There is a version of the Google Ads audit that most guides still teach. It involves checking conversion tracking, reviewing keyword match types, auditing negative keyword lists, assessing Quality Scores, and examining ad copy. That version was accurate and useful for a long time. It is still relevant today — but it is no longer enough.
The Google Ads platform that advertisers operated in 2019 and even 2022 has been significantly restructured. Exact match no longer means exact. Manual bidding has been functionally deprecated for most advertisers. Campaign types that once separated brand from non-brand, search from display, prospecting from retargeting, now bundle those audiences and networks by default. Visibility has been deliberately narrowed in some areas and selectively restored in others.
Google’s Director of Product Management for Search Ads, Brandon Ervin, appeared on the Ads Decoded podcast to discuss campaign consolidation, AI Max, and what advertiser control looks like today. The conversation was substantive. Ervin’s team has shipped real improvements: brand exclusions in Performance Max and Demand Gen, site visitor and customer exclusions, network-level reporting within bundled campaigns, improved search term visibility, and semantic modeling that reduces learning period risk during consolidation.
These are genuine improvements. They are also, in nearly every case, solutions to problems that Google’s own product decisions created. The ability to separate brand from non-brand inside Performance Max is not new functionality — it restores a distinction that previously existed by default. Network-level reporting within a bundled campaign is not an expansion of transparency — it restores reporting that was removed when bundling was introduced.
That framing matters when you are conducting an audit. A control that reinstates baseline visibility is not the same as a control that expands advertiser leverage. A 2026 audit must distinguish between the two.
Why the Old Google Ads Audit Framework No Longer Captures the Full Picture
Traditional Google Ads audit frameworks were built around direct controls. An advertiser could set an exact match keyword and be reasonably confident about what would trigger it. They could set a manual bid and know what they were paying per click. They could segment campaigns by network and see performance data by channel with minimal ambiguity. Structural audits mapped directly to performance outcomes because the connections between structure and result were observable and adjustable.
That relationship has been progressively loosened. The new control architecture is indirect. You are no longer telling the algorithm where to go and what to pay. You are feeding it signals, setting targets, and trusting that it will allocate spend efficiently within those parameters. When it does not — when ROAS is acceptable at a blended level but marginal returns have turned negative at the last $50,000 of spend — the dashboard does not tell you that. It shows a blended metric that is still within tolerance, and the algorithm keeps spending.
This is not a minor calibration issue. It is a structural change to where performance problems accumulate and where audit methodology must focus. An advertiser who audits their 2026 Google Search campaigns using a 2020 framework will identify the easy problems (broken conversion tags, missing ad extensions, duplicate keywords) and miss the expensive ones (non-incremental conversions inflating ROAS, spend crossing the diminishing returns threshold, weak inventory subsidized by strong search).
The audit framework presented here addresses both. It covers the foundations — the non-negotiable structural checks that every account should pass — and it addresses the deeper economic questions that determine whether the account is actually generating value or just reporting metrics that look like value.
The Audit Foundations: What Must Already Be in Place
Before examining the more complex economic questions, every Google Search Ads audit should verify that a set of structural baselines are in place. These are not sophisticated optimizations. They are prerequisites. An account that has not cleared these items cannot be effectively evaluated on the dimensions that matter in 2026.
Conversion tracking accuracy. Every conversion action should be verified: the tag fires on the correct page, there are no duplicate firings, the conversion window reflects the actual sales cycle, and the conversion value — where applicable — is populated correctly. If offline conversions are available (MQLs, SQLs, revenue, customer lifetime value), they should already be imported and weighted by downstream conversion rates. An account feeding Google’s algorithm only surface-level events like form submissions, without any revenue or pipeline stage data, is giving the algorithm an incomplete and potentially misleading picture of what a conversion is actually worth.
Ad extensions fully deployed. Sitelinks, callouts, structured snippets, image extensions, and call extensions should all be active. This is not primarily about Quality Score. It is about occupying real estate in the search results page and providing the algorithm with additional signals about relevance and engagement.
Automated bidding configured intentionally. Smart bidding is appropriate for most accounts, but the target itself requires deliberate setting. A Target CPA or Target ROAS set too loosely allows the algorithm to spend past the point of profitable return. A target set too aggressively may cause the campaign to under-serve. The target needs to be anchored to actual downstream economics, not to a number that looks reasonable in isolation.
