The Weekly Audit: Why Nothing Runs on Autopilot
A campaign is not a crockpot. Left alone, an ad account drifts, and it drifts in the platform's favor, not yours. Here is how that drift actually happens, and why we audit every account, every week.
A campaign is not a crockpot. You do not set it and walk away.
Left alone, a Google Ads account drifts. And it does not drift toward your interests. It drifts toward the platform’s, because the platform is the thing making changes while you are not looking. The account you launched in January is not the account you have in March, even if you never touched a single setting yourself.
That is the part most people miss. Doing nothing is not neutral. In a live ad account, doing nothing is a decision to let someone else steer.
A quiet example of drift
Let me show you how it actually happens, because it is almost never dramatic. Nobody breaks the account on purpose. It erodes.
Say a recommendation shows up in the account: expand your reach, it says, turn on this setting, add these broader terms. It looks helpful and it is set to apply on its own if no one intervenes. So it applies. For a few days nothing seems different. Then over the next 2 weeks the targeting quietly widens, and the spend starts flowing toward cheaper, broader, lower-intent traffic, because that traffic is easier for Google to buy.
Here is the cruel part. The conversion count might even tick up, because cheap traffic produces cheap conversions, so the dashboard looks fine or better. The cost-per-real-outcome is getting worse the whole time, but that is not the number staring back at you. A month later the closings are down, the client asks what changed, and the honest answer is that nobody changed anything, which is exactly the problem. The account changed itself while everyone assumed it was holding still.
The kinds of drift
Once you have seen it happen a few times, you start to recognize that drift comes from a handful of directions, and they are all working at once.
The platform changes things under you. Auto-applied recommendations, new defaults on new campaign types, settings that get migrated during some update you never read about. Google ships changes constantly, and its changes are aimed at Google’s goals.
The market moves. Competitors raise their bids, a new entrant floods the auction, seasonality shifts what people search for. Your costs creep up even though your account did nothing, because the auction around it got more expensive.
Things decay. A product feed goes stale. A landing page gets redesigned and a form moves. A tracking tag breaks in a site update and the attribution silently goes dark. None of these announce themselves.
And your own past decisions compound. A budget nudge here, a bid tweak there, each sensible on its own, adding up over months into an account nobody would design on purpose. Entropy is the default state of any account nobody is watching, and it pulls from every one of these directions at the same time.
The weekly audit
So once a week, every account gets a real look. Not a glance at the dashboard between other things. An audit.
Weekly is a deliberate choice, not a convenient one. It is fast enough to catch drift before it compounds into a wasted month, and it is slow enough that the data between audits has time to actually mean something. Look every day and you are reacting to noise, chasing normal daily variance like it is a signal. Look once a quarter and you are paying for a problem that has been quietly running for weeks. Weekly sits in the narrow band where the data is real and the damage is still small.
What the audit is for
I will keep this at the level of the questions, not the checklist, because the checklist is the part that took years to get right and is not the part that helps you.
The audit asks a consistent set of things. Is the spend going where it is supposed to go. Is the signal we optimize by still clean. Did the platform change something under us since last week. Is the money still tracking to real outcomes, or just to activity that looks like progress. Does the attribution still mean what we think it means. Those questions, asked on a schedule against a live account, catch almost everything before it gets expensive. The value is not any single question. It is that they get asked every week, whether or not anything looks wrong.
”Isn’t a weekly audit just busywork?”
It is a fair objection, and I get it. If most weeks you look and nothing is wrong, why keep looking? Is that not just process for the sake of process?
No, and the reason is the whole point. You cannot tell whether nothing is wrong without looking, because the failures are quiet. A broken tag, a widened audience, a slow cost creep, none of them ring a bell. The weeks where the audit finds nothing are not wasted. They are how you earn the confidence that nothing is wrong, which is a real thing you are buying.
Busywork is repeating an action that never changes what you do. An audit is the opposite. It changes what you do the moment it finds drift, and it does nothing loudly the rest of the time, which is exactly what you want from it. The month you skip it to save an hour is the month the quiet problem runs unopposed. Cheap to catch on a Tuesday, expensive to discover at the end of the quarter.
Whose advice is it, anyway
A real part of the weekly audit is deciding which of Google’s own suggestions to ignore, and that is not a small part.
The platform’s recommendations are not neutral. They optimize for Google’s revenue at least as much as for your result. Sometimes those two things point the same direction and the suggestion is genuinely good. Sometimes they point in opposite directions, and the suggestion will spend more of your money for less of your outcome. The only way to know which is which is for a person who is on your side to read each one and make the call.
Auto-apply takes that call away from you and hands it to the machine by default. And the machine’s judgment, however sophisticated, is not the judgment you are paying us for. Half of good account management is not the clever thing you add. It is the confident thing you decline, every week, when the platform politely suggests you widen the net.
Why it beats set-and-forget
The accounts that win are not the ones that were cleverest on launch day. Launch day is the easy part. They are the ones a person actually audits, every week, for as long as they run.
That is the part that is hard to sell and impossible to skip, because it does not photograph well. There is no dramatic before-and-after. There is just an account that stayed pointed at the right things because someone kept pointing it there. The work is not heroic. It is regular, and regular is the whole edge.
For now, at least, no amount of automation replaces someone competent looking at your money on a schedule and asking whether it is still doing its job. The platform will keep making changes while you are not looking. The audit is simply the decision to keep looking.