that it produces the first kind of page and not the second.

What our thin pages taught us

The thin listings did not come from carelessness in any single week. They came from a structure. A large part of the catalogue arrived in one bulk import: 9,620 live companies and 3,416 live products were last updated in December 2020, when that import ran, and nobody owned them after that. Nobody checked whether the businesses still existed; when we drew 1,000 company listings at random on 24 September 2026, 171 of their websites no longer led to the business as listed. And 5,092 software listings still show 2020 prices with no date; in a sample of 30 checked against the vendors' own pages, only 6 still matched.

The lesson for any content team is that volume without ownership decays. A page is not finished when it is published. Someone has to own it, re-check it and retire it when it stops being true.

Our response has four parts, and each maps to a role in the team shapes below:

• Rebuild thin listings in our own words from the business's own website, with sources kept.
• Retire businesses that have closed, with a page that says so rather than a silent deletion.
• Remove what nothing can be found for.
• Curate: a second sitemap, at /sitemap-curated.xml, that holds only the 233 pages we stand behind today. The first batch rewrote 48 listings and retired 2 as closed.

We also wrote down the editorial rules that every piece, including this one, must follow. Open with something we counted or observed ourselves. Trace every figure to a source someone can open. When a listing is published before we have been able to verify it independently, show a red "not yet independently verified" note on the page, and re-check it on every refresh run. Never invent a rating, price or benchmark.

What Google says it rewards

Google's own guidance is more specific than most commentary about it. Its page on creating helpful, reliable, people-first content, last updated in December 2025, asks publishers to judge their work with questions such as whether it provides original information, reporting, research or analysis, and whether it adds substantial value compared with other pages. It says that of experience, expertise, authoritativeness and trust, trust is the most important. It lists warning signs of content made for search engines first, including producing a lot of content on many topics hoping some will perform, and using extensive automation to produce content on many topics.

It also asks publishers to think about "Who, How and Why": who created the content, how it was created, and why. On the "how", it asks whether the use of automation, including AI generation, is "self-evident to visitors through disclosures or in other ways".

Google's position on AI itself was set out in February 2023 and has not been withdrawn: "Appropriate use of AI or automation is not against our guidelines." What is against its rules is using automation "to generate content with the primary purpose of manipulating ranking in search results". Its spam policies, updated in August 2026, define scaled content abuse as "when many pages are generated for the primary purpose of manipulating search rankings and not helping users", and give as an example using generative AI tools to generate many pages without adding value for users.

Read together, the guidance does not tell you to avoid AI. It tells you that a page must be worth reading whoever drafted it, that the reader should be able to see who stands behind it, and that making many pages cheaply is itself a warning sign. That is a description of an editorial process, not a writing tool. For more on how search is changing, see is SEO dying? and our own account in our search traffic journey.

The rules that now apply to AI text and reviews

Three sets of rules affect how a content team works in 2026.

The EU AI Act, Article 50. The transparency obligations in Article 50 have applied since 2 August 2026. Under paragraph 4, a deployer who publishes AI-generated or manipulated text to inform the public on matters of public interest must disclose that it was artificially generated. The obligation does not apply where the text "has undergone a process of human review or editorial control" and a person or company "holds editorial responsibility for the publication". The European Commission's guidance, published in July 2026, says a spelling or grammar check does not count; editorial control means someone with real authority who can approve, change or reject the substance, including checking facts and sources. If you publish to readers in the EU, a named editor who actually reviews each piece is now a legal safeguard as well as good practice. The Commission's page also notes a limited grace period, to 2 December 2026, only for the separate machine-readable marking duty on AI systems already on the market.

Fake reviews and testimonials in the United States. The Federal Trade Commission's rule on consumer reviews and testimonials, announced in August 2024, took effect on 21 October 2024. It bans fake reviews and testimonials, including ones generated by AI that misrepresent a real experience; paying for reviews conditioned on a particular sentiment; undisclosed reviews by company insiders; company-controlled websites posing as independent reviewers; suppressing negative reviews through threats; and buying or selling fake social media indicators. The FTC can seek civil penalties against knowing violators.

