How to Build Topical Authority From Scratch
Almost every failure in this work is a sequencing failure rather than a quality one. The pages were fine. They arrived in the wrong order, else the subject was chosen badly, else the programme stopped at the point where stopping costs most. This is the order, plus what decides each stage.
Choose A Subject You Can Actually Own
The subject has to be narrow enough to cover completely and broad enough to matter commercially. Both conditions, not one. Most failures are decided here rather than later, which is why this takes longer than anybody expects.
Why this is the whole programme in one decision. Everything inherits it.
The research, the pages, the linking and the budget all follow from what the subject is. Choosing badly means executing the rest perfectly against the wrong target, which is the most expensive way to fail.
What owning something means. Being the obvious source.
Not ranking for a phrase. Being the site somebody would send a colleague to on that subject, which requires having addressed it more completely than the alternatives.
The condition people skip. Genuine expertise.
You cannot own a subject you do not know. The awkward questions that separate a real source from a researched one cannot be produced without having done the work. That limit is set out in why topical authority outperforms keywords.
The commercial condition. The people searching must be buyers.
A subject you could own completely, searched by people who will never purchase anything, is an expensive hobby. Both tests apply and businesses tend to check only one.
Narrow Beats Broad Every Time
Complete coverage of a small subject beats partial coverage of a large one. Most businesses choose too broadly. There are reliable signs of having done so.
Why breadth feels safer. It looks like more opportunity.
A larger subject appears to offer more ways to be found. It also means competing against everybody who covers any part of it, while covering all of it yourself is not achievable.
How to tell you have gone too broad. Three signals.
You cannot state the subject in one sentence. You cannot list the questions that belong to it. Or the list you produce would take years to work through. Any of those means narrow further.
What too narrow looks like. Genuinely possible.
The map is finished in a few pages and nobody is searching any of them. That is the opposite failure and it is far less common, because narrowing feels like giving something up.
Why narrowing does not cost you breadth. The counterintuitive part.
Being known for one subject brings questions about adjacent ones, because people who trust a source ask it about neighbouring things. Breadth arrives as a consequence of depth rather than as an alternative route to it.
Find The Questions That Actually Exist
The map is built from what people genuinely ask rather than from what the business wants to say. Those two lists overlap less than anybody expects. The gap between them is where most content programmes go wrong.
What businesses produce instead. A list of what they sell.
Left alone, a business maps its own services, its own terminology and the things it finds interesting. That produces a structure organised around the seller rather than around the questions.
Why the difference matters so much. Nobody searches your structure.
People search their problem, in their words, at whatever stage they have reached. Content organised around what you offer meets them only at the point where they already know what to ask for.
What the research is looking for. Questions, not phrases.
The things people are trying to establish, including the ones that reveal a misunderstanding you could correct. Those are frequently the most valuable pages and they never appear on a list drawn from a business.
Where the mechanics live. Not here.
How that research is actually conducted belongs to our keyword research material. This page owns where it sits in the sequence, which is before anything is written and after the subject is fixed.
The Difference Between A Map And A List
A list of phrases is not a structure. The map is the relationships between the questions. That is what turns a set of pages into a cluster rather than a pile.
What a list gives you. Items with no order.
Research usually produces a spreadsheet of phrases with numbers attached. Nothing in it says which question comes before another, which are aspects of one thing or which a reader needs first.
What a map adds. The connections.
Which questions are versions of each other, which belong beneath which, then which sequence a person actually moves through. That information does not come from a research tool. It comes from somebody who understands the subject.
Why this decides the page count. Grouping happens here.
Several phrases that are one question become one page. A phrase concealing three separate questions becomes three. That collapsing and splitting is the work. Doing it wrong produces either duplication or gaps.
What it also produces. The linking plan.
The relationships in the map are what the internal links will express, which is why the map has to exist before anything is written rather than being reconstructed afterwards.
Why a tool cannot do this part. It is judgement.
Grouping questions requires knowing which are the same question in different words, which is a subject matter decision rather than a similarity calculation.
Build The Anchor First
The anchor page establishes what the subject is. The supporting pages establish depth within it. Building supporting pages first leaves them unanchored. That is a common and expensive mistake.
What the anchor does. Declares the territory.
It covers the subject at the level somebody arriving with a general question needs. It is what the supporting pages refer back to. Without it there is nothing establishing that the detailed pages belong together.
Why businesses start with supporting pages. They are easier.
A narrow question is quicker to write and produces a visible result sooner. The anchor is longer, harder and feels less urgent, so it gets postponed and frequently never written.
What unanchored pages look like. Individually good, collectively nothing.
A scattering of detailed pages with nothing tying them together. Each may be excellent. None of them establishes that the site covers the subject, which was the objective.
Why fixing it later is expensive. Retrofitting.
Writing the anchor afterwards means revisiting every supporting page to link it properly. The months in which those pages sat unanchored earned less than they should have.
What the anchor is not. A summary of the others.
It is the page somebody with the general question needs, which is a different thing from an index of your detailed pages.
Link It Together Deliberately
The internal linking is what turns pages into a cluster. It should be planned from the map rather than added at the end. The difference between those two approaches is visible in the result.
What planned linking looks like. Relationships expressed.
