How we took over an inherited Google Ads account for an online education platform with 200+ courses, fixed the conversion data first, then rebuilt the campaign structure around DSA and Performance Max - and ended up 3.8x above the client's ROAS target.
Oney.Studio · Performance MarketingAn online education platform selling professional courses in psychology, nutrition, rehabilitation, yoga, pilates and more - 17 directions, 200+ courses, 62,700+ graduates across 90+ countries. Courses are bought directly on the site, in three languages and three currencies, with a mobile app behind the login.
We inherited the account from a previous agency. Conversion tracking reported numbers that didn't match the CRM, so no bid strategy could be trusted. Coverage was narrow - a handful of manual search campaigns against a catalogue of 200+ course pages - and the client's benchmark was simple: a campaign is only worth running above 1.5x ROAS.
Trustworthy conversions first, full catalogue coverage second, and budget moved to whatever the numbers - not the habit - said was working.
Two conversion actions were firing against the same purchase and reporting several times more revenue than the CRM recorded. We audited the setup with the client's developers, ran verified test purchases end to end, removed the duplication and agreed one source of truth. Only after the numbers matched did we let the smart bidding strategies work from them.
One source of truthInstead of a few broad campaigns covering everything, we split the account by learning direction - psychology, nutrition, rehabilitation, pilates and stretching, military rehabilitation and the rest. Each direction gets its own budget, its own search intent and its own creative, so a strong niche is never subsidising a weak one inside the same campaign.
Structure by directionWith hundreds of course pages and new ones added constantly, manual keyword coverage will always lag behind the site. Dynamic Search Ads let the catalogue itself generate the coverage - surfacing long-tail course queries nobody would have written keywords for, and feeding the account a steady stream of new converting search terms.
Full catalogue coveragePMax started as a controlled test on a single direction. Once it cleared the 1.5x benchmark with room to spare, we scaled the budget and built the rest of the directions out one by one - adding new asset groups weekly, refreshing creative around the strongest performing angle, and letting the campaign type with the best return take the larger share of spend.
Test, prove, scaleSomeone searching for a nutrition qualification and someone searching for a pilates certification are buying two different futures. Messaging was written per direction, around the platform's real differentiators - a European mini MBA diploma, certification backed by a state university, ECTS credits, AI consultation simulators - and rotated based on what converted.
Same account, restructured around trustworthy data. Performance Max now produces more revenue than the entire legacy search setup - at a quarter of the cost per sale.
Conversion value / cost by campaign type, taken from the live Google Ads account. Client benchmark for a campaign to stay running: 1.5x.
With tracking reconciled against the CRM, budget and bid decisions stopped being guesswork. Every campaign is now judged against the same 1.5x line, and the ones that don't hold it get fixed or cut.
DSA surfaces demand for niche courses that no keyword list covered, and every new direction the client launches goes live in the account within days rather than months.
Most campaigns are still budget-limited, meaning the demand is there and the ceiling hasn't been hit - growth from here is a question of spend, not of finding what works.
No dashboard mock-ups and no rounded-up numbers. This is the campaign-level return as Google Ads reports it, in the account currency, sorted by conversion value.

The best-performing campaign returns 5.73x on spend, against a client benchmark of 1.5x.
Cost per sale on that campaign - roughly a quarter of what the inherited search setup was paying.
And the campaigns that don't work are in the same screenshot. That is what the 1.5x line is for.
We set the bar at 1.5x and expected it to be a stretch. The best campaigns are now several times that, and for the first time we're increasing budgets without hesitating.
- Head of Marketing · Online Education Platform · 200+ Courses · NDA
Search and Display campaigns built for efficiency and scale. Proprietary structure methodology that compounds over time.
Facebook and Instagram advertising that converts: audience segmentation, creative testing, and full-funnel alignment.
Precise targeting by job title, industry, and company size. Built for brands where professional context drives purchase decisions.
Performance-focused TikTok campaigns: creative strategy, audience testing, and conversion tracking built for fast-moving feeds.
Sponsored Products, Brands, and DSP campaigns structured to maximise visibility and sales velocity on Amazon's marketplace.
Traditional SEO paired with Answer Engine Optimisation and Generative Engine Optimisation: so your brand ranks everywhere users and AI models look.
End-to-end audit of your conversion funnel: landing pages, checkout flow, and offer structure reviewed and improved to turn more clicks into revenue.
No commitment required. We'll take a look at your current setup and tell you honestly what we see, and what we'd do differently.
No charge. No fluff. A straightforward conversation about your growth.