How to Audit Your PPC Agency's Google Ads Account

Your agency sends a PDF every month. Impressions up. Clicks up. Quality Score mentioned once. Nobody in the report has ever answered the only question you care about: did anyone actually open this account last month?

You do not need to hire a second agency to find out. Every piece of evidence you need is already sitting in the account you pay for, and Google exposes all of it for free.

Are you being managed, or just billed?

You can settle this in about 30 minutes with five checks, every one of them inside your own Google Ads account. None require a tool, a login you do not have, or a second opinion.

Before you start: make sure you are the owner of the account, not a guest in it. If your agency created the account under their manager account and you cannot log in directly, that is the first finding, and it is a serious one. Your ad account, your data, your history.

• The search terms report — what are you actually paying for, term by term?
• The conversion actions — is a hangup at your phone menu being counted as a lead?
• The negative keyword list — when was the last word added to it?
• Performance Max — is it quietly buying your own brand name back?
• The change history log — how many days per month did a human touch this account?

What does your search terms report say you are paying for?

Go to Campaigns, then Search terms under Insights and reports, set the date range to the last 90 days, sort by Cost descending, and read the top 50 rows. That single screen is the most honest report your account produces, because it lists the real queries people typed — not the keywords someone bought.

Google's own documentation is blunt about what to do with it: if a search term is not relevant enough to what you offer, add it as a negative keyword. That is the job. So a spend-sorted list full of terms you would never take a call from is not bad luck — it is work that was not done.

Read the top 50 with a highlighter. Highlight anything you would refuse to quote: DIY searches, job seekers, competitor names you do not want, free-service seekers, the wrong city, the wrong trade. Then total the Cost column on the highlighted rows.

The threshold we use: under 10% of 90-day spend on junk terms is normal drift. Over 25%, with no matching negatives added in that period, means nobody has read this report. Fix it with a real negative keyword strategy, not a one-time cleanup.

Is your conversion action counting phone-menu presses as leads?

Google counts a phone call as a conversion in one of two ways: by using AI to analyze call recordings and identify high-quality leads, or by counting calls that last longer than a minimum duration you set. If that minimum is set to a handful of seconds, a wrong number, a robocall, and a caller who hangs up at your phone menu all bill as leads.

Open Goals, then Conversions, then Summary. For every call conversion action, read the call length setting. For a service business, a real lead almost never resolves in under 60 seconds. If the setting is 15 or 30 seconds — or if call recording is off and the fallback duration was never raised — your lead count is inflated and every optimization built on it is aiming at the wrong target.

Check the counting setting too. Google's guidance is that the 'One' counting option is the right choice when you care whether a lead was generated, and 'Every' is for sales where each one adds value. A lead-gen account set to 'Every' turns one person submitting your form three times into three conversions. Same customer. Triple the score.

Then count the conversion actions marked as primary. If there are six — form fill, call, click-to-call, chat, PDF download, and a page view — you are not measuring leads, you are measuring activity. Long sales cycles need conversion tracking built for the real buying window, not a scoreboard tuned to look busy.

When was your negative keyword list last touched?

Open the negative keyword list, then confirm the last edit date in change history and filter to keyword changes. On any account running broad or phrase match, a list where the newest negative is more than 90 days old is an abandoned list — and on a busy account, more than 30 days is already a smell.

Note the cap while you are there: Google allows a maximum of 1,000 account-level negative keywords per account. If your agency is anywhere near that ceiling, they are working. If the account-level list is empty and campaign lists have not moved since onboarding, the search terms report is being ignored — which you already proved in the previous check.

One caveat before you swing: an account on tight exact-match keywords legitimately needs fewer new negatives than a broad-match account. Judge the list against the match types actually in use, not against a number you read somewhere.

Is Performance Max eating your branded search?

Keeping Performance Max off your own brand name is not a default — it is a thing somebody has to do. Google's instructions are explicit: to exclude branded traffic you first build a brand list, then apply it to the campaign as a brand exclusion. If nobody built the list, nobody excluded anything.

This matters because branded search is the cheapest traffic you own. People who type your company name were already coming. When Performance Max harvests those clicks, your blended cost per conversion drops, the monthly report looks fantastic, and you are paying to buy back demand your other marketing created.

Two checks. First, open the campaign settings and look for a brand list under Brand exclusions — present or absent, no interpretation needed. Second, read the Performance Max search terms and see whether your own company name is in there. Note that Performance Max negative keywords apply to Search and Shopping inventory only, so a negative list is not a full substitute for brand exclusions. If you are still deciding whether the campaign type belongs in your account at all, we wrote up where Performance Max fits for a service business.

What does the change history log reveal about how often anyone logs in?

This is the check nobody wants you to run. Google keeps change history for the past 2 years, and the User column shows the email address of the person who made each change through the Google Ads interface. Your agency cannot delete it, edit it, or explain it away.

Go to Campaigns, then Change history. Set the date range to the last 90 days. Filter the User column to your agency's email addresses. Now count the distinct days on which at least one change was made. That number is your management. Divide by three for a per-month figure.

Read it carefully, because the log tells you who is not a human. Google notes that changes made by automated systems appear under users like 'Google Ads system', and changes made through the Google Ads API — a third-party tool or a script — may show 'Google Ads API' or the tool's name. An account with 200 changes that all trace back to a script is an automated account with an invoice attached to it.

For a $2,000–$10,000/month service-business account, we would expect a human to make changes on roughly 8 to 20 distinct days a quarter. Two days a quarter, or zero human changes with a wall of system entries, is the answer to your question.

Which numbers mean 'this is normal' and which mean 'fire them'?

Start with the only externally measured numbers here. WordStream analyzed 16,446 US-based search advertising campaigns running between April 1, 2024 and March 31, 2025 and reported an average Google Ads click-through rate of 6.66%, an average cost per click of $5.26, an average conversion rate of 7.52%, and an average cost per lead of $70.11. Those are cross-industry averages, not your target — a plumbing emergency click and a personal-injury click do not live in the same universe. Use them as a sanity range, not a goal.

The thresholds below are ours. They are operator judgment, not measured law, and we are labeling them that way on purpose.

That last row is not a stylistic complaint. Google's own documentation says Quality Score is not a key performance indicator and should not be optimized or aggregated with the rest of your data, and that it is not an input in the ad auction. An agency leading a client report with Quality Score is either behind by a decade or counting on you being impressed. Two of these findings is a hard conversation. Four is a replacement.

Also open Auction insights and check your search impression share lost to budget versus lost to rank. Lost to budget means you are leaving winnable volume on the table. Lost to rank means the ads or the landing page are losing the auction. An agency that has never shown you that split is not managing a budget, it is spending one.

What do you do with the evidence?

Send one email with three attachments and a five-business-day deadline: the search terms CSV with the junk rows highlighted and totaled, a screenshot of your change-history day count, and a screenshot of your conversion action settings. Ask for a written response, not a call. Calls do not leave a record.

A real partner will already know the answers and will show you what they did instead. A bad one will explain that the algorithm handles it now, offer a strategy call, or send a longer PDF. If you get a longer PDF, you have your answer — and if you are stuck in a 12-month contract while it happens, that contract was written to protect them, not you. Everything we run is month-to-month for exactly this reason.

Whatever you decide, do not hand over the login. You own the account, the conversion data, and the two years of change history that let you run this audit again next quarter. That history is worth more than any agency relationship, including ours. If you want a second set of eyes on the account before you fire anyone, our paid ads management starts with exactly these checks, in writing, and you can also see how attribution should actually be wired before you spend another dollar. Get my free audit.