Search Partners and Display Network disabled for Search campaigns. This bears repeating because it remains common for these settings to remain enabled by default without any deliberate decision to include them. Search Partners and Display expansion deliver traffic at different quality levels than core Google Search. They should be evaluated separately and, for most accounts, excluded from pure Search campaigns.
Brand and generic campaigns separated. Brand queries and generic queries have fundamentally different cost structures, intent profiles, and competitive dynamics. Running them together in a single campaign structure makes it impossible to evaluate the economics of either accurately. Brand controls should be used to enforce this separation at the campaign level.
Negative keyword lists maintained actively. A negative keyword list that was built at account launch and never updated is not a functioning negative keyword list. The list should be reviewed regularly against search term data, organized into shared lists applicable across campaigns, and updated to reflect new patterns as query data accumulates.
Automatically created assets reviewed. Google can generate text assets automatically for Search campaigns. These should be audited for accuracy, brand tone, and factual correctness. Errors in automatically generated assets have appeared in accounts across industries and can cause compliance issues in regulated sectors.
Mobile vs. desktop performance tracked separately. Device performance gaps are frequently material. An account that shows a $45 CPA on desktop and a $110 CPA on mobile, without device bid adjustments to account for that gap, is paying desktop prices to fund mobile conversions that do not justify the spend.
Every account should clear these checks before the deeper audit begins. They are baseline requirements. The real work starts after that.
Signal Architecture: The New Source of Advertiser Control
In the direct-control era of Google Ads, the audit question was structural: Is the account set up correctly? Are keywords segmented properly? Are bids set appropriately? These are still valid questions, but they are no longer the primary levers.
In the current model, Google’s machine learning system makes the final call on which auctions to enter, what to bid in each one, and how to allocate budget across queries, audiences, and networks. The advertiser’s ability to influence that system is real, but it operates through data quality rather than direct instruction.
Brandon Ervin described this on the Ads Decoded podcast: “control still exists, it just looks different.” The accuracy of that statement depends entirely on whether advertisers understand what the new control architecture requires. Feeding the algorithm high-quality, economically meaningful conversion data is the most direct form of control available. Feeding it low-quality, surface-level data is the equivalent of giving a decision-making system bad inputs and then wondering why it makes bad decisions.
A signal architecture audit evaluates three dimensions:
Quality. What conversion data are you passing to Google’s algorithm? If you are passing only form fills or page visits, you are passing proxy events. If you are passing revenue data, pipeline stage, qualified lead status, or customer lifetime value, you are passing economically meaningful signals. The algorithm optimizes toward what it is told to optimize toward. If it is told to maximize form fills and form fills are a poor predictor of actual revenue, it will generate a high volume of economically meaningless conversions.
The fix is to import offline conversion data at the conversion event level. For B2B advertisers, this typically means connecting CRM data to Google Ads via the Google Ads API or a third-party connector, and passing lead stage or opportunity value back into the account as conversion values. For e-commerce, it means passing actual order value rather than a fixed conversion value, and weighting high-value customer segments appropriately.
Density. Does the algorithm have enough data to learn from? A campaign that generates fewer than 30-50 conversions per month in a Smart Bidding strategy is operating with insufficient data density. The algorithm requires volume to build a reliable model. Sparse conversion data produces noisy signals, and noisy signals produce erratic bidding.
For accounts with low conversion volume, the solution is often to import micro-conversions — not as primary conversion actions, but as supporting signals. A qualified phone call, a high-engagement page visit, or a catalog request may not be a purchase, but it is a meaningful step in the conversion path. Using these as value-weighted secondary signals, rather than as primary conversion events, can provide the algorithm with the density it needs without corrupting the economic signal.
Selectivity. Are you passing every conversion indiscriminately, or are you being intentional about what you tell Google counts as success? Net-new customer conversions are typically more valuable than returning customer conversions for growth-stage campaigns. High-value account conversions are more valuable than low-value account conversions for B2B campaigns. If everything is passed with equal weight, the algorithm treats everything as equally desirable and optimizes accordingly.
Selectivity means building conversion action sets that reflect your actual business priorities: new customer acquisition over repeat purchases when the objective is growth, pipeline stage over raw lead count when the objective is revenue quality, high-LTV cohorts over total conversion volume when the objective is customer economics.
