Positioning a generalist agency

The advice arrives in every conversation with a consultant, every panel at every small-agency conference, every LinkedIn post from someone who sold their firm and now coaches the ones that haven't: niche down. Pick an industry. Become the agency for fintech, or for dental practices, or for Series A SaaS, and stop trying to be everything to everyone. It is good advice, stated by people whose own income no longer depends on next quarter's utilisation rate. For a ten-person shop with a payroll due on the last Friday of the month, "stop taking work outside your niche" is not a positioning exercise. It is a bet that the new, narrower pipeline fills fast enough to replace the old, wider one before the gap empties the account. Sometimes that bet pays off. Often the firm quietly keeps taking the off-brand work anyway, feels guilty about it, and concludes that positioning is a luxury for agencies further along than they are.

That conclusion is wrong, but the standard advice earns it. The actual problem with "niche down" is not the goal — a sharper position is genuinely worth more than a vague one, for exactly the reasons laid out in our piece on pricing and positioning for small companies: a specific claim is checkable, and a checkable claim is what lets a buyer say yes without a lengthy education first. The problem is that the advice collapses "specialise" into one specific axis — industry — as though it were the only kind of narrowing available. It isn't. There are at least four, they behave very differently under cash-flow pressure, and the generalist agencies that get this right almost never pick the one everyone quotes.

Four ways to narrow, and they are not interchangeable

Industry is the axis everyone means by default. You become the agency for logistics companies, or veterinary clinics, or B2B SaaS. It produces the cleanest case studies and the easiest referrals, because "we do this for companies exactly like you" is the most persuasive sentence in sales. It is also the most expensive axis to adopt, for a reason the advice-givers rarely mention: it requires you to fire your existing client base. If your last twenty clients span six industries, committing to one of them means the other five stop referring you work that fits your capacity, starting immediately, while the new industry's referral network — which takes years to build, not quarters — has not yet started sending anything back. The ceiling this axis sets is real and the argument in our companion piece on pricing for why segment bounds price applies with extra force here: an industry segment bounds not just what you can charge but who is even allowed to hire you, and that second constraint is the one that empties a pipeline.

Problem is a different cut entirely. Instead of restricting who can hire you, you restrict what they hire you to solve — technical SEO, not marketing in general; brand naming, not brand strategy; checkout conversion, not e-commerce. Almost any industry can have this problem, so the client pool does not shrink the way it does on the industry axis. What sharpens is the pitch: instead of "we do digital marketing," you say "we fix sites that get traffic and don't convert it," and that sentence does real qualifying work in a first call without needing the prospect to belong to any particular vertical.

Process narrows around how you work rather than what you produce or who you serve. An agency that only does fixed four-week sprints, or that only works retainer with no project work, or that insists on shipping something live in week one rather than presenting decks first, has a process specialism. It is the least visible axis from the outside and the easiest to adopt without losing a single existing client, because most clients don't actually care about your internal process as long as the output is right — which also means it does the least work in a cold pitch, since "we work in two-week sprints" rarely closes a deal on its own.

Client stage narrows by where the buyer's company is in its life, independent of industry: pre-seed founders who need a first site fast and cheap, versus companies past their Series B doing a full rebrand with a procurement process attached. This axis behaves a lot like industry in that it changes who is allowed to hire you, but it is cheaper to adopt because stage-based referral networks — accelerators, specific investors, specific law firms that work with startups — are denser and faster-moving than industry associations, so the new pipeline can start filling sooner.

Why the loudest advice is the most expensive to take

None of this means industry positioning is a mistake. Plenty of the agencies everyone points to as positioning successes did pick an industry and it worked. But it is worth being honest about why that axis dominates the advice: it is the easiest one to explain in a keynote, because "the agency for X" is a one-line story with an obvious moral, and it photographs best in a case study — not because it is the axis that survives contact with a firm's actual cash position. A problem or process specialism is a harder story to tell from a stage — "we fix checkout conversion for anyone who has the problem" doesn't have the same clean logo wall as "we build for fintech" — but it is the one that lets a generalist agency sharpen its position without restricting its pipeline down to a fraction of its former size. The industry axis and the problem axis are not equally risky, and treating them as interchangeable versions of "specialise" is what leads a small firm to bet its runway on the expensive one.

