Most accounts we inherit are spending correctly in aggregate and incorrectly everywhere else. The total looks defensible. Underneath it, branded search is absorbing budget that should be buying new customers, upper-funnel video is being judged on last-click, and nobody has separated the campaigns that close sales from the campaigns that decorate the report.
What we change first
The scoreboard. Before we touch a bid, we split spend into three buckets: branded search that would largely have converted anyway, non-branded acquisition that genuinely wins customers, and defence against competitors bidding on your terms. Those three deserve different budgets, different targets and different patience.
Then we rebuild campaign structure around how people actually search, not around how your catalogue is organised. Match types get separated so harvesting works. Negatives get maintained rather than dumped in once at launch. Placement and dayparting modifiers get set from your own data instead of category defaults.
How budget gets set
Not by a target ACoS handed down at the start of the quarter. We work backwards from unit economics: what a customer is worth over their life with you, what you can afford to pay to acquire one, and where the margin floor sits once fees, returns and promotions are counted. That gives a CAC ceiling per product line. Bids move inside it.
Where a brand sells consumables, the calculation changes again — a first order is a down payment on a subscription, and paying more than a one-time buyer would justify is usually the right call. We make that trade deliberately and show you the arithmetic.
What you get every week
A written read. What we changed, what it cost, what it returned, what we are doing next, and which number to watch. Written to be forwarded to a CFO without translation.