Incrementality: The Metric Google Doesn’t Optimize For
Google’s algorithm optimizes toward reported conversions. It does not optimize toward incremental conversions. The difference between those two things is what most advertisers are missing when they look at their ROAS numbers and believe the account is performing well.
An incremental conversion is one that would not have occurred without the ad. A non-incremental conversion is one that would have occurred regardless — a user who already intended to purchase and simply clicked the ad on the way to completing a transaction they had already decided to make. Brand search campaigns capture a substantial portion of non-incremental conversions. Retargeting campaigns capture users who are already in motion. Performance Max and AI Max, by design, blend prospecting and retargeting signals, and they frequently over-index on warm brand traffic because that traffic converts at higher rates and is therefore more attractive to the bidding algorithm.
Ervin acknowledged this problem on the Ads Decoded podcast when asked whether AI-driven campaigns are over-indexing on warm brand traffic to inflate blended ROAS. His response pointed to partial solutions: use brand controls, theme your account more precisely, and use multi-campaign A/B testing. These are useful suggestions. They do not fully resolve the measurement problem.
A 2026 audit must address incrementality directly. At a minimum, this means:
Geo-based incrementality testing. Run a holdout experiment by pausing brand search or non-brand search in a matched set of geographic markets while leaving it active in another set. Measure the difference in conversions, revenue, and site traffic between the treatment and control groups. The difference represents the incremental impact of the campaign. This methodology is imperfect, but it is substantially more informative than relying on last-click or data-driven attribution, both of which credit the ad regardless of whether it caused the conversion.
Segmenting brand from non-brand search. If brand and non-brand traffic are running in the same campaign structure, they cannot be evaluated independently. Brand campaigns should be isolated and their performance evaluated separately from non-brand prospecting. The blended ROAS of a campaign that includes high-converting brand queries alongside lower-converting generic queries does not reflect the true cost of acquiring new customers.
Evaluating PMax and AI Max for brand traffic bleed. Performance Max campaigns, by default, can serve across all inventory including brand search queries. If a separate brand campaign is not capturing those queries with higher priority, PMax may be capturing them at higher CPCs than a dedicated brand campaign would pay, while attributing the resulting conversions to its own optimization engine. The audit should verify that brand queries are being correctly captured and attributed, and that PMax/AI Max is being held accountable only for the traffic that would not have converted through other campaigns.
Measuring lift, not just ROAS. Return on ad spend measures revenue divided by ad spend. It says nothing about whether the revenue was caused by the ad. Brands that have run incrementality tests frequently find that their true incremental ROAS is meaningfully lower than their reported ROAS, particularly for brand campaigns and retargeting. That gap represents the portion of spend that is capturing existing demand rather than generating new demand.
Marginal CPA: Why Your Last Dollar of Spend Is Almost Certainly Inefficient
Google’s Smart Bidding operates on a blended CPA. It will allocate spend until the average cost per conversion across all spend reaches the target CPA that has been set. What this means in practice is that the algorithm continues to bid in progressively lower-value auctions as the budget is consumed, and the last dollars of spend in any campaign period are typically the most expensive and least valuable.
This is not a bug. It is how auction-based systems work under volume-maximizing optimization. But it creates a real problem for advertisers who assume that all spend within a Target CPA is equally efficient. It is not. The first portion of spend — the high-intent, high-converting queries that the algorithm enters early in a budget period — may convert at a CPA significantly below the target. The final portion of spend, entered to fulfill the daily budget, may convert at a CPA significantly above what the business can justify.
Marginal CPA analysis maps this relationship explicitly. For a given campaign, you plot spend against incremental conversions at different spend levels. The resulting curve typically starts steep (high conversion rate per dollar) and flattens as spend increases (declining conversion rate per dollar). The point at which the marginal CPA — the cost of the next conversion — exceeds the maximum allowable CPA based on customer lifetime value is the point at which additional spend destroys rather than creates economic value.
Most advertisers are bidding well past that point and have no visibility into it because the blended average CPA remains within tolerance.
Conducting a marginal CPA audit requires the following steps:
First, pull historical data by day or week and plot total spend against total conversions for the period. Calculate the average CPA at multiple spend levels by isolating periods of lower budget from periods of higher budget. If the account has run budget experiments, those data points are valuable.