Sharpen the pitch before you restrict the intake

The sequence that survives a real P&L runs in the opposite order from the advice. Do not restrict who you'll take on first and hope the sharper story fills the gap. Sharpen the pitch first, on whichever axis fits your actual client history, and keep taking the work that pays the bills while the new pitch proves itself in real meetings.

Concretely: look at the last twenty deals, the same exercise described in the pricing piece for finding your real segment, but read it for a problem or process pattern instead of an industry pattern. If eight of your best-paying jobs were, underneath whatever the client called them, checkout redesigns — regardless of whether the client sold shoes or software — that is a problem specialism sitting inside a generalist client list, and you can start saying it out loud in pitches next week without turning away a single lead. The pitch changes before the intake does. You are testing whether the sharper claim closes better on cheap, reversible inventory — the words in a proposal — before spending the expensive, hard-to-reverse inventory: the client relationships you'd have to end to make an industry restriction real.

Only once the sharper pitch is demonstrably converting — not felt to be converting, actually converting, tracked the same way you'd track any change to a proposal template — does it make sense to start saying no to work outside it. Some firms test the new claim the way a software company tests a new plan structure before committing to it fully: a scoped, low-commitment pilot offered only to prospects who fit the new position, run alongside the normal proposal process rather than replacing it, so the firm can see whether the sharper story converts before betting the whole intake on it. It is the same underlying caution as the choice between a free trial, a demo and a freemium tier — let the buyer prove the fit cheaply before either side commits.

What to do with the clients who don't fit and still pay the bills

Every generalist agency doing this work has a set of clients who are, by the new definition, off-position — the industries or problem types you're not leading with anymore, paying invoices that currently cover payroll. The instinct is to feel a low-grade embarrassment about them, hide them from the new pitch deck, and hope they don't notice they've become a legacy segment. That instinct is backwards. Those clients are the reason you can afford to reposition at all, and treating them as a problem to be quietly wound down rather than a bridge to be maintained is how firms run out of runway mid-transition.

The better approach is to say plainly, internally, which accounts are bridge accounts and which are target accounts, and to staff and price them differently on purpose rather than by accident. Bridge accounts get serviced well and priced at whatever they're currently paying — this is not the moment to raise their rates to punish them for not fitting the new story. Target accounts, the ones that match the sharpened problem or process claim, get the new pitch language, the new case studies as they accumulate, and first call on senior time. A firm that tries to serve both groups identically, with identical marketing pointed at both, ends up with a homepage that describes the target segment and a client roster that mostly isn't it — which is precisely the drift a rebrand sometimes tries and fails to paper over, a failure mode we've gone through in our piece on the rebrand that was really a positioning problem. New visual identity does not fix a mismatch between what the homepage claims and who actually pays the invoices; only the underlying client mix does, and that mix changes slowly, on purpose, not with a logo update.

The one-sentence test

Whichever axis you pick, there is a single test for whether the repositioning actually happened or just got discussed in a meeting: can the homepage say it in one sentence, above the fold, without a qualifier that quietly widens it back out. "We help growing companies with their digital presence" is not a position on any axis — it is the sentence a generalist writes when it wants credit for specialising without doing the work of choosing. "We fix checkout flows that get traffic and don't convert it" is a problem-axis position, and it survives being read by a stranger who has never heard of the firm. "We build the first four weeks of a brand-new product, then hand off" is a process-axis position, equally legible cold.

The honest way to check your own homepage is the same test a pricing page fails when it hides its numbers behind a contact form: if the one sentence needs a follow-up question before a stranger understands who you serve and what you don't, the position isn't finished yet, whatever the client list already proves about where the real expertise sits.