Set the case up before comparing anything
Take a workflow product at £8 million of annual recurring revenue, split evenly between two segments. Segment A is independent agencies: 800 accounts paying an average £5,000 a year, self-serve with a light sales touch. Segment B is in-house teams at large enterprises: 40 accounts paying an average £100,000 a year, sold by a field team over long cycles. Both lines are £4 million.
State the question you are answering, because the case is otherwise ambiguous. I will read "focus" as where the product roadmap, the majority of engineering capacity and the go-to-market spend point for the next four quarters, and I will optimise for durable recurring revenue in three years rather than bookings next quarter. If the interviewer wants a different objective — cash in twelve months, or a valuation event — say so, because the answer genuinely flips.
Equal revenue is the trap in the prompt. Revenue is a product of account count, price and how long accounts stay, and two lines that match today can be on completely different trajectories. So the first move is to decompose rather than to argue.
The comparison that decides it
Five things matter and they can be assembled from data the company already has. Assume the following, each of which I would state as an assumption to be checked.
| Dimension | Segment A, agencies | Segment B, enterprise |
|---|
| Accounts and price | 800 at £5,000 | 40 at £100,000 |
| Annual logo churn | 25 per cent | 5 per cent |
| Net revenue retention | 85 per cent | 115 per cent |
| Cost to acquire | £3,000 | £40,000 |
| Gross margin after support | 80 per cent | 65 per cent |
| Addressable accounts | ~20,000 | ~300 |
Now work the arithmetic. In Segment A, each account contributes £5,000 times 80 per cent, or £4,000 of gross profit a year, so a £3,000 acquisition cost pays back in about nine months. In Segment B, each account contributes £100,000 times 65 per cent, or £65,000, so a £40,000 acquisition cost pays back in roughly seven and a half months. Both are healthy and they are close enough that payback does not decide it, which is the useful part — payback is where most candidates stop.
Retention is where they separate. Segment A loses a quarter of its accounts a year, which means 200 accounts a year must be replaced before any growth happens at all, costing 200 times £3,000, or £600,000 a year, purely to stand still. Segment B loses two accounts a year and the survivors expand, so the line grows about 15 per cent without a single new logo, which is £600,000 of growth from the same base that Segment A spends £600,000 to hold flat. That is a £1.2 million a year swing in the same £4 million of revenue, and it is invisible in the prompt.
Headroom pulls the other way and must be stated or the analysis is one-sided. Segment B has roughly 300 addressable accounts and you hold 40, so even at a 30 per cent share the line caps around 90 accounts, about £9 million. Segment A has 20,000 addressable accounts and you hold 800, so 4 per cent penetration — at 15 per cent it is 3,000 accounts and £15 million, and price may have room to move too. Segment B is better revenue with a ceiling in sight; Segment A is worse revenue with a much larger ceiling it is currently unable to reach.
The finding that reframes the question
Sit with the Segment A numbers for a moment, because they contain the real insight. A segment with nine-month payback and 25 per cent churn does not have a segment problem, it has a retention problem, and those are different diagnoses with different remedies. Before choosing between segments I would spend a week on one question: is Segment A's churn uniform, or concentrated?
If churn is concentrated — say most of it falls in accounts under three seats, or in accounts acquired through one channel, or in the first 90 days — then Segment A is really two segments wearing one label, and the retained part may look much more like Segment B on every dimension except price. Suppose half the 800 accounts are three seats or fewer and account for 80 per cent of the churn: that sub-segment churns at 40 per cent while the remainder churns at 10 per cent. The recommendation is then not "focus enterprise" but "stop acquiring the small tail and serve the rest", which changes the spend, the roadmap and the acquisition targeting all at once and costs far less than a segment pivot.
If churn is uniform at 25 per cent, the segment is structurally weak for this product and the choice becomes clean.
The recommendation, and what would reverse it
On the assumptions as stated, with uniform churn, I would focus on Segment B for the next four quarters, and I would say why in one sentence: it is the only line where product work compounds, because expansion inside retained accounts converts a feature into recurring revenue that does not have to be re-won every year.
Three things go with that recommendation, and their absence is what separates an adequate answer. First, focus is not abandonment: Segment A keeps a maintenance allocation, keeps its acquisition channel running at reduced spend, and explicitly does not get a price rise or a support downgrade in the same year, because a deprioritised segment that also gets worse churns faster than the model assumes and takes the revenue with it.
Second, the ceiling is a countdown: at 15 per cent growth Segment B reaches its practical cap inside four or five years, so part of this year's work is finding the next expansion axis, whether that is a second product for the same buyer or an adjacent enterprise vertical. Third, name the reversal condition — if the churn diagnosis comes back concentrated rather than uniform, or if enterprise win rate falls below the level that makes a £40,000 acquisition cost work, this decision is revisited at the half-year, not at the end of it.
What the interviewer is grading
The figures above are arithmetic on assumptions I chose, and an interviewer will not check them against anything. What they will check is whether you turned equal revenue into unequal revenue quality, whether you brought cost to serve and headroom in without being prompted, and whether you noticed that one of the two options was really a retention question in disguise. Candidates who reach a defensible answer purely from revenue and payback score adequately; the round is won by the one who asks whether Segment A is one segment at all.
The most common failure is picking enterprise because the numbers are prettier and then being unable to say what is lost. What is lost is optionality: 20,000 addressable accounts, a self-serve motion that scales without headcount, and a product that stays simple because no single customer can demand things. Forty accounts paying £100,000 will bend the roadmap towards themselves, and in two years the product may be unable to serve the self-serve segment even if you want it back.
Equal revenue is never equal, and the segment question is often a retention question wearing a segment's clothes — check whether the weak line is one population before you decide it is the wrong one.
© 2026 Preptima. Originally published at preptima.com (opens in a new tab).