We have run hundreds of Google Ads audits. The framework below is the one we still use, updated for how accounts work in 2026. The original version of this guide was written when expanded text ads, broad match modifier and manual CPC were the norm. None of those are the norm now. Responsive search ads are the only search ad format, Smart Bidding runs most budgets, broad match is the default suggestion in new campaigns, Performance Max sits in almost every ecommerce account, and conversion data flows through GA4 and enhanced conversions rather than a single pixel.
The structure is unchanged because it still works. Four sections, each with its own checks:
- Account structure. How campaigns and ad groups are organised, including search, Shopping, Performance Max and lead generation accounts.
- Campaign settings. The defaults that quietly waste money, plus bidding and budget checks.
- Ad groups. Responsive search ads, keywords and match types, search terms, assets and audiences.
- Everything else. Experiments, broken landing pages, conversion tracking quality, Performance Max and landing pages.
Work through them in order. Structure problems explain most of what you find later, and you will save time if you understand the account’s logic before judging its ad groups. A reusable checklist is at the end.
Section 1: account structure
Good structure is planned, not improvised. The best accounts mirror the website: a campaign per service or product category, ad groups per sub-topic, each ad group pointing at a page built for that topic. Start the audit by opening the site’s navigation next to the campaign list and checking whether one reflects the other.
Simple and extended structures
A simple structure suits smaller accounts. Campaign: Levi’s jeans. Ad groups: men’s, women’s, kids. The women’s ad group holds keywords like “women’s Levi jeans”, an ad about women’s Levi jeans and a landing page about women’s Levi jeans.
An extended structure segments further: a campaign for women’s skinny Levi’s with ad groups for general, dark wash and leggings. That works for a broad catalogue, but every extra segment is more to manage and many will never get enough traffic to optimise. Smart Bidding also needs conversion volume to learn, so over-segmentation hurts more than it used to. Look for the balance between relevance and manageability, and flag campaigns that are too thin to ever exit learning.
Things aren’t always as they seem
An odd-looking structure may be the result of years of testing. Before recommending a rebuild, open Change History and look at removed and paused campaigns. If a structure was tried and discarded, say so in the audit rather than proposing it again. Budget constraints also shape structure. The ideal layout is not always the affordable one.
Network and geo separation
Search campaigns should not be running on the Display Network. Check each search campaign’s settings for Display Network and Search Partners. Google still pre-selects both for new campaigns. Display expansion in a search campaign is almost always wasted spend; turn it off unless there is a specific reason. Search partners can stay if the segment report shows they convert at a comparable cost.
Review geo targeting campaign by campaign. For local businesses, check whether the account uses a double campaign structure: one campaign with tight geo targeting and broader keywords, another with broad geo and tight location-modified keywords such as “plumber Parramatta”. It gives the widest coverage and is a common audit recommendation.
Ecommerce structure
Ecommerce accounts carry hundreds or thousands of products, so the structure needs to scale. The layout we recommend:
- Search campaigns for categories. Skinny jeans, loose jeans, straight jeans, targeted with regular keywords.
- Dynamic Search Ads for product pages. Individual products generate too many long-tail queries to cover with keywords. A DSA campaign crawls product pages and matches the query to the page. Open the dynamic ad targets and confirm they cover product URLs only, not the blog or the contact page.
- Shopping, through Standard Shopping or Performance Max. Any retailer with a product feed should be running one of these. Which one, and how they interact, is a structural decision in its own right.
If both Standard Shopping and Performance Max are live, understand the precedence. Performance Max takes priority over Standard Shopping for the same products in the same account, so a Standard Shopping campaign running alongside a PMax campaign with the full catalogue will be starved. The audit should confirm that overlap is deliberate: for example, PMax for the catalogue and Standard Shopping restricted to a set of products the client wants to control manually.
Where Standard Shopping is used, the three-tier structure still applies: a low-priority catch-all with low bids, product-category campaigns at medium priority and medium bids, and a high-priority campaign for best sellers or high-margin items with the highest bids. Check that priorities and bids line up with the hierarchy.
