Before I audit anything, I ask one question most marketers skip.
Most audits start with the account. Mine start with a question, and it is the one I see skipped most often: what is this account actually for? Get that wrong and every recommendation after it is wrong too, no matter how clean the slides look.
People open the account, sort by cost, find the expensive keywords, find the campaigns “limited by budget,” and start writing findings. It feels productive. The trouble is that a finding only means something relative to a target. “This keyword costs a lot” is not a finding. “This keyword costs a lot against a target CPL of X, with these conversions” is a finding. Without the target, you are just narrating the interface back to the client.
So before I look at a single chart, I want three things settled. Is this account lead-based, where I am steering toward a target CPL or CPA, or is it revenue-based, where I am steering toward a tROAS? Are there secondary goals or constraints sitting on top of that, things like lead quality, a geo bias, a long sales cycle, brand restrictions? And then the question that sounds soft but is the sharpest of the three: what does “good” look like for this account right now? Not last year, not in the pitch deck. This quarter.
If I cannot answer those, I do not have an audit, I have a guess in a nicer font.
Here is where I part ways with how a lot of audits get delivered. When the goal is unknown, the temptation is to assume something reasonable, get on with the “real” work, and bury the assumption in a footnote on page nine. I do the opposite. If the goal is not confirmed, that is finding number one. I will write something like: goal not confirmed, audit assumed lead-gen with a CPL benchmark inferred from the live bidding strategy, please confirm before acting. It goes at the top, in plain language, because it is the thing that determines whether anything underneath it is trustworthy.
And confirming the goal is not the same as taking the client’s word for it. (This is the part people find uncomfortable, and I think it is exactly where the value is.) I have sat across from accounts where the client tells me, with total confidence, that they are a lead-gen business chasing cheap leads, and then I open the account and value-based bidding is quietly running the show. Those two things cannot both be true. When the stated goal and the live setup disagree, I do not smooth it over with “this might be because.” I put the contradiction on the table and ask which one is real, and the answer reshapes the whole audit. Recommending budget reallocation against the wrong KPI is genuinely worse than recommending nothing, because nothing at least does no harm.
The reason this matters more now, not less, is that the tooling has gotten very good at producing confident output. An AI pass over an account will happily hand you forty findings, ranked, scored, formatted, and every one of them silently assumes a goal nobody confirmed. The scarce thing was never the analysis. It is the willingness to stop, ask the unglamorous question, and refuse to proceed until it is answered. That is the part the operator still owns.
So if you take one thing from this: the first line of a good audit is not a number. It is a target everyone has agreed on. Establish that, use it as the benchmark in every finding after it, and the rest of the work finally has something to be measured against. Skip it, and you have written a very tidy opinion.