Spend some time inside established UK ecommerce businesses — the £5m to £50m bracket that sits between startup scrappiness and enterprise process — and a pattern emerges that has little to do with budget size. The businesses growing profitably are not necessarily spending more on growth than the ones standing still. They are spending it in a different order.
This piece is a read on how growth investment is actually being allocated in 2026, what has changed over the last two years, and the operating patterns that separate the compounders from the harvesters. It draws on the engagements, diagnostics and conversations that make up our consulting work rather than on survey data, so treat it as a practitioner's view rather than a census.
1. The demand-buying era is over; the diagnosis era has started
Between 2016 and 2022, the dominant growth model in UK ecommerce was demand buying: paid social and paid search scaled faster than costs rose, and the limiting factor was usually creative volume and appetite for spend. Most growth teams, agency rosters and board packs were built around that model.
Three things broke it. Customer acquisition costs rose structurally as auction density increased and privacy changes degraded targeting efficiency. Cheap capital disappeared, so 'growth at a loss, fixed later' stopped being an acceptable plan. And AI-mediated discovery started moving a meaningful slice of consideration behaviour into surfaces where paid reach does not follow.
The result is a quiet but widespread reallocation. The question boards are asking has shifted from 'how much should we spend on growth?' to 'where is growth actually being constrained?'. That sounds like a nuance. It is a different discipline entirely — and most organisations are not yet set up for it.
The scarce resource in 2026 is not budget. It is conviction about where the budget belongs.
2. Where the money is actually sitting
Across the businesses we assess, growth investment still clusters in the same places it did five years ago — largely because that is where the industry's supply side is built. The mismatch is not that these areas don't matter; it is that they are rarely the binding constraint.
| Where investment concentrates | Where diagnostics most often find the constraint |
|---|---|
| Paid acquisition and creative volume | Conversion experience and proposition clarity |
| New channel expansion (marketplaces, TikTok, retail media) | Repeat rate and second-purchase economics |
| Attribution and reporting tooling | Prioritisation and decision-making capability |
| Content production, increasingly AI-assisted | Non-branded and AI-mediated discovery readiness |
Read the right-hand column carefully. None of it requires a bigger media budget. Most of it is cheaper to fix than the activity currently being funded — which is precisely why it stays unfunded: nobody's retainer depends on recommending it.
3. The four patterns of the businesses that are compounding
The established businesses growing profitably through this period share operating patterns rather than tactics. Four show up consistently.
- They run a small, stable set of commercial numbers — blended contribution, new customer cost, fixed-window repeat rate, non-branded demand — and hold the definitions constant long enough for trends to mean something.
- They name one constraint per quarter and defend it. The plan fits on a page, includes an explicit stop list, and survives contact with a bad channel week.
- They treat agencies and specialists as execution capacity against a brief they own, not as the source of strategy. Briefs sharpen; scope churn falls.
- They fund diagnosis before expansion. A senior, independent read of the whole system precedes any significant new investment, and is repeated when the market or the mix shifts.
1
Named growth constraint per quarter
If the priority cannot be stated without an 'and', the diagnosis is not finished.
4. Where AI is genuinely changing the numbers
Stripping out the noise, AI is changing ecommerce growth economics in three concrete ways. Execution is getting cheaper — research, reporting, creative variation and analysis that once consumed junior capacity now happens in minutes, which raises the value of senior judgement relative to headcount. Discovery is being intermediated — assistants summarise and shortlist on specification, availability, delivery and reputation, which rewards machine-readable commercial facts over content volume. And measurement confidence is falling as journeys fragment across surfaces, which increases the premium on incrementality thinking over attribution precision.
None of these favour doing more. All three favour knowing precisely what to do next — the exact capability the demand-buying era allowed businesses to skip building.
5. What this means if you run a £5m–£50m business
The practical implication is uncomfortable but liberating. The next pound of growth investment is most likely to compound if it is spent on finding and removing a constraint, not on adding demand to a system that has one. That is a cheaper, faster and more defensible decision than the alternatives — but it requires someone senior, independent and accountable to the whole system rather than to a channel.
That is the shape the market is moving towards: fewer, better-briefed specialists; a small set of trusted numbers; one named constraint at a time; and senior commercial judgement as the layer that holds it together. The businesses that organise around this in 2026 will look, in retrospect, like they bought growth at a discount.