Custom labels and brand separation
Merchant Center is largely out of scope for a time-boxed audit, but custom labels are a quick check. Open a Shopping or PMax campaign’s listing groups and look for a label like “best seller” or “high margin”. If none exist, the client cannot bid or budget differently for the products that matter most.
Also check that brand traffic is separated from generic. A branded search campaign with its own budget, and brand exclusions applied to Performance Max, stops the automated campaigns from taking credit for customers who were already looking for the client.
Section 2: campaign settings
Settings are where defaults cost money. Most are set once and never revisited.
Location options
Open a campaign’s settings and expand Locations, then Location options. The default is “Presence or interest”: people in, regularly in, or who have shown interest in the targeted location. That means a campaign targeting Sydney can show ads to someone in Manila who searched for something Sydney-related. For almost every lead generation and local account, the right setting is “Presence” only.
If you find the default, quantify the damage. In Reports, open the predefined User Locations report, set the range to the last six months, and export it. Filter out the target country and sum the clicks and cost that remain. That number goes in the audit as a saving.
Ad schedule and device adjustments
Check whether schedules and device bid adjustments exist, then check whether they still do anything. With manual CPC or enhanced CPC, a schedule that runs ads only during business hours and a -30% mobile adjustment are meaningful. With Target CPA, Target ROAS or Maximise Conversions, Google ignores bid adjustments on schedule and device, with the exception of a -100% device adjustment. An account full of adjustments under Smart Bidding was probably migrated without review. Note it, because it tells you how closely the account is managed.
Bidding strategy and budgets
Record the bid strategy for every campaign. Then ask three questions.
- Does the campaign have enough conversions for the strategy? A Target CPA campaign with three conversions a month will never settle. Consolidating campaigns or using a portfolio strategy is usually the answer.
- Is the target realistic? A Target CPA set well below the historical CPA throttles volume. Compare the target to the trailing 90-day actual.
- Is the budget limiting? Check the Budget column for “Limited by budget” and the search lost impression share (budget) column. If a profitable campaign is capped, that is a recommendation with a number attached. Our regression method estimates what extra spend is likely to return.
Also check the conversion goals each campaign is optimising to. Campaigns can use the account default goals or their own. A lead generation campaign optimising to page views because someone left the default is a conversion tracking problem dressed as a bidding problem.
Auto-apply recommendations
Open Recommendations and check the auto-apply settings. Google can add keywords, change match types, raise budgets and adjust bidding without anyone approving it. Some auto-applied recommendations are harmless; some, such as adding broad match keywords or expanding final URLs, change the account’s strategy. Most managed accounts should have these off. If they are on, cross-reference Change History to see what has been applied automatically in the last 90 days.
Section 3: ad groups
The ad group holds the ads, keywords, assets and audience settings. This is the longest part of the audit, so sample. Review the five highest-spending ad groups properly and spot-check the rest.
Responsive search ads
Expanded text ads cannot be created or edited anymore. Legacy ETAs still serving is not a red flag on its own, but an ad group with no responsive search ad is. Check each sampled ad group for:
- At least one RSA, with enough distinct headlines to give Google combinations to work with. Fifteen headlines that say the same thing in different orders do not count.
- Pinning. Pinning every headline to a position turns the RSA back into a static ad and limits learning. Pinning one headline for compliance or a brand name is fine.
- Ad strength. Poor or Average on the top spenders is worth noting, though ad strength is a guide to asset variety, not a performance metric.
- Keyword relevance. Do the headlines contain the ad group’s core terms? Is there a clear call to action? Is dynamic keyword insertion used where it helps?
Keywords and match types
Ad groups with 50 keywords are still common and still too many. An ad group is one concept. Five to fifteen keywords usually covers it; if there are more, they can be split into tighter groups.
Match types are now broad, phrase and exact. Broad match modifier is gone, and phrase match absorbed most of its behaviour. Broad match has changed the most: paired with Smart Bidding and good conversion data it can work well, and Google now recommends it by default in new campaigns. Broad match on manual bidding, or in an account with weak conversion tracking, is a red flag. The audit question is whether match types fit the bidding strategy and the budget, not whether broad match exists.
