Eight years running paid media that has to pay for itself.

Test-ride bookings had to be driven across three markets with different dealer networks, different purchase behaviour and different search intent. Centrally translated copy does not convert in DE or FR — it converts in EN and gets translated into something nobody searches for.

• —Built and ran Search, Display and Performance Max per market rather than one campaign with language variants
• —Wrote market-specific ad copy against local search intent, not translations of the English original
• —Set geo-targeting to dealer catchment areas so spend matched where a booking could actually be fulfilled
• —Aligned landing pages per market and allocated budget on marginal CPA rather than splitting it evenly

Sustained 3% conversion rate on test-ride bookings across all three markets.

Conversion tracking ran under EU Consent Mode, which changes what signal the bid strategies actually receive. Bidding on modelled conversions in DE and FR is a different problem to bidding on observed conversions in the US, and the campaign structure had to account for it.

A publisher acquisition account carrying both B2C subscription and B2B objectives across search, social and programmatic. Each channel resolved on a different attribution window, so fixed monthly channel budgets consistently underfunded whichever channel was currently working.

• —Ran acquisition across Google Ads, Meta and The Trade Desk against separate B2C and B2B objectives
• —Reallocated budget on marginal CPA continuously rather than on a monthly planning cycle
• —Activated programmatic prospecting and retargeting — audience segment selection, frequency capping, creative rotation, pacing against CPM and CPA targets

B2C conversion rate held at 5–10% and B2B at 3–5%, sustained across the engagement.

The programmatic work ran inside an account architecture set at platform level. Campaign activation and optimisation were mine; seat structure, deal curation and supply path configuration were not.

Paid lead capture was running against a page that had never been built to convert, holding at 1.5%. There was no dedicated landing page, no design support and no existing marketing automation instance to build one in.

• —Stood up the HubSpot instance from zero — forms, lifecycle stages, lead capture, CRM workflows
• —Built the landing page from scratch, with no design or development support
• —Ran structured A/B tests on conversion elements and iterated the UX off the test data rather than off opinion

Conversion rate from 1.5% to 3.5% — a 133% lift.

Delivered inside a two-person marketing team, alongside full ownership of Google Ads, Meta and email for the rest of the client book.

Enterprise MDM is a long, committee-driven sale. Volume lead generation produces contact records, not pipeline. The client needed first conversations with specific named accounts, not more MQLs.

• —Built LinkedIn ABM targeting against a defined named-account list via Sales Navigator
• —Ran the paid campaigns and paired them with Lemlist and Apollo outreach sequences against the same list, so paid and outbound hit the same accounts rather than two disconnected audiences

Ten qualified first conversations in 45 days. The client renewed the brief twice.

A deliberately small budget. The constraint was account list quality and message fit, not spend — which is the correct constraint for ABM and the reason the result is transferable.

In large-catalogue retail, campaign performance is bounded by product feed quality long before it is bounded by bid strategy. Optimising bids against an incomplete feed moves nothing.

• —Built and scaled Performance Max and Shopping campaigns through Google Merchant Center
• —Managed feed coverage, asset group structure and inventory-level performance

Included here for scope rather than outcome — the reporting for this account sat with the client and I do not hold a figure I can stand behind publicly.