How Encore Kitchens got to a 28x return they can actually prove

Every month, Encore puts one number in front of its shareholders: the return on its marketing spend. It is a joint venture, so that slide gets read closely. For a long time the number on it was a guess, and most of the room half-knew it.

That is the story. Not the ads. The ads were the straightforward part. The hard part was earning a revenue number Encore could actually stand behind, in a business where a lot of the revenue never shows up.

Encore had grown into the largest operator in the country mostly on outbound sales and word of mouth. Paid search existed on paper, an account set up back in 2021 with a couple of broad campaigns bidding on terms like "dark kitchen London," but it had not been touched since. It still trickled in the odd lead. It was nobody's priority.

The deeper problem was that they were flying blind. They could see what the sales team was doing and almost nothing else. Inbound could not be tracked or valued, so there was no way to know which enquiries were worth chasing, where they came from, or whether marketing could ever pay for itself.

I rebuilt the Google Ads account and set up tracking so that, for the first time, Encore could see which campaigns actually produced leads. Then we wrote specific campaigns by product and region instead of a handful of broad ones, which was where a lot of budget had been quietly leaking. It sounds basic. It changed the conversation immediately. They went from having leads to knowing where leads came from.

As the business grew from one product into several, dark kitchens plus production units and day-rate kitchens, each got its own angle, and the account grew with it.

A lead count is a vanity metric if you cannot tell a tyre-kicker from a real prospect. Encore had no lifecycle stages, so every enquiry looked the same, and most of the inbound was low value.

I defined what a marketing-qualified lead, a sales-qualified lead and a real opportunity looked like for their business, then built the HubSpot workflows behind those definitions. With offline conversions feeding back into Google Ads, the platform started optimising toward opportunities created rather than raw form fills. We reworked the forms too, adding proper conversion points across the site and guardrails like a budget question to wave off the leads that were never going to buy.

Fewer leads came in. The ones that did were worth having.

Once the funnel was defined, I connected revenue back to its source. Encore went from "we have leads, but we don't know where they come from or which campaigns work" to tracking revenue right down to the campaign and the keyword that earned it.

For most companies that is the finish line. For Encore it was where the real problem started.

Encore's customers are food businesses, and food businesses churn. Hard. Deals were signed on 12 and 24 month contracts, but a meaningful share were never going to run their full term. Customers go under. Kitchens empty out halfway through. So the revenue figure in HubSpot was not real money. It was a best case.

This is the part most people get wrong. The standard move is to pull the revenue number out of the CRM, divide by ad spend, and call it your ROAS. For Encore that number would have been confidently, expensively wrong. Knowing what you have actually collected, versus what you have merely been promised, is the difference between spending with confidence and guessing.

Building this properly normally takes a team of developers on a multi-step attribution model, or a mountain of manual work every day. Encore had neither. I used automation to build a custom report that splits paid revenue into two honest buckets.

Collected is money already in the business: contracts from paid campaigns that have actually been paid. No forecasting, just cash received.

Projected is what is realistically still to come: the months left on each live contract, with a churn rate applied per customer type, because a young startup behaves nothing like an established brand, and pretending otherwise ruins the number.

Add the two, divide by spend, and you get a ROAS that survives scrutiny. Encore can see what they have, what they can reasonably expect, and the gap between them.

This is what fixed the board slide.

Across 2025, the churn-adjusted return on paid search was 28x, on roughly £100k of ad spend and around £2.75M in attributed revenue. Paid search is now the largest digital source of new customers Encore has, ahead of organic and direct combined.

The 28x is real, and the point is that Encore can defend it, line by line, to a board that reads the slide closely. It is built on money collected plus revenue projected against real churn, traceable to the campaigns and keywords that produced it, not a figure scraped out of a CRM and dressed up.

Encore's own advice is the same lesson the whole project turned on:

If you sign contracts and some of your customers churn, the revenue number in your CRM is fiction, and you are almost certainly spending against it. The wide-open market gave Encore the opening. The measurement is what let them keep it, and prove it. You cannot optimise a number you cannot trust. Fix the number first. The returns are the easy part.