Search terms and negatives
The search terms report no longer shows every query; low-volume terms are hidden for privacy. That makes the terms you can see more important, not less. Pull the last 90 days, sort by cost, and look for irrelevant queries that have been allowed to keep spending. Then check the negative keyword lists: are there shared lists at account level, are they applied to every campaign, and when were they last updated? An account with no negatives added in six months is not being managed.
Assets, formerly extensions
Ad extensions were renamed assets. Every campaign should have sitelinks and callouts at a minimum, and they should be relevant to that campaign rather than copied from the account level. Depending on the account, also look for structured snippets, call assets, lead form assets, price assets, image assets and the business name and logo. Check the automatically created assets setting too. Google can generate sitelinks and headlines from the landing page; some clients are fine with that, some, particularly in regulated industries, are not.
Audiences
Open the Audiences tab for each sampled campaign. Under Smart Bidding, audiences in observation mode feed the algorithm even without bid adjustments, so the question is whether the right segments are attached at all. Look for website visitor lists, converters, cart abandoners for ecommerce, engagement-based lists for lead generation, in-market segments and customer match lists built from the client’s CRM or email list. Customer match is underused and, where the client has a list of past clients, one of the easiest improvements to recommend.
Also check demographic exclusions and whether any adjustments or exclusions contradict performance. Bidding up on an age band that never converts is a finding.
Audience strategy matters most in high-CPC categories. A law firm account, where a single click can cost more than a day’s budget in another industry, lives or dies on who it excludes. We cover the specifics of that in our Google Ads for law firms service.
Section 4: everything else
With structure and ad groups covered, five remaining areas separate a well-run account from an adequate one.
Evidence of experimentation
Open Experiments in the left menu. Drafts are gone; experiments now run directly from a campaign and can test bidding, ad copy, landing pages and, through custom experiments, almost any setting. If no experiment has been run in the last year, the account is being optimised on opinion. Recommend a testing programme and point to the kind of hypotheses that pay off; our experiments case study is a good starting list.
404 checking
Sending paid traffic to a broken page is the purest waste in the account. In a large account you cannot click through every final URL. Open Tools, then Scripts, and look for a link checker. If there is nothing at account or manager level, recommend one. The scripts we use most are listed separately.
Conversion tracking quality
This is the most important check in the audit, and in 2026 it is harder to get right than it was. Everything above depends on it: Smart Bidding, broad match and Performance Max all optimise toward whatever the conversion actions say. Open Goals, then Conversions, and work through the following.
- One source per action. Each conversion should be tracked once, either by the Google Ads tag or by importing a GA4 key event. If “Form submission” exists as both a Google Ads tag conversion and a GA4 import, and both are set to primary, every lead counts twice and the bid strategy is being lied to. Pick one source per action. We lean toward GA4 imports where the GA4 setup is clean, because it keeps one measurement layer for the whole site.
- Primary and secondary. Only actions that represent real business value should be primary. Page views, scroll depth and clicks to the contact page belong in secondary, where they are reported but do not drive bidding.
- Status and recency. The Status column shows whether each action has recorded a conversion recently. “No recent conversions” on a primary action is either a broken tag or a dead goal. Check the Diagnostics tab for tag errors.
- Enhanced conversions. Enhanced conversions for web send hashed first-party data such as email address with the conversion, which recovers conversions that browser restrictions would otherwise lose. Enhanced conversions for leads lets a lead generation client import offline outcomes, such as a qualified lead or a signed client, back into Google Ads. Both should be on, and the diagnostics tab reports match rates. For a lead generation account, offline conversion import is the single biggest tracking upgrade available: it lets the account optimise to clients, not form fills.
- Consent mode. If the site runs a cookie banner, check that consent mode is implemented so Google can model conversions from users who decline. Without it, declined users disappear from the data entirely. This is mandatory for advertisers reaching the EEA and UK, and increasingly relevant in Australia as privacy law tightens.
- Attribution. Data-driven attribution is the default and the older rule-based models apart from last click have been retired. Confirm the model is consistent across actions, and that nobody is comparing a last-click report against a data-driven one and calling the difference a performance change.