Second, segment by campaign. Marginal returns are not uniform across all campaigns. High-intent, specific campaigns typically have steeper curves (faster diminishing returns) than broader prospecting campaigns. A marginal CPA analysis at the account level obscures these differences.
Third, compare marginal CPA to customer lifetime value. A $50 CPA for a customer who generates $150 in lifetime value is profitable. A $50 CPA for a customer who generates $60 in lifetime value is marginally profitable on paper but may be unprofitable when churn, support costs, and acquisition overhead are accounted for. The algorithm does not know the difference unless you tell it through conversion value data.
Fourth, consider lowering targets deliberately to test whether reduced spend with a tighter target actually improves overall return. This is counterintuitive for advertisers and for Google’s sales organization, which has no incentive to recommend it. But a lower Target CPA forces the algorithm to be more selective, bidding only in auctions where the expected conversion probability justifies the bid. The result is fewer total conversions at a lower CPA, which may produce better economics than more conversions at a higher blended CPA.
Query Resolution and the AI Max Matching Problem
The quality of Google’s query matching has changed significantly over the past several years, and not uniformly for the better. Broad match has always introduced variability into what queries trigger which ads, but the variance has increased substantially. AI Max, launched in May 2025 as Google’s most automated search campaign type, has accelerated this trend. Advertisers running AI Max campaigns have reported increases in the volume of irrelevant queries flowing through the system — queries with no apparent connection to the advertiser’s business, keywords, or intended audience.
Ervin acknowledged on the podcast that some AI Max matches “can look a little wonky” and stated that his team is working on exposing the model’s reasoning for specific query matches. That is a useful direction. It does not yet resolve the immediate auditing problem, which is that advertisers need to evaluate query quality now, with the data they have.
A thorough query resolution audit involves:
Extracting the full search term report. The search term data available in Google Ads is not comprehensive — Google withholds some query data to protect user privacy — but the available data is sufficient for meaningful analysis. Export the full search term report for the audit period and classify each query by intent tier.
Intent tier classification typically looks like this: Tier 1 queries are high-intent, specific matches directly aligned with the advertiser’s product or service. Tier 2 queries are relevant but lower-intent — informational, research-phase, or broadly topical. Tier 3 queries are tangentially related, ambiguous, or low relevance. Tier 4 queries are irrelevant and should be excluded via negative keywords.
Calculating performance by tier. Compare CPA and, where available, revenue or customer lifetime value by query tier. The typical pattern is that Tier 1 queries convert at the lowest CPA with the highest downstream value, and conversion performance degrades through Tier 2 and Tier 3. If the campaign is spending heavily against Tier 3 and Tier 4 traffic, the blended CPA may be within tolerance but the actual economics are being supported disproportionately by Tier 1 traffic, with Tier 3 and 4 consuming budget that is generating little return.
Auditing match type distribution. Google’s recommendation engine consistently pushes advertisers toward broad match. Broad match has legitimate uses — it can expand reach efficiently when the campaign has strong conversion data and accurate negative keyword coverage — but it requires active management. An audit should examine what portion of conversions are coming from exact intent matches versus broadly matched queries, and whether the performance differential justifies the match type mix.
Measuring the impact of AI Max on query quality. For accounts running AI Max, the audit should specifically compare query relevance and conversion quality in AI Max campaigns against standard Search campaigns running in the same account. If AI Max is generating more total conversions but at a lower quality level (higher CPA, lower downstream value), that tradeoff should be explicit and deliberate rather than invisible.
Maintaining a structural negative keyword schema. The shift toward larger, more consolidated ad group structures — driven by Google’s consolidation recommendations — makes query control via negatives more important, not less. Fewer ad groups means that performance differences between query types cannot be managed by adjusting bids at the ad group level. They must be managed through exclusions. A structural negative keyword schema — a hierarchical set of negative lists that reflects the intent tiers of the account — is essential for any consolidated account structure.
Network Economics: What Bundled Campaigns Are Actually Spending On
Performance Max campaigns serve across all of Google’s inventory: Search, Shopping, Display, YouTube, Gmail, and Discover. AI Max campaigns extend the same bundling logic to Search-focused campaigns with significantly expanded matching. The value proposition, as Google presents it, is that the algorithm will allocate spend to whichever channel drives the best performance at any given moment.