Each link exists because the map says those two pages relate, appearing where a reader would benefit from following it. Somebody could explain every link on the page.
What retrofitted linking looks like. Blocks at the bottom.
A list of related articles appended to every page, generated rather than chosen. That is a component rather than a structure. Readers learn to ignore it.
Why the difference shows. Context.
A link inside a sentence carries the meaning of that sentence. A link in a list carries nothing except adjacency, so the same two pages connected either way are not equally connected.
Where this connects to structure. The same decision.
How pages are grouped and how they link are one exercise, covered in what is information architecture in SEO.
The practical consequence for commissioning. It affects the brief.
Writers need to know what else exists and what it says, which means the map has to be shared rather than pages being commissioned individually against a title.
Half A Cluster Performs Worse Than You Expect
Partial coverage underperforms what its individual pages would suggest. That tempts businesses to abandon the approach at precisely the point where finishing is what produces the result.
The one sentence of reasoning. The mechanism is not yet running.
The value comes from a body of work rather than from pages, so half a body of work does not deliver half the value. The full argument is in why topical authority outperforms keywords.
What that means for planning. A hard rule.
Do not start a subject you cannot finish. Scope it to what you will genuinely complete rather than to what would be ideal, because an ideal map abandoned partway is worse than a modest one completed.
The budget consequence. Commit to the whole thing.
Funding a cluster in monthly increments with a review at each stage sounds prudent and builds in the temptation to stop at the worst possible moment. The commitment should be to the cluster.
What to do if the money genuinely runs out. Narrow rather than abandon.
Redefine the subject to something smaller that the existing pages complete, rather than leaving a partial map. A finished small cluster beats an abandoned large one.
Why we raise this before starting. It is not an excuse afterwards.
Said in advance, this is a planning constraint. Said in month five, it is indistinguishable from a supplier explaining away a disappointment.
How Long It Takes
Months rather than weeks. Beyond that, anybody quoting you a period is inventing it, so what follows is what actually decides the answer for your situation.
The size of the subject. The largest factor.
A subject with a dozen genuine questions and one with two hundred are different projects. The map produced in block four is what tells you which you have, which is another reason it comes before any scheduling.
The starting position. Whether anything is already there.
A site with existing standing on adjacent subjects is not starting from the same place as a new one. That difference is substantial and it is why comparisons between businesses are unreliable.
The rate you can genuinely produce. Usually the real constraint.
Not how fast pages could be written. How fast pages that pass the awkward-questions test can be written by people who know the subject. That is slower and it is the real limit.
What we will not do. Give you a figure.
No period appears anywhere on this page, because nothing supports one and a number invented to reassure you becomes a stick to be beaten with later.
What to ask a supplier instead. A better question.
Not how long. How many pages the map contains and what rate they can sustain. Those two answers produce a range you can both stand behind.
When To Start The Next One
A second subject started before the first is finished usually damages both. The discipline is sequential. It is harder to hold than any other rule on this page.
Why businesses start a second early. The first looks slow.
Partway through, the first cluster is in the difficult stage from block seven. Starting something else feels like making progress while the disappointing thing settles. It guarantees two half built clusters.
What splitting does to both. Neither reaches the threshold.
The same production capacity divided across two subjects means both take twice as long to complete, so both spend twice as long in the stage where they underperform.
When the first is genuinely finished. The map is covered.
Every question on the map has a page and the pages are linked as planned. Finished is a state you can check against the map rather than a feeling or a date.
What to do while waiting for results. Not start something.
Improving what exists, deepening the pages that are attracting the most difficult questions, else handling the structural work. Useful things that do not divide the production capacity.
The exception. Genuinely separate capacity.
A business with two independent teams and two independent subjects is not splitting anything. That is rarer than businesses believe when they want to start a second.
What It Does Not Fix
Three things this approach will not overcome, however well the sequence above is executed. Knowing them beforehand is worth more than discovering them at the end.
A site nothing can reach. The precondition.
Authority in a subject does nothing for pages that are never fetched. That is a floor rather than a competing priority. It applies before any of this becomes relevant.
A business with nothing to say. The hardest one.
Where there is no genuine expertise, the pages are researched rather than known and they read that way. No amount of structure compensates. This is the situation where we would decline the work.
A market where the searcher does not want what you sell. The least obvious.
Where demand sits at the explanation stage and you are selling a service, complete coverage of the service reaches almost nobody. That is a stage problem rather than a coverage one and it is covered in how search intent evolves.
What it also does not replace. Judgement about volume.
Coverage means answering the questions that exist rather than generating pages until a number is reached, which is the line drawn from the production side in what is programmatic SEO. The full series is on the advanced SEO guide.
Do not start
what you
cannot finish.
We scope a subject to what will genuinely be completed rather than to what would be ideal, build the map before anything is written and commit to the whole cluster. If your subject is too broad to finish, else you need enquiries faster than this produces them, we will say so first.
What the national tier covers:
The national tier is £1,550 a month. It suits businesses competing beyond one town. We will say so if a lower tier would serve you better.
Every guide.
One practice.
Semantic search, machine learning, topical authority, programmatic production, rendering, log files, crawl allocation, information architecture, content pruning and how strategy changes with scale.
