← All articles
Automation5 min read

Should You Listen to Your Google Ads Rep?

What to accept from a Google Ads rep call, what to test first, and how to turn each suggestion into a dated experiment with a rollback plan.

The campaign is two weeks old and has six conversions. Your Google rep has a slot on Thursday and a short agenda: switch it to Target CPA, turn on AI Max, and let them "enable a few optimizations" while you are both on the call. They sound sure. They talk to hundreds of accounts. You talk to yours.

I have taken this call more times than I can count, and I have come out of it with good advice exactly often enough that I never decline it outright. What changed my results was a rule I adopted years ago: nothing a rep suggests goes live without a written hypothesis, an end date and a way back.

Short answer: Listen to your Google Ads rep, but implement nothing on the call. Accept the fixes (tracking, policy, broken setup) straight away. Treat every growth suggestion, from Target CPA on a thin campaign to AI Max, as a dated experiment with one success metric, an end date and a written rollback.

The takeaways

  • Google's own Target CPA help page says you can start with no conversion history, and the same page tells you to judge performance over 30 days with at least 30 conversions. Six conversions is a guess.
  • Custom experiments split the budget evenly between control and treatment, per Google's help docs, but only on Search, Display, Video and Hotel campaigns. Shopping and App campaigns are excluded.
  • Write the old settings down before any change. The rollback is the part of the test nobody plans.

When is a Google Ads rep worth listening to?

A Google Ads rep is worth listening to whenever the problem has one correct answer. Broken conversion tracking, a disapproved asset you cannot decode, a policy flag, a feature you cannot find in the interface: reps see these daily and can often escalate them faster than a support ticket. Take that help and say thank you.

Reps are also a decent early-warning system. Whatever they push this quarter tells you where Google is putting its product weight, which is useful context for planning even when you decline the pitch.

The trouble starts with the growth items, because those have no single correct answer. Whether a budget raise, a bid strategy switch or broader matching pays off depends on your margins and your conversion volume. The rep sees neither in full. I wrote about the same split for the Recommendations tab in which Google Ads recommendations to apply: repairs get done, growth gets tested.

Why is Target CPA on a new campaign the classic bad pitch?

Target CPA on a campaign with a handful of conversions is the classic bad pitch because the bidding system has almost nothing to aim with. Google's Target CPA help page says advertisers "can start using Target CPA with no conversion history". The same page recommends evaluating performance over the last 30 days with at least 30 conversions.

Both statements are true. They answer different questions. The first is about whether the setting works technically. The second is about when you can tell whether it worked for you.

On six conversions, a target is a number you made up, and the system will throttle delivery to chase it. If the CPA you type in is too tight, spend stalls and you learn nothing. If it is too loose, you have handed over the budget with a label on it. My default is to keep collecting conversions on a less constrained strategy and revisit the target once the 30-day window holds enough events to mean something. The broader trade-offs live in choosing a Google Ads bid strategy.

How do you turn a rep's suggestion into a test?

You turn a rep's suggestion into a test by writing four things before anything changes: the hypothesis, the one metric that decides it, the end date, and the result that triggers a rollback. "AI Max will lower cost per qualified lead by the end of the month, and we revert if CPA rises past our ceiling" is a test. "Let's see how it goes" is a subscription.

Where you can, use Google's own Experiments tool. A custom experiment runs inside the original campaign and splits the budget evenly between control and treatment, so a bad week in the market hits both arms. When the result is in, you apply the change to the original campaign or end the experiment.

Custom experiments are not available for Shopping or App campaigns. There the fallback is a dated before-and-after: note the change date, compare equal windows, and treat a small difference as noise. For AI Max specifically, the trial setup is in should you use Google AI Max.

What does a good rollback plan look like?

A good rollback plan is a snapshot of the old settings plus a trigger you agreed on before launch. Before any rep-suggested change, I copy the current bid strategy, targets, budgets, match types and asset settings into a note with the date. The change history in Google Ads records what moved, but it will not tell you which state you trusted.

The trigger matters more than the snapshot. Pick the number in advance: a CPA ceiling, a minimum conversion rate on qualified leads, a share of spend going to search terms you would never bid on. When the trigger fires, you revert on the same day without a second meeting.

One rule holds on every call: you make every change yourself. If a rep "pilots" something in your account, nobody owns the end date. That is how auto-apply settings and half-forgotten tests survive for months in inherited accounts.

What should you say on the next rep call?

On the next rep call, ask three questions about each suggestion: which metric it should move, by how much, and over what period. A useful rep will answer with specifics you can write into a test. A vague answer ("it should improve performance overall") tells you the suggestion is on a list, not tailored to your account.

Then ask for the suggestions in an email. That gives you the paper trail, takes the pressure out of the call, and lets you sort fixes from growth items with your own data in front of you.

This is the same discipline I built into Adscalr's automation layer: it recommends by default, logs every rule evaluation, and full-auto only runs once you opt in. If you want to see how recommend-first automation works with your own thresholds, the automation page shows the setup.

This is the thinking behind Adscalr.

See the product →