The problem with that framing is that it assumes perfect cross-channel measurement. In practice, attribution across Google’s owned properties is not neutral. YouTube view-through conversions are measured differently than Search click conversions. Display and Discover impressions are credited differently than Search clicks. When all of these channels are bundled into a single campaign, and the campaign is evaluated on a blended ROAS metric, the result is a single number that obscures which channels are actually generating value and which are being cross-subsidized.
The structural incentive here is worth naming directly: Google has both the highest-value inventory (core Search) and lower-value inventory (Display, Discover, Gmail, tail YouTube) running within the same campaign structures. When an advertiser evaluates the campaign on its blended performance, the surplus value generated by high-intent Search queries effectively subsidizes the lower-performing inventory. The advertiser’s spend is being allocated across a broader set of placements, with less visibility into which placements are driving results.
A network economics audit requires:
Breaking out performance by network. Where network-level reporting is available within PMax or AI Max, extract it and compare CPA and conversion value by network. This data has become more accessible following Google’s recent transparency improvements, but it requires deliberate extraction and analysis.
Evaluating cross-subsidization. If Search is converting at a $30 CPA and Display is converting at a $95 CPA within the same campaign, the blended $45 CPA is being held down by Search performance while Display spend continues. Identifying this pattern requires network-level disaggregation.
Comparing Search campaigns to PMax/AI Max on the same queries. Where both campaign types are running, analyze search term overlap and compare CPA by query for matched queries. If PMax or AI Max is serving against the same high-intent queries as a dedicated Search campaign but at higher CPCs, the bundled campaign is competing with and potentially cannibalizing the more cost-efficient structure.
Auditing Search Partners separately. Search Partners — the network of third-party sites that serve Google search ads — frequently deliver traffic at different quality levels than core Google.com Search. For many advertisers, Search Partners traffic converts at a significantly higher CPA. This should be verified by pulling the Search Partners network data and comparing performance to core Search. For most accounts, Search Partners should be excluded unless there is affirmative evidence that they contribute positively to account economics.
Value Redistribution: Where Your Best Search Budget Goes
The cumulative effect of the structural dynamics described above produces a specific pattern that is worth naming as a distinct audit category. Call it value redistribution.
High-intent search queries — the queries that represent genuine purchase intent from users who are actively looking for what you sell — are among the most valuable advertising inventory that exists. There is a limited and relatively inelastic supply of these queries. Advertisers who are well-positioned to win these queries, who have relevant ads and competitive bids, generate a substantial surplus on that inventory.
What bundled campaign structures do is take that surplus and spread it across inventory that could not generate it independently. This happens through four mechanisms:
Non-incremental conversions — primarily brand search and retargeting — inflate reported conversion counts, making overall performance appear stronger than the incremental results justify. This signals to the algorithm that the account is performing well, which sustains spend at levels that would not be justified by true incremental performance alone.
Loose match types expand where ads appear. Broad match and AI-expanded matching pull in queries that are tangentially related to the advertiser’s keywords. These queries convert at lower rates, diluting the intent precision of the campaign. More spend flows to lower-intent traffic, funded by the budget that would otherwise concentrate on high-intent traffic.
The absence of clean marginal return visibility means that the spend threshold at which returns turn negative is invisible. The algorithm continues to bid in progressively lower-value auctions without any signal from the account’s performance reporting that the incremental economics have turned.
Network bundling hides channel-level performance. Weaker networks — Display, Discover, Gmail, low-intent YouTube — are evaluated alongside Search within the same campaign metrics, obscuring their individual economics.
The practical effect is a redistribution of value from the advertiser’s best inventory to inventory that Google needs to monetize but that generates lower value for the advertiser. An audit that identifies this pattern and quantifies its scale gives the advertiser a basis for restructuring spend toward a more concentrated, higher-value configuration.
