Introduction
Your client's display budget went from $80K to $52K in attributed revenue this quarter. The campaigns look healthy in the platform reports. CTR is stable, impressions are pacing, the bidding strategy has not changed. But the conversion column tells a different story, and the renewal conversation is three weeks away.
Display and video campaign underperformance in 2026 rarely arrives as a single broken setting. It compounds across four layers that most agencies audit in isolation when they should audit them in sequence. Tracking degrades quietly under Consent Mode V2. Targeting drifts as audience signals decay and PMax pushes spend toward inventory the algorithm finds cheap rather than valuable. Creative fatigues faster than refresh cycles can keep up, with CTR decay now measurable within 7 to 14 days on prospecting audiences. Media buying gets evaluated against the wrong attribution model, so display and video lose credit for the upper-funnel work they actually do.
This audit framework was built for mid-market agency Media Buyers and Campaign Managers who need a defensible process, not a checklist. It moves from the data layer outward, since every optimization downstream of broken tracking optimizes against a distorted picture. By the final stage, you have a prioritized fix list with owners, impact ranges, and verification metrics. That document is what protects the retainer.
The platform direction in 2026 makes this work non-optional. Google removed Display and Video from Performance Planner in March. Bundled campaign formats hide channel-level economics by design. Standalone impression-share planning is being deprecated in favor of conversion-tracked structures. Agencies that audit on the old logic will keep finding symptoms. The framework below finds causes.
Why Display and Video Campaigns Fail Differently Than Search in 2026
Display and video campaigns produce the kinds of impressions that algorithms reward, even when those impressions produce nothing of value for the advertiser. Search Engine Land documented in March 2026 that bundled campaign formats like Performance Max and Demand Gen offer limited visibility into which networks drive results, making it hard to cut the underperforming ones. The structural effect of that bundling is value redistribution, where strong query revenue subsidizes weak display placements inside the same blended report.
The platform side adds its own pressure. Google removed Display and Video from Performance Planner on March 9, 2026, along with impression share, top impression share, or absolute top impression share as primary metrics. Agencies that built quarterly forecasts on those plans lost the ability to view or edit them. The signal is direct. Standalone display and video planning around impression-share metrics is being deprecated, and audits need to move toward conversion economics.
A 2024 study by Optmyzr across 24,702 Performance Max campaigns found that 82% of advertisers were running PMax alongside other campaign types, and PMax consistently underperformed those other campaigns when they competed for the same traffic. The implication for an audit is that the conclusion "the campaign is underperforming" is rarely a single-layer answer.
Stage 1: Tracking and Measurement Diagnosis
Before adjusting a single placement, prove the data is real. Trackingplan estimated in 2026 that standard Google Ads conversion tracking is projected to miss 30 to 50% of actual conversions due to cookie restrictions and privacy changes. An audit that skips the tracking layer optimizes against a distorted picture.
Conversion Setup Validation
Open the conversions table in Google Ads and check each row for current status. Then verify the source. Then confirm whether each conversion sits as primary or secondary. Common failures cluster around four patterns that almost every account hides somewhere.
Conversion source duplication: GA4 imports and native Google Ads tags counting the same purchase, halving reported CPA and confusing Smart Bidding.
Misclassified events: an eCommerce purchase tracked as view_item from a misconfigured GTM trigger, which distorts the conversion volume Smart Bidding optimizes toward.
Incorrect counting method: lead actions set to "Every" instead of "One," which inflates form-fill volume when the same prospect submits twice.
Stale primary actions: a primary conversion with zero recent fires while bidding still references it as the optimization target.
A practical confirmation step is to submit a real test purchase, watch it land in GA4 Realtime, wait 24 hours for the GA4 Events report to confirm, then check Google Ads Conversions after 48 hours. If the three numbers diverge by more than a few percent without an obvious cause, the data layer is the audit's first fix.
Consent Mode and Privacy Signals
Consent Mode V2 is now load-bearing for measurement in the EEA and UK. The April 2026 GA4 update made it a hard requirement for GA4 data to flow accurately into Google Ads conversion tracking in the EEA and UK, and introduced stricter behavioral modeling thresholds globally. The Trackingplan analysis noted that recent consent mode changes affecting server-side tracking and cross-platform discrepancies account for 19% of issues, and common fixes often miss nuanced denials involving ad_storage and ad_user_data.
A further structural shift takes effect June 15, 2026. After that date, the Google Signals toggle in GA4 will no longer determine whether advertising data is shared with Google Ads, and Consent Mode (specifically ad_storage) becomes the primary control for Google Ads data collection and identifier access. Agencies running EU traffic should validate the timing of consent updates inside GTM Preview, not the banner UI, since consent state matters at the moment tags execute.
Enhanced Conversions and Modeled Measurement
Enhanced Conversions remain the single most effective recovery mechanism. Enhanced conversions send hashed first-party customer data to Google alongside conversion events, and advertisers typically see a 5% to 15% increase in reported conversions after implementing them. For accounts running advanced Consent Mode, modeling activation requires at least 700 ad clicks over 7 days per country/domain pair, seven full days of data collection, and sufficient consent rate. Audit the modeled-versus-observed split inside Google Ads under Goals and Conversions Summary, since the ratio explains why reported volume can look stable while raw observed conversions decline.
