Category Pages That Rank: Writing Intros at Scale
The category page is a directory's main SEO surface, and most are left empty. What a 150-300 word intro must contain, how to write 40 without duplicating them, and when a category should not exist at all.
Your AI chat can build this directory.
Describe the niche, watch the agent design the fields and fill the catalogue. Free plan, no card.
Open almost any niche directory and look at a category page. A heading, a filter bar, a grid of cards, nothing else. No context, no reason for the page to exist beyond the listings it holds. That page is the biggest SEO asset the directory owns, and it has been left empty because writing forty of them sounds like a fortnight of work nobody wants to do.
It is a fortnight of work. It is also the difference between a directory that ranks and one that quietly never does.
The page that actually ranks
Homepages of small directories rarely rank for anything useful: the term is too broad for a new site to win. Listing pages rank occasionally, for brand queries the listed business was going to win anyway.
The category page is different, because it matches how people actually search. Nobody types "directory of accountants". They type "accountants for freelancers in Manchester" — a query with a modifier, an intent and an implied shortlist. That is a category, or a filtered view of one, and it is the level of specificity where a small site can compete against a large generic one.
It is also the page AI answers reach for. Asked to recommend options in a narrow space, an assistant wants a page that names the options, states the criteria and shows the trade-offs. A bare grid of cards gives it nothing to quote.
What the intro has to do that the grid cannot
The listings answer "who is available". The intro has to answer everything a visitor needs before that list means anything.
Define the boundary. What is in this category and what is deliberately not. "Accountants who work with sole traders and freelancers, not practices whose smallest client is a limited company with staff." That sentence alone removes half the visitor's uncertainty.
Name the selection criteria. Three or four things that genuinely differentiate entries here: pricing model, certification, whether they handle a particular filing, minimum engagement size. This is the part that gets quoted, and the part that turns a grid into a decision.
Give the honest state of the market. How many entries there are, roughly what they cost, what most have in common, where the gaps are. If two thirds of a category charge a retainer and a third charge hourly, say so.
Tell the visitor what to do next. Which filter to apply, which adjacent category to try if this one is not right.
That is 150 to 300 words. Below 150 you are labelling, not explaining. Above 300 you are pushing the listings below the fold, which is the thing the visitor came for.
Writing 40 intros without writing the same one 40 times
This is the real problem. Forty near-identical paragraphs with the category name swapped in is the definition of scaled thin content, and it is exactly what search engines have got better at detecting.
The fix is to template the structure and never the sentences.
Write from the data, not from the template. Before writing, pull three facts out of the entries actually in that category: how many there are, the range of a typed attribute they share, the most common value of another. Every intro then carries at least three statements that could not appear on any other page of your site. This is where a typed listing schema pays off — if entries carry five to eight structured attributes rather than a blob of description text, the differentiating facts are already sitting there.
Apply the swap test. Take any sentence you have written and mentally replace the category name with a different one. If the sentence still reads true, delete it. "Finding the right supplier can be a challenge" survives the swap on every page in your directory, which is precisely why it is worthless.
Vary the shape, not just the words. Some categories are best explained by price band, some by geography, some by a technical distinction, some by who the customer is. Let the category decide the angle. Forty intros that all open with a definition read as a template even when the sentences differ.
Draft fast, then edit properly. Generating a draft from your own listing data is fine. Publishing it unedited is not. Budget ten minutes per category to add the thing only you know: the gap in the market, the caveat, the reason two entries that look alike are not.
Do them in batches of five. Forty in one sitting produces forty versions of the same voice by entry twelve. Five a day for eight days produces pages that read as though someone was paying attention.
Do not create a category with four entries
A category with three entries is a worse page than no category at all. It disappoints the visitor, it dilutes the internal link graph, and it gives search engines a set of near-empty pages to judge the site by.
The working threshold is 5 to 10 entries minimum before a category exists, and you want a credible path to 15. Below that, the listings belong in the parent category with a filter or a tag.
The corollary matters more: 8 well-populated categories beat 30 sparse ones. Operators over-split early because splitting feels like progress. Then every page is thin, every intro is padding, and the structure has to be rebuilt. Start coarse. Split when a category passes roughly 25 entries and the split is one a visitor would recognise, not one that only makes sense to you.