A Complete Google Search Ads Audit Framework for 2026
Bringing all of the above together, a complete 2026 Google Search Ads audit should be structured as follows:
Phase 1: Structural Foundations Verification
- Confirm conversion tracking accuracy, including offline conversion imports
- Verify automated bidding targets are anchored to downstream economics
- Check that Search Partners and Display Network are disabled for pure Search campaigns
- Confirm brand and generic campaigns are separated with brand controls enforced
- Review all active negative keyword lists for completeness and recency
- Audit automatically created assets for accuracy and brand compliance
- Confirm ad extensions are fully deployed
- Check device performance and verify appropriate bid adjustments
Phase 2: Signal Architecture Assessment
- Evaluate the quality of conversion data being fed to Smart Bidding (surface events vs. revenue data)
- Assess conversion volume and data density against Smart Bidding requirements
- Review conversion action configuration for appropriate selectivity (new vs. returning customers, high-value vs. low-value segments)
- Confirm offline conversion imports are functioning and weighted appropriately
- Verify that high-LTV customer signals are being passed for audience-level optimization
Phase 3: Incrementality Analysis
- Segment brand from non-brand spend and evaluate each independently
- Identify the proportion of conversions that are likely non-incremental (brand, retargeting, existing customer)
- Review the geo-test framework for incrementality measurement
- Assess whether PMax/AI Max brand controls are properly configured
- Evaluate the blended ROAS against a reasonable estimate of incremental ROAS
Phase 4: Marginal CPA Modeling
- Plot spend against incremental conversions for major campaigns
- Estimate marginal CPA at different spend tiers
- Compare marginal CPA at current spend levels to maximum allowable CPA based on customer LTV
- Identify campaigns that have likely crossed the diminishing returns threshold
- Evaluate whether reducing targets would improve overall account economics
Phase 5: Query Resolution Review
- Extract and classify the full search term report by intent tier
- Compare CPA and conversion value by intent tier
- Quantify the proportion of spend allocated to Tier 3 and Tier 4 queries
- Audit match type distribution and evaluate whether broad match is generating net-positive returns
- For AI Max campaigns, specifically compare query quality to standard Search campaigns
- Review and update the structural negative keyword schema
Phase 6: Network Economics Analysis
- Extract network-level performance data from PMax and AI Max campaigns
- Calculate CPA and ROAS by network (Search, Display, YouTube, Discover)
- Identify cross-subsidization patterns
- Evaluate Search Partners performance versus core Google Search
- Compare PMax/AI Max query overlap with dedicated Search campaigns
The argument for this level of audit rigor is straightforward. Google’s automation is genuinely powerful and in many cases produces results that manual management cannot match. The goal is not to argue against automation. The goal is to give automation the highest-quality inputs, evaluate its outputs at an economic level rather than a metric level, and identify when structural features of the platform are working against the advertiser’s interests rather than with them.
The improvements Google has shipped — brand exclusions, network reporting, better search term visibility, semantic consolidation — are meaningful. They are also, in large part, improvements to visibility that the platform previously had and removed. An audit that simply accepts the current state of controls as the baseline will miss the degree to which those controls represent a net reduction in advertiser leverage compared to what was available five years ago.
The practical question is not whether Google’s AI is good at optimization. It is. The practical question is: what does the AI optimize toward, who benefits from the default configuration, and what does an advertiser need to do to ensure the optimization engine is working in their economic interest rather than the platform’s?
Signal quality, incrementality measurement, marginal return analysis, query classification, and network disaggregation — these are not advanced tactics for sophisticated advertisers. They are the minimum requirements for understanding what your Google Search Ads spend is actually doing in 2026. Anything short of that is not an audit. It is a dashboard review.
Frequently Asked Questions: Google Search Ads Audit 2026
Q: What is a Google Search Ads audit and why does it matter in 2026?
A Google Search Ads audit is a structured review of a Google Ads account that evaluates whether the account’s structure, targeting, bidding, and data inputs are aligned with the advertiser’s business objectives. In 2026, audits matter more than they did in previous years because the platform has shifted from direct advertiser controls to algorithm-mediated optimization. Advertisers who do not audit the inputs to Google’s AI — their conversion data quality, incrementality assumptions, and query matching configuration — are effectively ceding control to a system optimized for volume rather than advertiser-specific economics.
Q: How is auditing Google Search Ads in 2026 different from auditing in 2021 or 2022?
In 2021, an audit focused primarily on structural elements: campaign segmentation, keyword match types, bid strategy configuration, and Quality Score optimization. These elements are still relevant but no longer primary. In 2026, the most important audit dimensions are economic: Are you feeding the algorithm the right conversion data? Are your conversions incremental? Are you spending past the point of diminishing returns? Is your best search budget being redistributed to lower-value inventory? These questions require a fundamentally different analytical approach than traditional structural audits.
