An inherited account contains a few years of decisions made by people who are no longer available to explain them. The point of the first hour is not to fix anything. It is to find out which numbers can be trusted.
1. Conversion actions – about fifteen minutes
What is the account optimising towards, is it Primary or Secondary, and does the count match what the business’s own database says?
This comes first because every other measurement depends on it. A misfiring tag does not just add noise – it points the bidding system at the wrong target, and then every keyword, audience and creative judgement downstream is made against a distorted scoreboard. We wrote about the five ways this usually breaks.
2. Budget distribution – ten minutes
Sort campaigns by spend, descending. Then ask, for the top five, whether the order matches what the business says its priorities are.
It frequently does not. Budgets accumulate. A test from two years ago is still running at $4,000 a month because nobody turned it off, while the campaign that maps to this year’s strategy is capped at $800 and losing impression share to budget every day.
This step takes ten minutes and is often worth more than everything after it.
3. Brand versus non-brand – five minutes
Split them if they are not already split. Brand search converts at several times the rate of everything else and costs a fraction as much, so an account that blends them reports a number that mostly measures how well the company is known.
Scale the budget on a blended account and the reported ROAS falls, which looks like the new spend failed when in fact the old spend was never doing the work.
4. Search terms – fifteen minutes
Pull the search terms report for the last ninety days and for the ninety days before that, and compare.
We are not looking for the obvious junk – anyone can find that. We are looking for drift: the queries the account matched then and does not now, and the opposite. Broad match and Performance Max both move over time, and a campaign whose match profile has shifted towards research intent will show declining performance with no change anyone made.
5. Structure – ten minutes
Now, and not before, the structure is worth looking at. Specifically: how many campaigns are competing for the same queries, whether the segmentation corresponds to something real (margin, geography, product line) or to something historical, and whether there is enough conversion volume per campaign for the bidding to learn anything.
That last one is the usual finding. Thirty campaigns with four conversions each per month cannot support automated bidding. Six campaigns with twenty can.
6. Creative and landing pages – whatever is left
Ad copy, extensions, and the page the click lands on. This goes last not because it does not matter – it matters enormously – but because it is the part most likely to be judged wrongly if steps one to five have not been done.
What comes out of it
A list, ordered by what would move the number most, with the things we would leave alone marked as such. Not forty flags. Usually five or six items, of which one or two are worth doing this week.
That is what the free audit is, and it is done by a person reading the account rather than a tool generating warnings.