How to Read a Google Ads Report Without Being Lied To

Most agency reports are not dishonest. They are selective, which does a similar job while staying true.

Here is how to read one in five minutes.

The numbers that look good and mean little

Impressions. How many times your ad was on a screen. It costs nothing to generate and proves nothing. An account spending badly produces impressive impression counts.

Clicks. Better, but still only traffic. You pay for these, so a large number is a cost as much as an achievement.

Click-through rate. The most misleading of the three, because a high rate can be bad news. A vague ad that appeals to everybody gets clicked by everybody, including all the people who were never going to book. A specific ad that filters people out has a lower rate and better leads.

Average position, or wherever it has been renamed to. Being top is not an achievement if top is expensive and the people clicking were wrong.

If a report leads with these and goes quiet afterwards, that is the signal.

The numbers that actually matter

Conversions, and immediately: what is being counted as one?

This single question changes the meaning of the whole report. A conversion might be a submitted form, a phone call lasting more than sixty seconds, a tap on your number, or somebody arriving on the contact page. The last two are not enquiries. If "conversions" includes clicks on a phone number that never connected, the number is traffic wearing a better name.

Cost per conversion. Spend divided by whatever that is. Only useful once you know the definition, and then it is the most useful figure in the report.

Cost per actual job. Almost never in the report, because it needs your side of the data: how many enquiries became work. Multiply cost per conversion by the number of enquiries it takes you to book one, and you have the real figure.

If you know what a customer is worth over time, you can put the two together and see whether the account is ahead. Toolbox Tribe's marketing ROI calculator does that arithmetic without a spreadsheet.

Do the sum yourself once

Five minutes with a pen, and it reframes every report you read afterwards.

Take the spend for the month. Divide by the number of enquiries you can actually remember receiving, not the number in the report. Then multiply by however many enquiries it takes you to book one job. That is what the account is charging you per job.

Compare it to what a job is worth. If a bathroom is worth several thousand and the account is costing a few hundred per booked job, the conversation is about scaling up rather than about whether it works. If those two numbers are close, nothing in the report matters except that.

Most people have never done this sum, which is why arguments about click-through rates go on as long as they do.

The report that is usually missing

The search terms report. The actual phrases people typed before your ad showed.

This is the most revealing document in any account and it is the one that least often appears in a monthly summary. It is where you find out you have been paying for apprenticeship searches, DIY guides and a town three hours away.

Ask for it every month. If the answer is that it is not available, that is the finding. The related job of turning those into negatives is where most accounts stop leaking.

Four questions that settle most of it

1. What counts as a conversion? Everything else depends on the answer.
2. What were the top twenty search terms by spend? Were they relevant?
3. What changed this month and why? A report with no decisions in it describes an account nobody touched.
4. What is the account spending on that is not producing? A good answer names something. "Everything is performing well" is not an answer.

What a good report looks like

Short, and it contains decisions.

It says what was changed and why, what happened to the thing that was changed last month, what is being watched, and what is being proposed next. The numbers are there to support those sentences rather than to fill a page.

A twenty page document of charts with no decisions in it took longer to produce than the work it describes, and that is usually literally true.

Whose account is it

Worth checking if you have never asked. The account should be in your name with your billing on it, and you should have access to it rather than only to the reports.

If leaving means losing the history, the structure and the learning, you are not really looking at your own data. That is a positioning problem, not a reporting one, and it is why my own arrangement is that the account is yours from the start.

The short version

Skip to cost per conversion. Ask what a conversion is. Ask for the search terms.

If those three are available and the numbers hold up, the rest of the report is decoration. If any of the three is missing, that absence is the most informative thing in the document.