Dimitri Bukys Start the diagnosis

Cases

Three accounts, and what actually moved them.

Each one has the same shape: what was wrong, what was built, why it worked, and the part that did not go to plan. Ratios only, because the absolute spend and revenue belong to the client rather than to my marketing.

Case 01 · UK ecommerce and B2B · March 2025 to now

Built from zero, and then it outsold their own stock.

A manufacturer selling direct and to trade buyers, with advertising running and no way to tell what any of it earned. Structure, tracking, research and a testing rhythm, plus the autumn it dropped below breakeven and what I did about it.

Read the case →

ROAS 3.7Blended, whole programme
Year twoStill scaling

Case 02 · Bookings, Bali · October 2025 to March 2026

Nobody could tell which ad caused a booking.

A surf camp selling multi-week packages internationally. The ads had not stopped working, the data telling anyone whether they worked had. Measurement rebuilt first, then scaled. Includes the restructure that destabilised the account.

Read the case →

+165%Bookings, over six months
ROAS 5.4Google Ads

Case 03 · Events venue · Three years

When every channel stopped working at once.

Rather than guess at replacements I read their own customer history to find where buyers actually came from, which surfaced a category of marketplace nobody was advertising on. Then the offer was rebuilt around when people decide rather than when they show up.

Read the case →

+250%Company revenue, three years
30%Paid share of company revenue

A fourth document, and it is not a case

The sample diagnosis is the entry product rather than a client story: 402 ads from a real account, replayed day by day against a set of kill rules, showing what the account would have kept. It is the closest thing to watching me work before hiring me.