Attribution Model Audit
A campaign labeled underperforming under last-click attribution is often correctly performing under data-driven attribution. The Improvado 2026 audit framework notes that strict last-click systematically undercounts upper-funnel channels like display, video, and brand campaigns that create awareness but do not drive immediate conversions. The Digital Applied benchmarks add quantitative weight to the same point. Multi-touch and view-through attribution typically credit display 30 to 50% higher conversion contribution, so if your stack runs strict last-click, treat display benchmarks as a floor and instrument view-through measurement before declaring a channel underperforming.
The audit step here is twofold. Confirm the attribution model in Google Ads matches the GA4 conversion source. Then inspect the conversion path report in GA4 to see how often display or video appears in the assist position. A display campaign producing 60% of assists but 4% of last-click credit may be paying for itself through view-through value the headline report ignores.
Stage 2: Targeting and Audience Diagnosis
Once the data is trustworthy, the next question is whether the campaign reached the right people in the first place. Targeting failures in display and video tend to be quiet, since the volume of impressions makes performance look active even when the audience composition is wrong.
Audience Signal Quality in Performance Max and Demand Gen
For Performance Max running display and video inventory, audience signals act as a directional input, not a constraint. The Promodo 2026 checklist recommends to add segments for your most valuable customers, those with high LTV or high average order value, and exclude audiences with low ROAS. The audit work is to verify that the audience signals actually represent high-value users rather than generic in-market lists copied from a template.
Three audience-side checks are worth running on every account.
Customer match list freshness: any list older than 90 days is decaying as users churn or change email addresses, which weakens the signal Google uses for lookalikes.
Exclusion hygiene: existing customers, recent converters, and low-LTV segments should be excluded from prospecting, especially in PMax where the algorithm otherwise re-targets the warmest part of the funnel and inflates blended ROAS.
Brand exclusion configuration: brand exclusions applied at the PMax level prevent the campaign from absorbing branded search traffic that a dedicated branded Search campaign should handle.
Network and Placement Breakdown
Network bundling hides the channel-level economics that an audit needs to expose. ALM Corp's 2026 audit framework recommends advertisers 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. For a standalone display campaign, the equivalent check is the placements report.
Look at the top 50 placements by spend. Watch for mobile gaming apps. Watch for made-for-advertising sites. Watch for content categories with no contextual relevance to the product. Their presence signals that the algorithm has found cheap impressions the audience targeting allowed. Excluding those placements often produces a measurable CPA improvement within 7 to 10 days, since spend reallocates toward inventory with better contextual alignment.
Frequency Distribution Analysis
Frequency caps are one of the few media buying levers still under direct agency control on most platforms. The Digital Applied 2026 benchmarks identify empirical sweet spots of 5 to 7 impressions per user per week for prospecting, 3 to 5 for retargeting. Pull the frequency report from Google Ads, DV360, or the relevant DSP, then check the long tail of the distribution. If 15% of impressions are being delivered to users at 20-plus weekly frequency, the campaign is burning budget on people who have already decided.
Stage 3: Creative and Asset Diagnosis
Creative is where underperformance compounds fastest. Hawky AI's 2026 fatigue analysis defines a confirmed fatigue case as CTR dropping 15% or more from its 7-day rolling baseline, CPM rising 10% or more in the same window, hook rate declining 20% or more, or frequency exceeding 3.5 on prospecting audiences, with two or more signals breaching together.
Fatigue Versus Audience Saturation
The most common diagnostic error is treating audience saturation as creative fatigue, since the symptoms overlap. AdsGo's 2026 playbook makes the distinction operational. If one creative declines while others in the same ad set remain healthy, it is creative fatigue, and if all creatives decline simultaneously, it is likely audience saturation. The audit step is to chart CTR decay across every active creative on the same axis. Divergent curves point to creative refresh as the fix, while convergent decline points to audience expansion.
For prospecting campaigns, Finsi AI's 2026 fatigue research notes that fatigue typically starts at frequencies of 2.5 to 4.0, depending on creative quality and audience relevance, while retargeting tolerates higher frequencies of 6 to 8 because those people already know the brand. Use these thresholds as a starting benchmark, then compare the account's own history to set realistic refresh cadences.
Asset Group Composition in Performance Max
For PMax-driven display and video inventory, asset group depth is a leading indicator of output quality. Two Spouts' audit guide recommends working through the asset rating system in Google Ads, since the platform categorizes assets as Low, Good, or Best, and focusing first on the assets rated as Low identifies the underperforming elements that may negatively impact campaign effectiveness. The audit step is to count assets in each rating tier per asset group, then flag any group where more than a third of headlines or descriptions or images sit in the Low band.
Promodo's 2026 PMax checklist recommends a minimum input package that covers text, image, and video so the algorithm has multiple combinations to optimize. Thin asset groups are one of the most common underperformance patterns documented in the SearchEngineJournal 2026 review of PMax. When inputs are limited or generic, the output reflects it, and separate asset groups by product category or audience segment is usually a better structure than one asset group per campaign.
