These are the numbers.
Here's what was actually behind them.

What was broken

The ads were fine. The problem was downstream. I tracked drop-off at every stage of the funnel and found the product page was where people were leaving. Clicks were coming in, intent was there, and then nothing. The page wasn't converting.

What I did

Full audit of the product page, competitive research on what was working in the space, then rebuilt the landing page from scratch. Wrote the video scripts, set up A/B tests across placements to find what actually held attention, and kept optimizing until the data gave a clear answer.

What changed

Conversion rate doubled. Once the page held, the paid media performance had somewhere to go. ROAS moved from 2x to 6x over the next six months.

What was broken

When I went deep into the account, the search terms alone were eating $70,000 in 4 months on queries that had nothing to do with the business. The campaign structure was overbuilt, brand campaigns were cannibalising non-brand, and YouTube was running an action campaign with the wrong audience entirely. The bidding was trying to solve a targeting problem.

What I did

Used scripts to do a full search term audit across every campaign. Stripped out everything that wasn't working, rebuilt the structure, and kept only the keywords that had a track record. For YouTube, I went back to the audience persona and built three custom audience segments from scratch, using keyword signals, relevant websites, and app behaviour. Clean signals in, clean results out.

What changed

CAC dropped 23% within the first month. Average order value also went up, because we were now reaching people who actually wanted to buy.

What was broken

PMax was one big campaign with everything lumped together. No product differentiation, no signal about lead quality. Google was optimising for volume, not value. On the social side, the account was heavily reliant on second-party audiences that were producing low-quality applicants and were on their way to becoming obsolete. The retargeting pool was thin. There was no real creative testing happening anywhere.

What I did

Restructured PMax by product category and lead quality signals. Integrated CAPI so the system learned to prioritise high-quality applicants from the start, not just anyone who submitted a form. On social, built a proper creative and audience testing framework, expanded into first-party and third-party audiences, rebuilt the retargeting pool, and pulled back reliance on second-party data that wasn't producing.

What changed

322% more leads year over year. Marginal CPA dropped 32%. The same paid media budget was now doing three times the work. This growth also became the primary engine behind the company's 7x expansion over the following 18 months.

What was broken

Two problems running in parallel: a PMax shopping campaign with a weak feed and no real structure, and a generic search campaign that wasn't built to capture high-intent buyers. The feed had poor titles, missing attributes, and nothing to help Google understand which products were worth fighting for. The landing pages weren't doing their job either.

What I did

Rebuilt the feed structure and titles, added proper attributes and custom labels. For search, built SKAGs for the highest-value generic terms with ad copy and landing pages written specifically for each intent. Layered in Google custom audience segments and ran a script-based search term audit to find every pocket of waste and every pocket of opportunity.

What changed

Closing the gaps in both campaigns added $1.2M in recoverable revenue over the year. Not from increasing spend. From stopping the leaks.

What was broken

The account had no systematic way to learn. Creatives ran until fatigue, audiences were never challenged, and landing page variants were built on gut feeling rather than data. There was no decision framework for what gets scaled and what gets cut. Every optimisation was reactive.

What I did

Built a structured A/B testing framework across creative formats, audience segments, and landing page variants running simultaneously. Every test had a clear hypothesis, a defined measurement window, and a decision rule. Winners got scaled. Losers got replaced with the next hypothesis. The learning compounded over time.

What changed

45% ROI improvement across the tested channels. Not from one breakthrough, but from dozens of small, deliberate wins that stacked. That's what a proper testing programme looks like.

What was behind it

This result didn't come from one thing. It came from everything working together. A PMax structure that knew the difference between product categories. A social framework finding real buyers instead of cheap leads. CAPI telling the algorithm what "qualified" actually meant. A retargeting pool that was actually populated. And a testing culture that meant the system kept getting smarter every week.

What compounded

The 322% lead growth was the foundation. But what made it 7x growth was that lead quality improved alongside volume. The business wasn't just getting more leads, it was closing more of them. When acquisition infrastructure works properly, growth compounds. That's what happened here.