Failed searches are the cleanest signal for what to create next: if people repeatedly search your site for something that returns nothing, that is a category or collection with demand already attached.
Categories or collections
Categories are the taxonomy: mutually exclusive, structural, part of the URL hierarchy, ideally one per listing. They answer "what kind of thing is this".
Collections are the cross-cuts: indexable landing pages that group listings by a shared trait regardless of category. "Open source", "under £50 a month", "family-run since before 1990", "ships to the EU". A listing can sit in many collections and it breaks nothing.
The distinction matters because most modifier queries people search are collections wearing a category's clothes. "Free accounting tools for freelancers" is a collection. Forcing it into the taxonomy creates an awkward category overlapping three others; building it as a collection gives you a clean, indexable page drawn from entries you already have. DirectoryFast keeps the two as separate concepts for this reason — the taxonomy stays stable while collections multiply around it.
Rule of thumb: if a listing could reasonably belong to two of your categories, one of them should have been a collection.
How categories should link to each other
An orphaned category page is one search engines find late and value little. Three links fix most of it.
- Parent down, child up. Every category links to its subcategories, every subcategory links back. Obvious, frequently missing.
- Sideways, to two or three genuine siblings. In prose, with a reason: "if you need someone who also handles payroll, try [X]". A sideways link buried in footer nav is ignored; the same link inside a sentence is not.
- To the collections that draw from this category. The highest-value link on the page, because it moves the visitor from a broad set to a filtered one, which is the direction intent travels.
Keep every category reachable within two clicks of the homepage. A page four clicks deep is crawled rarely and ranked badly, whatever the intro says.
Thin page, useful page: the signals
The difference is not word count. A 400-word page can be thin and a 180-word page can be excellent. What separates them:
- Unique facts per page. Three or more statements true of this category and false of every other one. Zero means thin, whatever the length.
- Populated listings. Twelve complete entries carrying real attributes have substance four stubs do not.
- Content that changes. Entries added, updated and re-ordered signal a maintained site. A page untouched for two years signals an abandoned one.
- Genuine internal demand. If nobody on your own site ever navigates or searches into a category, it is unlikely to earn external traffic either.
- Answers, not descriptions. A page that explains how to choose is useful. A page that describes what the category contains is a label.
If a category cannot pass the first item, do not fix the intro. Delete the category and fold its entries upward.
Where this goes wrong
Publishing all 40 intros in one unedited batch. The whole set carries the same rhythm, openings and padding. Reviewers and algorithms both notice.
Creating categories before entries exist. An empty taxonomy is a set of thin pages waiting to drag down the rest of the site.
Keyword-stuffing the intro. Repeating the category name eight times in 200 words reads badly to humans and adds nothing for machines.
Letting the intro drift out of date. An intro that says "eleven providers" when the category holds thirty is worse than no number. Tie the count to the data or leave it out.
Splitting too early. Thirty categories of four entries is the most common structural mistake in new directories, and the most expensive to unwind once URLs are indexed.
Writing intros for pages that cannot be indexed. On a noindex free subdomain, none of this work counts. Get the custom domain sorted first, then write.
FAQ
How long should a category intro be? 150 to 300 words. Long enough to define the boundary and the selection criteria, short enough that the listings stay visible without scrolling far.
Can I generate category intros with AI? As a first draft from your own listing data, yes. Published unedited across forty pages, no — that is precisely the pattern that gets a directory classified as scaled thin content.
Where does the intro go, above or below the listings? Above, kept tight. A long block of text pushed below the grid is read by almost nobody and helps less than operators assume.
Should every category have the same structure? Same skeleton, different angle. Identical structure with identical phrasing is a template; identical structure with category-specific substance is consistency.
How many categories should a directory of 200 listings have? Usually 8 to 15. Enough that each holds 10 or more entries, few enough that a visitor can hold the whole structure in their head.
What do I do with a category that never grows? Fold it into its parent and redirect the URL. A permanently thin page costs you more than the handful of listings it holds are worth.
Build your directory in a conversation →