Format and Inventory Tier Alignment
Format choice is now more economically meaningful than industry vertical for display performance. Digital Applied's 2026 benchmark dataset reports that rich media units average 1.84% CTR and video display formats deliver 73% higher CTR than static banners, while native units hit 1.16%, and standard display banners average just 0.46%. The same dataset shows that average CPM on the Google Display Network sits at $3.12, while private marketplace deals now average $8.20 and CTV display reaches $24.50. A campaign that targets a brand-safe audience on open exchange at GDN CPMs is buying a different product than the same campaign on PMP inventory, even when the creative is identical.
The audit question becomes whether the creative format and the inventory tier match the campaign objective. Upper-funnel awareness goals justify the CTV premium when completion rates and dwell time matter, while direct-response retargeting on open exchange may be the right call for budget-constrained accounts.
Stage 4: Bidding, Budget, and Media Buying Diagnosis
Media buying issues often look like creative problems on the surface, since both produce flat conversion graphs. Separating them requires inspecting the bidding configuration against the conversion data it actually receives.
Bid Strategy Realism
WordStream's PMax and display audit framework opens with a question agencies often skip. Realistic expectations, since the first thing to check on bid strategies is whether people have realistic ideas of what they can achieve. A target CPA set 40% below the account's historical average forces the algorithm to either pull back spend or chase low-quality conversions to meet the math. Either outcome looks like underperformance.
The audit step is to pull the conversion history for the last 90 days, compute the median CPA, and compare it to the current Target CPA. A gap above 20% in either direction warrants either a target adjustment or a strategy change.
Smart Bidding Data Density
Smart Bidding accuracy degrades sharply below certain conversion thresholds. For Target CPA, an account producing fewer than 30 conversions per month leaves the algorithm with insufficient data to optimize against the macro goal. The Vijay Bhabhor 2026 guide suggests tracking micro-conversions when conversion volume is low, since if you have fewer than 30 conversions per month, track micro-conversions alongside macro-conversions, which gives Smart Bidding enough signals to optimize effectively. The audit step is to map conversion volume per campaign against the bid strategy in use, then flag any campaign where the math does not support the strategy.
Budget Allocation and Pacing
Mid-month pacing reports often expose allocation problems that monthly performance summaries miss. WordStream's audit guidance asks whether budget distributions across campaigns are in line with performance, whether the campaigns with better performance have more budget while the underperformers have less, and whether there are campaigns that are budget-capped that need an increase to perform better. The audit step is to compare each campaign's daily budget to its impression share lost due to budget. A campaign hitting 100% of its budget cap for the last 14 consecutive days is either ready to scale or wasting impression share on a flawed strategy.
Incrementality and Diminishing Returns
The Search Engine Land 2026 analysis of automated bidding flags a measurement gap most accounts ignore. Google uses a blended cost-per-action, so for example the first $50K of spend might return a $30 CPA, while the next $50K might return $120, and with automation, money is spent until the blended metric falls within tolerance, meaning the last dollar is not spent efficiently. For display and video campaigns specifically, this matters because non-incremental conversions inflate the headline number while the marginal conversion gets more expensive.
A practical audit method, even without dedicated incrementality tools, is to plot weekly spend against weekly incremental conversions and estimate marginal CPA at each spend tier. Where the marginal CPA exceeds the account's contribution margin, the campaign is past the point of efficient scale.
Stage 5: Turning the Audit Into a Prioritized Fix List
An audit that ends with findings is incomplete. The Improvado 2026 framework structures recommendations into impact tiers, and that structure works well for agency-client conversations. Tier 1 covers high-impact, low-effort changes that can be implemented within 30 days, like pausing underperforming campaigns, fixing broken tracking, and reallocating budget to top-performing channels. Tier 2 covers strategic changes that require planning and resources, like implementing a new attribution model. Tier 3 covers long-term initiatives that require significant investment.
For most mid-market agency audits, the prioritization sorts as follows. Tracking fixes and consent mode validation go first, since every downstream optimization depends on clean data. Bid strategy realism and budget reallocation come next, since they produce measurable change within the first reporting cycle. Audience exclusions follow on a 30-day window. Placement blocks and asset group restructuring sit on a 60-day window. Strategic rebuilds, like attribution model migration or account restructuring, sit on a quarterly plan.
The deliverable is a single document with each finding mapped to an owner, an expected impact range, and a verification metric. That structure is what turns a campaign management services engagement from a reactive monthly report into a defensible, repeatable process.
A Note on the 2026 Platform Direction
The deprecations stacking up across Google's ecosystem point in one direction. Standalone display and video campaigns optimized around impression metrics are being phased into bundled, conversion-optimized formats. Agencies that build their audit practice across the four diagnostic stages above will be in a stronger position to defend channel value when the platform reporting layer keeps changing underneath them. The four stages are tracking, then targeting, then creative, then media buying.
The audit is the product. The fix list is the deliverable. The retainer is the outcome.
