How to build topic authority without publishing generic content.

A site that cannot be crawled, contradicts its own canonical URLs or takes eight seconds to render has a ceiling, and no amount of writing raises it. That is why technical work comes first and why it is genuinely urgent.

It is also why so many businesses are surprised when the technical work finishes and nothing happens. Removing an obstacle reveals whatever was behind it. If what was behind it is thirty pages that could have been written by anyone, the outcome of a perfect technical audit is thirty perfectly accessible pages with nothing to distinguish them.

Authority is the part that has to be built rather than cleared. It is slower, it is harder to outsource, and it is the only part a competitor cannot copy in an afternoon.

The test that settles it: pick any page and ask whether a competent competitor could have published the same page without knowing anything the business knows. If they could, the page is coverage, not authority.

Authority gets talked about as a score, which makes it sound like something a tool measures rather than something a business builds. It is more useful as four separable properties, each of which can be worked on directly.

• DEPTH ON A DEFINED SET

  Several pages covering one subject from different angles, so the site has said more about it than a single article could. Depth is what breadth is usually mistaken for.

• FIRST-HAND MATERIAL

  Something the business knows because of work it has done: a method, a failure pattern, a decision rule, a constraint it has hit. Google’s content guidance is explicit about originality and first-hand expertise.

• CONSISTENCY

  The same facts, stated the same way, wherever they appear. Contradictions between a business’s own pages become confident errors in anything that summarises it.

• DURABILITY

  Content that is still correct next year, and an owner for the parts that are not. Authority accumulates only if the earlier work stays true.

The objection usually means the business has no case studies, no proprietary data and no research budget, and concludes that it therefore has nothing original to publish. That conflates evidence with expertise.

What a working business has instead is the accumulated result of doing the work: the questions clients actually ask, the decisions that keep going wrong, the sequence that has to happen in a particular order, the thing everyone assumes that turns out not to be true. None of that requires a dataset and none of it is available to somebody who has not done the work.

That material is also harder to copy than a statistic. A number can be cited by anyone; a method that reflects how a specific business operates cannot be borrowed without becoming obviously secondhand.

A business describes itself dozens of times across a site: in service pages, in an about page, in article bylines, in metadata. Where those descriptions drift, a system assembling an answer about the business has to pick one, and it will pick without telling anyone.

This is cheap to fix and almost never done. Decide the canonical statement of what the business is, who it serves and what it does not do, then make every page agree with it — not by repeating a paragraph verbatim, but by never contradicting it.

The same applies to terms. A business that calls the same thing three names across three pages has made its own subject harder to associate with it, for a reader and for anything reading on the reader’s behalf.

Authority is built by finishing rather than by starting. A single subject covered from every angle a reader needs is a defensible position; twelve subjects covered once each is a backlog that reads as thin from the outside.

1. 01

   Choose the subject the business wants to be known for

   Decides

   Which work do we want more of?

   Output

   One subject, chosen commercially rather than by search volume.

2. 02

   Inventory what is genuinely first-hand

   Decides

   What do we know because of work we have done?

   Output

   A list of methods, failures and decision rules — the raw material.

3. 03

   Map it against the reader’s stages

   Decides

   What does somebody need at understand, decide and implement?

   Output

   A small set of pages that complete the subject rather than sample it.

4. 04

   Settle the canonical facts

   Decides

   How do we describe ourselves and our terms, everywhere?

   Output

   A statement every page can be checked against.

5. 05

   Publish and connect

   Decides

   Which page is the hub, and what links into it?

   Output

   A connected group rather than isolated articles.

6. 06

   Set a review owner before starting the next subject

   Decides

   Who keeps this true?

   Output

   A named owner and a review interval — or a decision not to publish it.

This is where most content programmes actually stop. The first-hand knowledge is real and it belongs to whoever does the work — usually the founder or the most senior practitioner, the two people with the least uninterrupted time and the least appetite for drafting. Asking them to write produces three articles and then silence.

Treating it as an extraction problem rather than a writing problem changes what is being asked of them. Nobody who knows a subject struggles to explain it out loud to somebody asking reasonable questions; the difficulty is the blank page, the structure and the second draft, and none of those require the expert.

The workable division is that the expert supplies the substance and reviews the result, and somebody else does everything in between. That is a twenty-minute conversation and a fifteen-minute review per article, which is a commitment a busy person can actually keep — unlike a writing habit.

The questions that produce usable material are narrow. What do clients get wrong before they arrive? What did you have to stop doing? What does everyone assume that is not true? What would you check first? Broad prompts produce generalities; those four produce the specifics that cannot be found anywhere else.

The signal that extraction is working: the draft contains at least one thing the person writing it could not have known. If it does not, the conversation collected opinions rather than experience.

Each of these is a way of appearing to build authority while building only page count. Google’s guidance describes the underlying pattern directly: content produced primarily for search systems rather than for people.

• SUMMARISING THE FIELD

  An article assembled from the top results for its own topic. It is accurate, unnecessary, and structurally incapable of being the best answer.

• SCALED PRODUCTION

  Publishing a lot quickly, whether by outsourcing or by generation. Volume without first-hand material is the exact thing search guidance treats as abuse.

• THE SAME ARTICLE IN FIVE VARIANTS

  One idea rewritten for five phrasings. It divides the evidence about which page is authoritative and doubles the maintenance.

• BORROWED CREDIBILITY

  Statistics, awards, certifications or client outcomes that are not the business’s own. A number that cannot be sourced is worse than no number, and one that is invented is a different problem entirely.

The honest statement about timescale is that nobody can give one. Technical corrections can register quickly; being recognised as a source on a subject is a function of how much has been published, how consistent it is, how long it has been true and what else exists — and the last of those is outside anyone’s control.

What can be said is the shape. Early work produces very little visible return, because a site with two pages on a subject is not yet distinguishable from a site with one. Return arrives when the group is complete enough that a reader can stay, which is also when internal links start describing a genuine structure.

That shape has a practical consequence: judging the programme after the first two pages will always say stop. The first honest review point is after one subject is covered completely.

The work above produces something real and largely invisible in the first months, which makes it vulnerable to being cancelled by whoever is asked to justify it. That is a measurement problem rather than a content problem, and it has its own answer.

It also assumes the pages themselves are structured so that a machine can use them — substance and form are separate disciplines, and a page can have one without the other.

1. Technical work removes a ceiling; it does not create value. A perfect audit over generic content produces perfectly accessible generic content.
2. Authority is four separable properties: depth on a defined set, first-hand material, consistency, and durability.
3. “We have nothing to write about” usually means no case studies. The repeated questions, recurring mistakes and declined work are the material.
4. Contradictions between a business’s own pages become confident errors in anything that summarises it — settle the canonical facts and never contradict them.
5. Cover one subject completely before starting a second: twelve subjects sampled once reads as thin from the outside.
6. Treat it as extraction, not writing: the expert supplies substance in a short conversation and reviews the result — everything between is somebody else’s job.
7. Summarising the field, scaled production, phrasing variants and borrowed credibility produce page count and nothing else.
8. Nobody can give a timescale honestly. The first fair review point is after one subject is covered completely, not after two pages.