Q: What is signal architecture in the context of Google Ads?
Signal architecture refers to the set of data inputs that an advertiser provides to Google’s machine learning system to guide its optimization decisions. This includes the types of conversion events being tracked, the values assigned to those events, the offline data imported from CRM or sales systems, and the audience signals provided to inform targeting. In the current Google Ads model, signal architecture is the primary mechanism through which advertisers exercise control. An algorithm fed accurate, economically meaningful data will make better decisions than one fed surface-level proxy metrics.
Q: What does incrementality mean in Google Ads, and how do I measure it?
Incrementality refers to the portion of conversions that were caused by the ad — conversions that would not have occurred without the ad being served. Google’s optimization engine reports all attributed conversions as though they were causally produced by the ad. In reality, a significant portion of brand search conversions, retargeting conversions, and some PMax conversions represent users who would have converted anyway. Measuring incrementality typically requires geo-holdout experiments, where campaigns are paused in test markets while remaining active in control markets, and the conversion difference between the two is measured.
Q: What is marginal CPA and why does it matter for Google Ads audits?
Marginal CPA is the cost of the next incremental conversion — what you are actually paying for each additional conversion as you spend more. It is different from average CPA, which is the total spend divided by total conversions. Google’s Smart Bidding manages to an average CPA target, meaning it will continue spending until the blended average falls within the target. But the marginal cost of conversions at higher spend levels is typically substantially higher than the average. Auditing marginal CPA reveals whether the last portion of the advertising budget is generating returns above or below the economic threshold that the business requires.
Q: How does AI Max affect my Google Ads audit process?
AI Max for Search, launched in May 2025, extends Google’s automated matching to cover a broader range of queries than traditional keyword targeting. It can match ads to queries outside the explicit keyword list based on the algorithm’s interpretation of the business’s relevance. This introduces both opportunity and risk. The opportunity is reach expansion; the risk is query quality degradation. Independent analysis of AI Max campaigns has found median revenue increases alongside median CPA increases. Auditing AI Max campaigns requires specific attention to query classification, comparison of match quality against standard Search campaigns, and evaluation of whether AI-expanded queries are generating economically justified returns.
Q: Should I use broad match in Google Ads in 2026?
Broad match has evolved significantly. When used with Smart Bidding and high-quality conversion data, it can expand reach efficiently to queries that convert. However, it requires active management through negative keywords and regular search term audits. The audit question is not whether broad match is appropriate in principle — it can be — but whether it is generating net-positive returns in your specific account. Classify search terms by intent tier and measure CPA and conversion quality by match type. If broad match traffic is converting at a CPA significantly above the average, and the downstream value is lower, the match type configuration should be adjusted.
Q: How do Performance Max campaigns affect a Google Search Ads audit?
Performance Max campaigns run across all of Google’s inventory simultaneously, including Search. They can cannibalize dedicated Search campaigns by bidding on the same high-intent queries, sometimes at higher CPCs. They frequently blend brand and non-brand, prospecting and retargeting traffic in ways that inflate reported ROAS without generating proportional incremental lift. A complete audit must evaluate PMax performance at the network level, verify that brand controls are properly configured, and assess whether PMax campaigns are generating incremental conversions or primarily capturing traffic that dedicated Search and brand campaigns would have converted at lower cost.
Q: How often should I audit my Google Search Ads account?
A full audit of the type described here — covering structural foundations, signal architecture, incrementality, marginal returns, query resolution, and network economics — should be conducted at minimum quarterly. Structural checks (conversion tracking, negative keywords, ad extensions, bidding targets) warrant monthly review for active accounts with meaningful spend. Marginal CPA analysis should be reviewed whenever budget or target changes are made, as these changes shift the equilibrium point. Query term audits should be reviewed weekly for high-volume accounts or accounts running AI Max or broad match at scale.
Q: What data do I need to conduct a marginal CPA analysis?
At minimum, you need historical data showing spend and conversions at different budget levels over comparable time periods. If the account has run budget experiments or geographic tests at different spend levels, those provide the most reliable data points. Otherwise, segment weekly data by spend level and plot spend against conversions. Calculate the average CPA in low-spend periods versus high-spend periods. The difference between those CPAs, adjusted for market conditions, provides an estimate of marginal returns. For a more precise analysis, using Google’s Performance Planner to model expected conversion volume at different budget levels can supplement historical data.
