Book a free review →

Revenue up 160%. ROAS up 192%. We changed the goal first.

When the right metric is in place, everything else has something to optimise toward.

Client Admiral Business, UK
Results after First 6 months vs previous year
160% ↑ Revenue
192% ↑ ROAS
135% ↑ Sales

The first thing we changed wasn't the campaigns. It was the goal.

Admiral Business sells insurance policies to SMEs across the UK. Their products differ substantially in value — a courier policy and an employer's liability policy might look identical in a conversion count but deliver very different revenue. That value data was already being imported into Google. Nobody was using it.

We came on board at the same time as a completely new marketing team. No attachment to how things had been done before, no defensiveness about inherited decisions. From the first conversation, the team wanted to understand the thinking behind every change — not just improve the metrics.

The goal at the time was CPA — which treats all policy sales as equally valuable, regardless of whether the policy is worth £100 or £1,000. Once you see the maths, CPA stops making sense immediately. The new team didn't just accept that — they got it straight away and pushed us to go further.

The target should be ROAS, not CPA. That was agreed before we touched anything in the account.

"The first thing we changed wasn't the campaigns. It was the goal. Once that was right, everything else had something to optimise toward."

Seven issues. None obvious from the surface.

Conversions were tracking. Some terms appeared to be performing. But the numbers underneath told a different story.

01
Goal not matched to account structure Policies varied too much in value to optimise by CPA. A courier policy and an employer's liability policy each counted as one conversion — but the revenue difference was significant. Value data was sitting in the account unused.
02
67% of search terms triggering multiple keywords Campaigns were structured by match type rather than product — a setup from an earlier era of Google Ads. With broadened match types, it created data overlap: the same searches firing across multiple campaigns simultaneously, making performance decisions harder and reporting unreliable.
03
Broad match absorbing budget at poor returns Thousands spent through broad match at a ROAS around half the target, versus phrase and exact match, which came in only slightly under target. Broad match was pulling in searches with no connection to business insurance.
04
Smart Bidding paying the wrong amounts CPCs far exceeding what the terms were worth. High-ROAS keywords were being underbid and missing traffic. The automation was spending more where it should have spent less, and less where it should have spent more.
05
Tens of thousands of pounds spent on irrelevant searches, returning almost nothing Searches with clearly irrelevant intent accumulating spend over time, with almost no return.
06
High-cancellation terms inflating reported results Certain short-term policy searches generated apparent conversions — but the policies weren't being kept. Reported revenue didn't reflect what was actually being retained.
07
PMax too broadly structured to perform All asset groups in a single campaign with irrelevant targeting. Ad headlines unpinned, allowing Google to assemble any combination — repetitive headlines, no CTA hierarchy, copy that didn't filter out low-intent clicks.

Seven changes. In the right order.

The new team understood that some changes would cause short-term dips before delivering long-term gains. When we explained that adding negative keywords would temporarily reduce reported ROAS before it improved, they trusted the reasoning and backed the decision without hesitation. That kind of trust is rarer than it sounds.

FIRST
Changed the goal Switched from CPA to target ROAS so the account would spend more to acquire high-value policies and less on cheaper ones. This came before anything structural — and it only worked because Admiral Business was willing to question the original brief.
STRUCTURE
Rebuilt campaigns by product type Campaigns restructured around insurance product rather than match type. Keyword overlap removed. Data consolidated so performance could actually be read and acted on.
PMAX
Segmented PMax properly and retested it Split by product type, with asset groups organised around relevant themes and audiences. Retested with a setup that matched how business insurance buyers actually search.
BIDDING
Replaced Smart Bidding with manual CPC Conservative opening bids based on proven and expected performance. Higher where terms had demonstrated returns, lower where they hadn't. No more inflated CPCs on searches that weren't converting.
NEGATIVES
Paused high-cancellation terms Short-term policy searches added as negatives. Reported revenue now reflected policies that were actually being kept.
KEYWORDS
Negative keywords built properly Extensive list built from search term analysis and proactive research — blocking categories of irrelevant traffic, not just individual terms.
ADS
Ads rewritten with proper structure Pinned elements throughout. Headline 1 keyword-relevant, Headline 2 a clear CTA that deters non-buyers, Headline 3 a trust signal. Two or more ads per ad group testing different angles — price, ease, service.

High season vs high season. Same window, twelve months apart.

We started working with Admiral Business at the beginning of high season. Rather than compare to the months before — which would flatter any seasonal business — we measured what mattered: performance against the same period the previous year.

160% Revenue increase
192% ROAS improvement
135% Sales increase
“

Admiral Business came in as a completely new marketing function — no legacy positions to defend, no reluctance to question what had come before. When we told them that pausing high-cancellation terms would reduce reported conversions before improving real ones, they didn't push back — they asked what else we should be looking at. That's not a common response. The pace and scale of what was achieved in six months is directly connected to that working relationship.

Melissa Ros · Founder
More Results
Audit · Exterior Cleaning · UK 57% 178% more conversions. 57% lower CPA.
Read more →
Consultation · Mortgage Provider · Ireland 4 One session. Four problems found — including one that had stopped the campaign spending.
Read more →

Want an honest look at your account?

20 minutes. Enough to spot the major issues, understand what you need, and give you a realistic idea of what we'd do and what it would cost.

Get a free review →