How to tell if your Google Ads agency is doing the work

Five checks you can run yourself in your own account, the change history that cannot be dressed up, and the four reasons the problem is sometimes not the agency.

Paying a Google Ads agency puts a client in an awkward position: the person judging the work is the same person receiving the report that describes it. And reports are built to look good.

This is not about distrust. It is about five things you can check yourself, in your own account, without being an expert and without asking anyone’s permission. If they come out well, you have a good agency and now you know. If they come out badly, you have a concrete conversation instead of a feeling.

The monthly report is not evidence

A report can be entirely true and still say nothing. Impressions, clicks, CTR and “cost per conversion” are real metrics; the problem is that none of them answers the only question that matters: did this bring customers, and did they cost less than they are worth?

Worse: all four can improve without the business improving. CTR rises if you target people already searching for you by name. Cost per conversion falls if you count as a conversion something that happens far more often than a sale.

Five checks you can run yourself

CheckWhat you are looking forBad sign
Search termsWhich real searches the money went toTerms unrelated to what you sell, month after month
Negative keywordsThat the list grows over timeEmpty list, or unchanged since launch
Conversion actionsWhat counts as a conversion and which one bidding optimisesEverything counts the same: a page view equals a sale
Budget vs structureWhether what works has room to scaleCampaigns limited by budget while others underspend
AdsWhen they were written and how many variants existThe same copy for a year

The first two go together and reveal the most. The search terms report shows what people actually typed before your money was spent, which is not the same as the keywords somebody chose. If searches unrelated to what you sell show up there, and show up again the following month, nobody is reading that report.

The third moves the most money. If everything counts as a conversion equally, the bidding system is optimising toward the most frequent event, which is almost never the most valuable one. The exact question: which conversion actions are included in the column automated bidding uses?

The fourth is quick to see: if the campaigns bringing the good customers are capped by budget while others do not spend what they are allocated, there is a reallocation nobody has done.

The one check that settles it

Google Ads keeps a change history: what was modified, when, and by which user. It is the only part of the account that cannot be dressed up in a report, because nobody writes it — the platform does.

Open it and look for three things:

  1. Is there regular activity, or a spike the week the account was built and silence afterwards?
  2. Do the changes resemble what the report says? If the report talks about “continuous creative optimisation” and the history records no new ad in six months, there is a gap worth asking about.
  3. Are the changes suspiciously clustered? A burst of tweaks three days before each monthly meeting says something about how the account gets prioritised.

An honest caveat: lots of activity is not the same as good work. With automated bidding, constant touching hurts. What you are looking for is not change volume but correspondence between what was done and what you were told.

Questions a good agency answers calmly

  • Which conversion action is the account optimising for right now, and why that one.
  • What were the three searches that consumed the most budget last month.
  • Which campaign would you scale if I raised budget 30%, and which would you cut if I lowered it.
  • How much of what the platform reports as a conversion turned into a sale.
  • What did you change last month that did not work.

The last one tells you the most. An account genuinely being worked accumulates experiments that failed, and whoever works it remembers them. “Everything has been going well” sustained for six straight months is less reassuring than it sounds.

When the problem is not the agency

Worth saying because it happens often: there are well-run accounts performing badly for reasons that sit outside the account.

  • The landing page. If traffic arrives somewhere that does not convert, no amount of bid optimisation fixes it.
  • Response speed. A lead answered 48 hours later has already bought elsewhere. That does not show up in Google Ads and gets paid for in Google Ads.
  • Price or offer. No campaign compensates for being out of the market.
  • Budget. Below a certain threshold the account never leaves the learning phase and everything looks erratic.

Before switching provider it is worth ruling these four out. Changing agency with a broken landing page produces exactly the same result, three months later.

We wrote separately about the budget question and how to frame it: how much to invest in Google Ads in Chile.

FAQ

  • I do not have access to the account. Is that normal? No. You should have at least read access to your own advertising account, and the account should be registered under your company, not the agency’s. If it is not, that is the first thing to fix — before any performance discussion.
  • The report shows lots of conversions. Is that good? It depends entirely on what counts as a conversion. If a page view, a click on a phone number and an actual sale all count as one, the total says nothing. Ask which conversion actions are included in the optimisation column.
  • How often should campaigns be touched? There is no correct frequency, and touching things constantly can hurt automated bidding. What matters is that changes are deliberate and explainable — not that there are many of them. A long silence with declining results is the actual warning sign.
  • Should I change agency if I find problems? Not necessarily. Most of these findings are fixable and worth raising first. How the conversation goes tells you more than the findings themselves.