Q: What is value redistribution in Google Ads, and how do I identify it in my account?
Value redistribution is the process by which surplus generated by high-intent, high-converting search queries gets allocated across lower-value inventory through bundled campaign structures and loose matching. It manifests as a blended performance metric that appears acceptable but masks the underperformance of specific channels or query types. To identify it, break down performance by network within PMax campaigns, classify search terms by intent quality, calculate CPA at the query tier level, and compare the economics of high-intent traffic against the blended account average. A large gap between Tier 1 query performance and overall account CPA is a signal that value redistribution is occurring.
Q: Can I conduct this type of audit without third-party tools?
Most of the audit dimensions described here can be evaluated using Google Ads’ native reporting and Google Analytics 4, with data exported to a spreadsheet for marginal CPA modeling. Full incrementality testing requires experimental design that can be implemented within Google’s native campaign experiment framework. Third-party tools can streamline data extraction and visualization, but they are not required. The more important prerequisites are access to full search term data, conversion data segmented by type and value, network-level campaign reporting, and the ability to segment performance by spend tier or time period.
Q: What is the most common mistake advertisers make in Google Search Ads audits today?
The most common mistake is auditing what is easily visible — account structure, keyword lists, ad copy, Quality Score — and treating those findings as a complete audit. The structural elements of the account are a necessary starting point, but they do not reveal whether the conversion data is economically meaningful, whether reported conversions are incremental, whether spend has crossed the diminishing returns threshold, or whether high-value budget is being redistributed to low-value inventory. These economic questions are harder to measure and require more analytical work, which is precisely why they are routinely missed. They are also where the largest performance improvements for sophisticated advertisers reside.
Q: How do offline conversion imports improve Google Ads performance?
Offline conversion imports allow advertisers to feed post-click conversion data back into Google Ads from external systems. For B2B advertisers, this typically means passing CRM data — lead qualification stage, opportunity value, closed revenue — back into the ad account so the bidding algorithm knows which lead types are most valuable. For e-commerce advertisers, it can mean passing actual order value, return rate-adjusted revenue, or repeat purchase data. When the algorithm has access to this downstream data, it can optimize toward conversions that are actually valuable rather than proxy events that may or may not correlate with revenue. This is one of the highest-leverage improvements available in most accounts that have not already implemented it.
Q: How does this audit approach apply to smaller advertisers with lower budgets?
The framework scales with account size. Smaller accounts with lower conversion volumes will find the marginal CPA analysis less statistically robust, and formal incrementality testing may require time horizons that are not practical for small accounts. However, the foundational principles apply regardless of scale: feed the algorithm the best quality data available, separate brand from non-brand, disable Search Partners unless there is evidence they are contributing positively, maintain active negative keyword lists, and audit search terms regularly. Signal architecture — passing meaningful conversion data rather than surface events — is arguably more important for small accounts than large ones, because smaller budgets have less room to absorb the cost of poor signal quality.
The standards for effective Google Search Ads management have shifted from structural configuration to economic intelligence. The advertisers who perform well in the current environment are not necessarily the ones with the most sophisticated campaign architectures — they are the ones who feed the algorithm the right data, measure what the algorithm actually produces rather than what it reports, and maintain a clear view of when and how platform incentives diverge from advertiser interests. An audit that addresses these questions honestly is not a defensive exercise. It is the basis for knowing whether your advertising spend is generating real economic returns or producing metrics that look like performance on a dashboard.
About ALM Corp
ALM Corp is a digital performance marketing consultancy that works with B2B and B2C brands to build paid search programs grounded in economic accountability rather than platform metrics. The firm provides Google Ads audits, paid search strategy, conversion data architecture, and ongoing campaign management for clients who need more from their advertising spend than dashboard-level reporting.
The audit framework described in this article reflects ALM Corp’s standard approach to evaluating search accounts: starting with structural foundations, assessing signal quality, measuring incremental performance, modeling marginal returns, and identifying where platform mechanics diverge from client interests. Whether you are taking over an underperforming account, evaluating the performance of a current agency, or preparing for a budget planning cycle, ALM Corp provides the analytical depth that complex paid search environments require.
To learn more about ALM Corp’s Google Ads audit services or to schedule a consultation, visit www.almcorp.com.
