Ask ten practice owners how to choose a niche for your accounting firm and most will describe the same process. Look at your current client list, find the sector you have three or four of, and declare yourself a specialist. It works often enough that nobody questions it. The problem is that your existing client list is a record of who happened to find you, not a map of where demand sits.

Our position is straightforward. A niche is a marketing decision before it is a service decision. You are choosing which search terms you intend to rank for, which audience you intend to advertise to, and which set of problems your website will speak to. You can make that choice using external data, and you can test it without refusing a single enquiry.

What follows is the sequence we use when firms ask us to work this out with them.

Staying general feels like the safe option because it keeps every door open. In practice it has a cost that shows up in three places.

The first is acquisition. A generalist site competes for terms like "accountant near me", where you are one of forty firms saying broadly the same thing to the same person. Cost per enquiry is set by whoever is willing to bid most, and the enquiry arrives with no context about what the prospect actually needs.

The second is conversion. When a prospect cannot see any difference between your firm and the two others they contacted, the only remaining variable is fee. AccountingWEB's coverage of the generalist trap makes the same point, and it matches what we see in enquiry data: undifferentiated firms get asked about price on the first call far more often.

The third is delivery. A portfolio spread across twenty sectors means constant context switching. Every job has different record keeping, different software, different seasonal rhythm. Workflows never standardise because there is nothing repeating often enough to standardise around.

None of that means generalist firms cannot grow. It means their growth is expensive and manual, which is usually why the owner is still the main source of new work at year eight.

Before you consider what you enjoy, establish what exists. Two datasets settle most of the argument.

Business registry data tells you how many companies of a given type operate in your addressable area. Companies House publishes SIC-coded company data, and you can filter it by sector, incorporation date, and region. That gives you the size of the pool and, more usefully, the rate at which new entities are forming. A sector with high incorporation volume is a sector where businesses are actively shopping for an accountant right now.

Search data tells you whether those businesses look for help online in language you can rank for. Some sectors have real search volume behind terms like "accountant for [sector]". Others have almost none, because buying happens through trade bodies or word of mouth. A niche with no search demand is not disqualified, but it means paid social and outbound carry the whole load, which is a different build.

Cross-reference the two and you get a shortlist that is defensible on numbers. Sectors with sufficient entity volume, active formation, and searchable intent. That shortlist rarely matches the sector the owner assumed they would choose.

Niche is often treated as a synonym for industry. It does not have to be.

A vertical niche is defined by what the client does. Construction subcontractors, veterinary practices, freight operators. A horizontal niche is defined by the work you do or the situation the client is in. Firms coming off a failed software migration, businesses crossing the VAT registration threshold, companies preparing for sale in the next thirty six months.

Horizontal niches are underused and often easier to enter. They have less competition on search, they cut across sector so your addressable pool stays large, and the trigger event gives you a clear moment to advertise into. A business hitting the registration threshold has a deadline. That is a far stronger acquisition signal than "owns a company in sector X".

The strongest positions usually combine both. Not "we do ecommerce" and not "we do international VAT", but the intersection: ecommerce sellers dealing with cross-border VAT after a marketplace change. Narrow enough that the page writes itself. Broad enough that the pool supports a pipeline.

Write your position as one sentence describing who and what. If you cannot, the niche is still too vague to build a landing page around.

This is the part that resolves the fear of turning work away, and it is the reason most firms stall on this decision.

You do not need to fire clients, rebrand the practice, or refuse general enquiries to specialise. You need a set of pages and campaigns pointed at one audience while everything else carries on as it is. The firm keeps its name, its existing portfolio, and its general homepage. Underneath sits a dedicated page for the chosen niche, written in that sector's language, with its own enquiry form and its own campaign behind it.

Run it for a defined period and read three numbers. How many enquiries the niche page produces relative to general enquiries. What proportion of them convert. Whether the work that arrives is closer to standardised than what you were doing before.

If the numbers hold, you widen it. More pages, more terms, more spend. If they do not, you have lost some campaign budget and a few pages rather than a third of your client base. When a firm asks us whether a niche will work, this is the answer we give: build the smallest version that produces real data, then decide.

Two things shift, and the second one matters more than most owners expect.

Acquisition gets cheaper per client because the message matches the search. A page written for one sector, addressing the specific compliance and record keeping issues that sector faces, converts at a materially higher rate than a general services page receiving the same traffic. That is the mechanism behind most of the search results we see in practice, including Prads at Wings Online Filings, who went from five to seven enquiries a month to fifteen to sixteen, and nine new clients, within the first two and a half months of an SEO architecture and content system going live.

Onboarding gets standardised. When incoming clients share a sector, they share a document set, a software stack, and a set of first year questions. That means one onboarding workflow instead of twenty improvised ones, and it is the point at which growth stops adding administrative load in proportion to headcount.

The niche decision is usually framed as a marketing question. Its largest effect lands in operations, because repeatable clients make repeatable processes possible.