Fake reviews in the United Kingdom. The Competition and Markets Authority published its guidance on fake reviews on 4 April 2025, reflecting rules under the Digital Markets, Competition and Consumers Act that took effect that month. Businesses that publish consumer reviews have duties to prevent and remove fake ones, and concealed incentivised reviews are banned.

For a content team the practical point is the same in every jurisdiction: testimonials, case studies and quoted reviews need a real, traceable source, and the team must keep the evidence. We hold about 20 reviews across 87,205 company listings, and we say so, rather than implying our rankings come from reviews.

The roles a content team needs

A content team needs five kinds of work covered. In a small team one person covers several; in a larger one each becomes a job.

1. Editor. Decides what gets published, holds editorial responsibility in the Article 50 sense, and signs off every piece. Owns the standards document and the corrections log.
2. Subject experts. The people who know the topic from doing it: your engineers, accountants, customer support leads, or outside specialists. They supply the experience Google's guidance asks for, and they check drafts in their field.
3. Researcher or data person. Counts things, pulls your own data, opens primary sources and records them. This is the role that produces the "our finding" opening that makes a piece worth reading.
4. Search and distribution. Knows what questions people are asking, maintains the site's structure and internal links, prepares versions for email, social and partner channels, and watches the numbers.
5. Design and video. Charts, diagrams, cover art and short explainers. Good charts of your own data are often the most-shared part of a piece.

Every role is human-accountable. AI tools can help each of them; none of them is replaced by a tool.

AI-assisted drafting with a person accountable

We use AI tools in our own workflow, and we think most teams should. The rule that makes it safe is simple: a named person is accountable for every published sentence, and the process is designed so that person can actually check it.

• Research first, draft second. The researcher gathers the facts and sources before anyone drafts. A tool drafting without sources will fill gaps with plausible text; a tool drafting from a supplied source pack has much less room to do so.
• Every figure is traced. Each number in a draft must link to the page it came from, which someone has opened. If it cannot be traced, it is removed, however likely it looks.
• The expert reads for truth; the editor reads for standards. Two different checks, done by two different people where the team is big enough.
• Keep a record. Which tool, which sources, who reviewed, who approved. That record is your evidence of editorial control.
• Disclose where readers would ask. Google's advice is that disclosures are useful where someone might reasonably wonder how the content was created. A short note on method at the end of a data piece is usually enough.

Sourcing and fact-checking standards

Write these down, keep them to a page, and apply them to every piece:

1. Primary sources over secondary: the official statistic, the regulator's page, the company's own filing, the named survey publisher's own page. Never an aggregator's table or an uncited blog figure.
2. Every source listed with a title, link, publisher and date, or "read on" date if the page has none.
3. Our own data described with its method and its limits: the window, the sample, what it does not show.
4. No invented ratings, prices, benchmarks, quotes or testimonials.
5. Items not yet independently verified carry a visible note saying so, and are re-checked on a schedule.
6. Corrections published openly, with the date.
7. Pages have an owner and a review date; stale pages are updated, marked as dated or retired.

Team shapes for one, three and eight people

One person

The founder or a single content lead does everything, so the discipline has to come from the calendar. Publish less, and make each piece count: one piece of original work a fortnight is better than five thin ones a week. Borrow subject experts from inside the company for an hour per piece. Use AI tools for research summaries, outlines and first drafts, and never skip your own source check. The chief executive acts as the second reader for anything that makes a claim about customers, competitors or money.

Three people

Role	Covers
Editor and lead writer	Standards, sign-off, editorial responsibility, most long pieces
Researcher and writer	Data pulls, source packs, fact checks, shorter pieces, page refreshes
Search, distribution and design	Site structure, internal links, charts, email and social versions, measurement

Subject experts from the wider company review drafts in their field. One day a week goes to refreshing and retiring existing pages, not new ones.

