Tax as part of the experiment, not the paperwork after it
It’s tempting to file corporation tax under "things my accountant handles once we’re making real money." But for a company running deliberate tests — a paid pilot, a small batch of paying customers, a limited regional launch — tax isn’t a separate track that starts later. It’s a variable that changes the economics of the test itself: how much of each pound of revenue you actually keep, when cash leaves the business, and how much runway a successful pilot buys you before the next investment decision.
This doesn’t mean tax determines whether an idea is good. A test that shows real customer demand and a workable margin is still evidence, whatever the tax treatment. But the net cash a test produces — the number that tells you whether the model can start funding itself — depends partly on rules that many early-stage teams underestimate.
The thresholds, and why "small company" isn’t one fixed rate
Since April 2023, UK corporation tax has worked on three tiers rather than one flat rate. Companies with profits of £50,000 or less pay the small profits rate of 19%. Companies with profits above £250,000 pay the main rate of 25%. In between, profits are taxed at the main rate but reduced by marginal relief, which is designed specifically to smooth the jump rather than let it happen in one step.
That smoothing is the part founders tend to misjudge. It’s not a second flat rate — it’s a taper that produces an effective rate somewhere between 19% and 25% depending on exactly how much profit falls in that middle band, and accountants commonly describe the effective marginal rate on profits in that zone as running higher than either headline figure, sometimes cited around the mid-20s, because of how the relief is calculated. The precise figure for any given company depends on its exact profit, its accounting period, and whether it has associated companies — which is exactly why the article can’t respectably hand you a simplified formula and call it done. HMRC publishes a marginal relief calculator for this reason, and the sensible move for any founder approaching these bands is to check current guidance or run the numbers there rather than eyeball it.
What’s useful for a founder to hold onto isn’t the formula — it’s the shape of the curve. Below £50,000, your rate is stable and easy to plan around. Above £250,000, it’s also stable, just higher. The zone in between is the one where a small change in profit changes your effective rate, and that’s precisely the zone many early-growth companies pass through in their first profitable year.
| Profit band | Rate logic | Cash-flow implication for a test | What to watch for |
|---|---|---|---|
| £0 (loss or breakeven) | No corporation tax due | All revenue from a pilot goes to funding the next one | Losses can usually be carried forward or back against other years’ profits |
| Up to £50,000 | Small profits rate, 19% | Predictable, easy to model into a pilot budget | Threshold is divided by the number of associated companies |
| £50,000–£250,000 | Main rate reduced by marginal relief, tapering between 19% and 25% | Effective rate shifts as profit grows within the band; harder to forecast precisely | Use HMRC’s marginal relief calculator rather than a rough mental estimate |
| Above £250,000 | Main rate, 25% | Stable but higher; larger companies may pay in instalments | "Large" status (roughly £1.5m–£20m profit) brings quarterly instalment payments |
Rate matters less than you’d think — timing matters more
Here’s the part that trips up early-stage teams even more than the percentages: corporation tax isn’t due when you earn the profit. It’s due nine months and one day after the end of your accounting period. That gap is a genuine planning asset if you use it deliberately, and a genuine trap if you don’t.
Imagine a company running a six-month pilot that turns a modest profit. The tax on that profit isn’t a monthly drag on the pilot’s cash flow — it’s a bill that lands well after the accounting period closes, potentially after you’ve already decided whether to extend the pilot, hire someone, or reinvest the proceeds into a second test. That’s good news for short-term cash flow, but it creates a specific risk: teams sometimes read early profitability as "we can afford to reinvest everything," without setting aside the portion that will be owed nine months down the line. When that bill arrives, it can force a hire to be delayed or a second experiment to be underfunded — not because the business model failed, but because the tax timeline wasn’t built into the plan.
The reverse mistake also happens: treating any positive profit as immediately taxable at 19% and holding back too much cash too early, when in fact a loss-making prior year, allowable expenses, or reliefs like R&D tax credits or capital allowances on equipment might reduce what’s actually owed. Either error distorts the real signal you’re trying to read from a test — which is whether the underlying unit economics work, not just whether the top-line number looked healthy for a quarter.
Where the mental model breaks: associated companies and short periods
A second place simple assumptions fail is when a founder runs more than one entity — a holding company, a sister brand, a joint venture with a co-founder’s other business. The £50,000 and £250,000 thresholds aren’t fixed per company; they’re divided by the number of "associated" companies, broadly meaning companies under common control. Two small companies that each individually look like they qualify for the 19% rate can, once linked, both fall into the tapered band instead. The same distortion happens with a first accounting period longer or shorter than twelve months, which proportionately adjusts the thresholds. None of this is exotic — it’s standard for group structures and early-stage spin-outs — but it’s exactly the kind of fact that a quick "we’re small, so we’re fine" assumption will miss.
Building tax into how you document a test
If your team keeps any kind of decision log after a pilot — and it’s worth doing even informally — tax timing deserves its own line, not a footnote. Three questions are enough: What was the accounting period for this test’s profit? When is the resulting tax actually due? And does our reinvestment plan assume gross proceeds or the amount left after that bill? Answering those turns a vague "the pilot made money" into a usable input for the next test.
None of this replaces professional advice — a company’s actual position depends on facts (structure, associated entities, accounting period, applicable reliefs) that no general article can know. But treating corporation tax as a design variable, rather than a distant compliance step, is itself a useful discipline: it’s one more place where careful observation of what actually happens to your cash beats an assumption about what should happen.
