How to do a Google Ads audit in the PMax era: a 6-step checklist
A Google Ads audit in 2026 is mostly a Performance Max audit wearing a checklist as a disguise. Structure, Quality Score, and search terms still matter, but PMax now controls the largest slice of most budgets and hides its search terms, its placements, and half its logic behind automated reporting. This is the order I actually work through an account, built to surface where PMax is quietly cannibalizing brand search, where Smart Bidding doesn't have enough conversions to trust, and where budget caps are starving the campaigns that deserve more.
How much account history to pull before judging anything
Pull at least 90 days of history; 12 months if the business has any seasonality, because a snapshot of a slow month reads as a broken account when it's just August. Get the conversion actions list, the tag setup (Google tag, GTM, or an offline import pipeline), and, if Performance Max is running, the asset group breakdown - PMax hides most of its behavior behind that view.
Last thing before you touch the account: ask what the business actually needs the campaign to do. A Target CPA of $60 means nothing on its own - it means something once you know the margin behind the sale.
- Admin or read access to the full account, including MCC hierarchy
- 90 days of history minimum, 12 months for seasonal accounts
- Full conversion actions list and tag/tracking setup
- Performance Max asset group and listing group breakdown
- The real business target: CPA ceiling, ROAS floor, or margin
Map the account and campaign structure
Start with the skeleton before judging any number. List every campaign, its type - Search, Performance Max, Display, Demand Gen, App - and its budget. Look for campaigns competing against each other for the same queries, which happens constantly once an account runs both Search and PMax on overlapping terms; Google's auction doesn't warn anyone when a business is bidding against itself.
Check whether budgets are shared or individual, whether conversion goals are set at account level or per campaign (account-level goal sets are the newer default and change what each campaign is actually optimizing toward), and whether geography and language targeting still match where the business sells. It's common to find a campaign still targeting five countries a client stopped shipping to two years ago.
Flag anything untouched in six months or longer. A campaign nobody has opened since it launched is either fine or silently draining budget, and the only way to know is to look.
Pull Performance Max apart and find where it hides waste
PMax blends Search, Display, YouTube, Gmail, Discover, and Maps into one campaign type. Open the Insights tab and the search terms insights view - a limited proxy for real PMax queries rather than the full report Search campaigns get - and check whether brand terms are showing up inside it.
This is where most PMax waste hides: if the business has strong branded search and hasn't excluded brand from PMax, a large share of its reported conversions are just intercepting people who were going to convert anyway, through organic or branded Search, at a much lower blended cost. I've seen accounts where branded traffic made up a third of what PMax was claiming credit for.
Check asset group count and freshness next. Too many asset groups splits the signal Google needs to learn; too few means no real coverage. Stale audience signals - built once at launch and never revisited - cap performance without anyone noticing, the same way an ignored negative list does. If the client will allow it, a controlled pause of PMax for a short window, compared against the pre-launch baseline, tests how much of its reported result is incremental versus borrowed from other channels.
Check Quality Score and RSA ad strength
Quality Score is a diagnostic, but it's a fast way to find where the account is paying a premium. Pull it at keyword level for the highest-spend ad groups rather than trusting an account-wide average, which smooths over the two or three keywords actually dragging cost up. Expected CTR, ad relevance, and landing page experience are the three components worth separating - a low landing page score points to the page, and no amount of RSA testing fixes that.
For Responsive Search Ads, aim for at least a Good ad strength rating and treat Excellent as a nice-to-have rather than a target worth chasing at the expense of everything else. Check pinning first: an ad group where every headline is pinned to a fixed position has effectively turned an RSA back into a static ad, defeating the combination testing it exists for. Most accounts I open are running eight or nine headlines when twelve to fifteen genuinely distinct angles give Google's algorithm enough material to find a winning combination. Every ad group carrying real spend should have at least two or three RSAs running against each other - a single ad with nothing to compare against can't show what's actually working.
Comb the search terms report and build negatives
With broad match now the platform default recommendation, paired with Smart Bidding, the search terms report is where an account's real intent match - or mismatch - shows up first. Pull the last 30 to 90 days sorted by spend with zero conversions, and read every line above a meaningful spend threshold rather than skimming.
The patterns repeat across almost every account: informational queries starting with 'how to' or 'free,' competitor brand names pulling clicks from people with no intention of switching, and categories so far from the offer that they signal the match type has drifted. Build the negative list from that data, and put it at the shared, account level rather than scattered per campaign - a shared list is the only version that actually gets maintained six months later.
One structural limit worth flagging to the client directly: PMax doesn't expose granular negative keywords the way Search campaigns do, only account-level and brand exclusion lists. That means tightening PMax waste happens through asset groups and audience signals - a different lever than most advertisers are used to pulling.