Eight people

Role	Number	Covers
Head of content (editor-in-chief)	1	Standards, legal questions, final sign-off, corrections
Section editors	2	Commissioning and editing in two subject areas
Researcher or data analyst	1	Own-data studies, source checking, method notes
Writers	2	Drafting from source packs, with subject experts
Search and distribution	1	Structure, internal links, channels, measurement
Designer or video producer	1	Charts, diagrams, cover art, short explainers

At this size, split the checks: the section editor edits, the researcher fact-checks, and the head of content signs off. Keep a page register with owners and review dates.

A weekly workflow

• Monday: plan. Review last week's numbers and reader questions. Agree this week's pieces, each with an owner, a subject expert and a question it answers.
• Tuesday: research. Build the source pack. Pull or count your own data. Decide what the "our finding" opening will be. If there is no finding, reconsider the piece.
• Wednesday: draft. Draft from the source pack, with AI assistance if you use it. Mark every figure with its source.
• Thursday: check. Expert review for accuracy, fact check for every figure, editor review for standards and disclosure. Fix or cut.
• Friday: publish and maintain. Publish, add internal links from related pages, prepare distribution versions. Spend the rest of the day on refreshing, marking or retiring older pages.

Measuring beyond clicks

Clicks matter, but on their own they reward the wrong thing. Our ranking pages drew hundreds of thousands of impressions and almost no clicks; a click-only report would have told us to make more of them. Track a small set instead:

1. Clicks and click rate by page type, so you can see which kinds of page earn attention, as our listing pages did.
2. Share of pages with an owner and a review date inside the last year.
3. Share of figures with a primary source, checked on a sample each month.
4. Corrections per month, and how fast they were made.
5. Pages retired or rebuilt, because removing weak pages is part of the job.
6. What readers did next: enquiries, sign-ups, replies, links from other sites, questions from sales calls that cite the piece.
7. Reuse inside the company: how often sales, support or leadership send a piece to a customer.

Checklist for the chief executive

1. Name one editor with authority to approve, change or reject every piece, and put it in writing.
2. Write a one-page standards document: sourcing, disclosure, reviews and testimonials, corrections.
3. Decide which AI tools the team may use and for which steps, and require a record of each piece's sources and reviewers.
4. Give every page an owner and a review date; budget time every week for maintenance.
5. Make "open with our own finding" a rule; fund the research and data role before adding more writers.
6. Measure what readers do next, not just clicks.

The same principle, a named person accountable for what a tool helps produce, applies to engineering. We set that out in how CEOs and CTOs should structure technology teams.

Sources

• Creating helpful, reliable, people-first content: self-assessment questions, trust as most important, "Who, How and Why", disclosure of automation, search-engine-first warning signs. Google Search Central, updated 10 December 2025.
• Google Search's guidance about AI-generated content: appropriate use of AI is not against guidelines; automation to manipulate rankings is. Google Search Central, 8 February 2023.
• Spam policies for Google web search: definition of scaled content abuse and the generative AI example. Google Search Central, updated 28 August 2026.
• Article 50: Transparency obligations for providers and deployers of certain AI systems: text of paragraph 4 and the editorial responsibility exception. EU AI Act text (read 25 September 2026).
• Transparency obligations under Article 50 of the AI Act: application from 2 August 2026, the marking grace period, and what counts as human review or editorial control. European Commission, 24 July 2026.
• FTC announces final rule banning fake reviews and testimonials: practices prohibited and civil penalties. Federal Trade Commission, 14 August 2024.
• The Consumer Reviews and Testimonials Rule: questions and answers: effective date of 21 October 2024 and insider disclosure. Federal Trade Commission, November 2024.
• Fake reviews (CMA208): UK guidance on prohibited review practices and duties of businesses that publish reviews. Competition and Markets Authority, 4 April 2025.
• TrustList catalogue counts, 24 and 25 September 2026: listings with fewer than 50 words, companies with no description, the 2020 import, the random sample of 1,000 websites, the price sample, reviews, and the curated sitemap. TrustList Google Search Console data, 92 days to 18 September 2026. Google AdSense decision, 25 September 2026.