- Auto-tagging and GA4 linking. Confirm auto-tagging is on in account settings and the Google Ads account is linked to the GA4 property. Without both, you lose click-level reporting in GA4 and cannot import key events.
- Conversion values. For ecommerce, confirm transaction value is passing through. For lead generation, consider assigning different values to different actions so Target ROAS or Maximise Conversion Value has something to optimise against.
Auditing Performance Max
Performance Max needs its own pass because it hides most of its workings. Check:
- Brand exclusions. Without them, PMax will happily bid on the client’s own brand and report a wonderful ROAS. Confirm a brand list is excluded, and that there is a separate branded search campaign.
- Asset groups. One asset group per product category or service, with a full set of headlines, descriptions, images and video. Missing video means Google generates one, which clients rarely love.
- Audience signals and search themes. Signals are a starting point, not targeting. Check they are relevant and that search themes cover the categories the asset group sells.
- Final URL expansion. On by default, it lets PMax send traffic to any page on the site. Check whether it is on, and whether URL exclusions keep traffic off the blog, careers page and other non-commercial pages.
- Search terms and channel reporting. PMax now reports search terms and channel performance. Pull both. If most spend is going to Display and YouTube with weak conversion, or the search terms are mostly brand, the campaign is not doing what the client thinks.
- Negative keywords. Campaign-level negatives are available for PMax. Check whether any have been added, and whether the account-level negative lists apply.
- New customer acquisition settings. If the client cares about new customers rather than repeat buyers, check whether the goal is set and whether a customer list defines “existing”.
- Listing group overlap. If Standard Shopping also runs, confirm which products each campaign covers, per the structure section above.
Landing pages
A perfect account sending clicks to a weak page is still a weak account. In GA4, compare engagement rate and key event rate by landing page for paid traffic. In Google Ads, add the Landing page experience column at keyword level and the mobile speed score at landing page level. Pages with low engagement and below-average landing page experience go to the top of the recommendations. Our guide to landing page optimisation for PPC and SEO covers what to fix.
The audit checklist
Copy this into your audit template. Each line is a yes, no or not applicable, with a note.
Structure
- Campaigns mirror the site’s service or product hierarchy
- Segmentation is manageable and each campaign has enough volume for its bid strategy
- Change History reviewed for discarded structures
- Display Network and Search Partners deliberately set on every search campaign
- Geo targeting correct; local accounts use a double campaign structure where appropriate
- Ecommerce: category search campaigns, DSA on product pages, Shopping or PMax with clear precedence
- Custom labels present in the feed
- Brand separated from generic; brand exclusions on PMax
Settings
- Location options set to Presence; user location report run and waste quantified
- Schedule and device adjustments consistent with the bid strategy
- Bid strategies fit conversion volume; targets realistic against 90-day actuals
- Budget-limited campaigns identified
- Campaign conversion goals correct
- Auto-apply recommendations reviewed and mostly off
Ad groups
- Every ad group has an RSA with varied headlines and limited pinning
- Ad copy relevant, with a call to action
- Five to fifteen keywords per ad group
- Match types fit the bidding strategy; broad match only with Smart Bidding and solid tracking
- Search terms reviewed; negative lists shared, applied and recently updated
- Sitelinks and callouts on every campaign; other assets as relevant; automated assets setting reviewed
- Remarketing, customer match and in-market segments attached; demographic exclusions sensible
Everything else
- Experiments run in the last 12 months
- Link checker script in place
- One source per conversion action; primary and secondary correctly assigned
- Enhanced conversions for web on; offline conversion import for lead generation
- Consent mode implemented where a cookie banner exists
- Auto-tagging on; GA4 linked; attribution consistent; conversion values passing
- PMax: brand exclusions, complete asset groups, URL expansion reviewed, search terms and channel reports pulled, negatives applied
- Worst landing pages identified by engagement and landing page experience
The time saving comes from the order, not from skipping steps. Structure first, then settings, then a sample of ad groups, then tracking. If you would rather have someone run it for you, that is what our Google Ads management team does every week.
