Most owners who ask how to niche down an accounting practice already accept the argument for it. What stops them is the risk calculation. The general practice pays the wages now, and narrowing the market feels like voluntarily shrinking the pipeline before you have proven anything.

That framing is the problem. Niching is a decision about what you say publicly and what you build your acquisition system around. It is not a decision to refuse work. The firms that get this wrong are the ones who announce a specialism, redesign the website, and then find the sector was chosen on enthusiasm rather than evidence.

This guide sets out the sequence we use when repositioning a practice: read the portfolio for commercial evidence, validate that the sector generates measurable demand, define the position tightly enough to be useful, then rebuild the acquisition layer around it. It also covers the part most articles skip, which is how to carry your existing clients through a repositioning without them feeling abandoned.

None of this requires a leap of faith. Every stage produces data you can act on before the next one starts.

Before looking outward, run an analysis of the clients you already have. Export the last two full years of fee income by client and add three columns: sector, realised margin, and hours of unbilled admin. Most practice management systems will give you the first two directly. The third is usually a judgement call from whoever handles the work.

What the analysis usually reveals

Firms are consistently surprised by two things. The first is that the largest clients by fee are rarely the strongest by margin. The second is that admin load clusters by sector rather than by client size, because sector determines record quality, software, and how many mid-year questions arrive.

• Which sectors deliver the highest realised margin per hour, after write-offs
• Which sectors generate repeat advisory work rather than compliance only
• Which sectors your team handles without escalating to you
• Which sectors you would decline if the pipeline were fuller

A sector that scores well on the first three is a candidate. A sector where you have three profitable clients and a partner with genuine sector background is a strong candidate, because you already have the technical depth and the case studies.

The clustering test

Look for clusters of four or more clients in the same sector who arrived independently rather than through one referral chain. Independent arrival suggests the market is finding you for a reason, which is a better signal than a single well-connected client who introduced everyone else. If the cluster came from one source, you have a relationship, not a niche.

Portfolio evidence tells you what you are good at. It says nothing about whether enough of those businesses exist, or whether they actively look for an accountant. Both need checking before you commit.

Population size

Companies House data is public and downloadable, and SIC codes let you count how many active companies sit in a sector, where they are, and how many incorporate each month. Incorporation rate matters more than total population, because new companies are the ones actively choosing an accountant. A sector with a modest total population but steady new registrations gives you a renewable market.

Search behaviour

Check whether the sector searches for sector-specific accounting help. There is a meaningful difference between a market that searches for a generic term and one that searches for a sector term. If the sector-specific searches barely register, you can still win the niche, but you will need outbound and referral routes rather than relying on search.

Ability to pay

Some sectors have real complexity and thin margins. Complexity without margin produces demanding clients at compliance fees. Look at the sector's typical turnover and whether the complexity is something the client sees as risk. Businesses that face a specific penalty exposure, such as CIS treatment in construction or VAT registration thresholds across marketplaces in online retail, understand why specialist advice costs more.

Run all three checks. A sector that passes population and search but fails on ability to pay will fill your calendar and flatten your margin.

Most firms that claim a specialism define it too broadly to change anything. "We work with construction" describes a quarter of the SME economy and tells a prospect nothing.

A workable definition has two axes. The first is who the client is: sector, entity type, turnover band, and stage. The second is what you actually do for them, which is usually a specific recurring problem rather than the full compliance list.

Examples of definitions that work

• Subcontractor-heavy construction companies turning over between one and five million, needing CIS treatment and reverse charge VAT handled without monthly correction work
• Dental practices in single-site ownership approaching a partnership change or sale, needing valuation and structuring alongside compliance
• Marketplace retailers selling into more than one jurisdiction, needing VAT registration thresholds monitored and platform data reconciled

Each of these is specific enough that a business owner reading it recognises themselves in one sentence. That recognition is the entire mechanism. It is also specific enough to write a landing page, a set of search pages, and a targeting brief against, which the broad version is not.

The stage variable

Sector is the obvious axis, but business stage is frequently the better one and far less crowded. Firms that position around a transition, such as approaching a first VAT registration, preparing for sale within three years, or moving from sole trader to limited, face almost no direct competition, because most practices organise around service lines rather than moments.

This is the objection that stops most firms, and it is largely a misunderstanding of what changes.

Repositioning changes what you market. It does not change your engagement letters. Your existing clients outside the niche keep their service, their contact, and their fee structure. What changes is where new fee income comes from.

What to tell existing clients

In practice, very little. Clients rarely audit their accountant's website. The ones who notice will ask whether you still handle their work, and the answer is yes. If you want to be proactive, a short note explaining that the firm is building deeper expertise in a defined sector reads as investment rather than retreat.

Managing the transition ratio

Work in stages rather than switching over in one move.

• Months one to six: niche clients form a small minority of the portfolio. All new marketing is niche-directed. General enquiries are still accepted.
• Months six to eighteen: the niche grows to somewhere around a third of new fee income. You start declining general work that carries poor margin.
• Beyond eighteen months: the niche carries the majority of new business. General clients remain but are no longer replaced when they leave.

Some practitioners deliberately cap the niche at around forty per cent of the client base to avoid sector concentration risk. That is a defensible position if your chosen sector is exposed to a single policy or economic cycle. It is a weaker argument in sectors with structural, permanent complexity, where the specialist work is not going anywhere.

A niche that lives only in your head produces no commercial change. The position has to be visible in the places where the sector looks, which means rebuilding three layers.

Search architecture

A single page saying you work with a sector will not rank. What ranks is a structured set of pages covering the specific questions that sector searches for: the treatment they are unsure about, the deadline they keep missing, the software integration they cannot get working. Each page targets one query and answers it properly. Volume matters less than coverage of the sector's actual search behaviour.

An entry engagement

Prospects rarely move straight from a website visit to a full compliance switch. A defined, scoped piece of work gives them a lower-risk starting point. A review of the sector's most common exposure area works well, because it produces a written output the client keeps and usually surfaces further work. Keep the scope fixed and the deliverable concrete.

Proof from within the sector

Two written case studies from the niche outperform a page of general testimonials. The prospect is checking whether you have handled their situation before. A short account of the problem, what you did, and the measurable outcome answers that faster than any claim about experience.

We combine registry data and search data at the start of this process for exactly this reason. It tells you which sectors have both population and demand before anyone writes a page.

Specialisation produces results on a slower curve than paid acquisition, so measuring the wrong thing at the wrong time will make a working position look like a failure.

Leading indicators, months one to three

• Proportion of enquiries that arrive from the niche, whether or not they convert
• Whether enquiries mention the specific problem you positioned around
• Time from first contact to proposal, which usually shortens with sector-specific enquiries

Lagging indicators, months six to twelve

• Average fee on new niche clients compared with your general average
• Realised margin per hour on niche work as the process standardises
• Proportion of new fee income originating in the niche
• Referrals arriving from within the sector rather than from your general network

The margin figure is the one to watch. Specialisation should reduce the time each engagement consumes, because you stop rebuilding your approach for every client. If margin per hour has not improved by month twelve, the problem is usually that the delivery process was never standardised, rather than that the niche was wrong.

One firm we work with rebuilt its website around a defined position and saw monthly visitors quadruple with ten to fifteen enquiries in the first month after launch. Another moved from five to seven enquiries a month to fifteen to sixteen across the first two and a half months once the search architecture matched the position. Both outcomes came from alignment between the position and the system, not from the position alone.

Each stage produces evidence you can act on before starting the next. Nothing here requires you to commit before you have data.

Four failure patterns account for most abandoned repositioning attempts.