Verify conversion tracking and enhanced conversions
Every other finding in the audit sits on top of this one. If the primary conversion action isn't the business's actual KPI - a qualified lead, a purchase, a booked call - then every bid strategy, every Quality Score fix, every budget decision downstream is optimizing toward the wrong signal without anyone noticing.
Check for double counting first: a Google tag and a GA4 import both firing on the same event is a common setup mistake, and it quietly inflates every conversion count and every reported ROAS in the account by whatever the overlap rate turns out to be. Then check enhanced conversions: whether the diagnostics status inside the conversion action shows it's actually receiving matched data, a deeper check than the toggle alone provides. Hashed first-party signals like email and phone recover conversions that cookie and app-tracking restrictions would otherwise drop, and an account running without enhanced conversions live in 2026 is bidding on a measurably thinner data set than one that has it configured correctly.
For accounts bidding to Target ROAS, check that conversion values are dynamic and pulled from real order data.
Judge the bid strategy, then read auction insights and pacing
Target CPA and Target ROAS need volume to learn from. A rough rule used across the industry is somewhere around 30 to 50 conversions in the last 30 days per campaign before Smart Bidding has a real chance to stabilize; below that, expect wide day-to-day swings that look like a broken algorithm but are really a data-starvation problem. A campaign sitting on Target CPA with eight conversions a month is just guessing at that point - consolidate it into a bigger campaign, switch to Maximize Conversions without a strict target while it builds history, or accept a manual bid for a while.
Auction insights is worth reading for impression share and overlap rate more than for position, which matters less than it used to. A competitor's impression share climbing steadily over a few months is worth flagging before it ever shows up in CPA - it's a leading indicator of where costs are headed.
Budget pacing closes the loop: check Lost Impression Share (budget) across every campaign, because it tells you directly whether strong campaigns are capped while weaker ones spend freely. A frequent finding is a solid Search campaign losing 20 to 40 percent of its impression share to budget constraints while a mediocre PMax campaign, sharing the same account budget pool, spends without friction.
Habits that undermine an audit
A few habits undermine the accuracy of an otherwise thorough Google Ads audit.
- Judging performance on a 7-day window when Smart Bidding needs 30 days minimum to show its real pattern
- Skipping Performance Max because it's harder to diagnose, when it's exactly where the biggest waste usually sits
- Blaming the algorithm for a dip that's actually seasonal, without checking the same period a year earlier
- Auditing without knowing the business's real margin, so recommendations optimize the wrong number
- Treating Quality Score as the goal instead of a proxy for the CPC and impression share it actually affects
- Recommending a full rebuild instead of a ranked fix list, which stalls the client from acting on anything at all
Realistic ranges and the fix list you hand over
The ranges hold across enough accounts to be a fair expectation of what a specific one will turn up. Hand the client a fix list ordered by effort and impact.
A workable order: fix conversion tracking and enhanced conversions first, since everything downstream depends on it. Add PMax brand exclusion and tighten asset groups second. Clean the search terms report and rebuild the negative list third. Then revisit bid strategy once volume and tracking are both trustworthy, and only after that touch ad copy and Quality Score, which move slower and matter less than the first three.
| Finding | How often it shows up | Typical spend impact |
|---|---|---|
| PMax running with no brand exclusion | Most accounts with strong branded search | 15-35% of PMax conversions overlap brand |
| Search terms report never reviewed | Roughly half of self-managed accounts | 10-25% of spend on zero-conversion queries |
| Target CPA/ROAS on fewer than 30 conversions/month | Very common on newer or narrow campaigns | CPA swings of 40-60% week to week |
| No enhanced conversions configured | Still common outside dedicated PPC teams | Measurable under-reporting of real conversions |
| Strong campaign capped by shared budget | Frequent in accounts split across many campaigns | 20-40% lost impression share on the best performer |
| Single RSA per ad group | Common in older, untouched accounts | Weaker ad strength, less data for Google to learn from |
FAQ
How long does a Google Ads audit take?
A structured pass through this checklist runs two to four hours for a single mid-size account, longer if Performance Max needs a real incrementality check or the account spans multiple MCCs. A surface-level review can be done faster, but it misses most of what actually costs money.
How often should an account get audited?
Quarterly is a reasonable baseline, plus an extra pass after any major change - a new PMax campaign, a bid strategy switch, or a tracking migration. Accounts that just changed something need a closer look sooner than the calendar would otherwise suggest.
Can I trust the conversions Performance Max reports?
Partly. PMax reporting doesn't separate incremental conversions from ones it would have gotten anyway through brand search or Display remarketing. Excluding brand terms and comparing a pre/post launch baseline gets you closer to the real number than the dashboard alone.
I will take a second look at your account
I go through the account and tracking, then hand back a prioritized fix list.
