Skip to main content
    Audit Findings

    50 ecommerce Google Ads audits£1.2M in recoverable spend, one repeating pattern

    We audited 50 ecommerce Google Ads accounts spending £10k+/month. The waste wasn't in the places agencies tell you to look. Here's what we actually found.

    11 min readApril 2026By Chris Avery

    Most Google Ads audits come back with a tidy executive summary and five generic recommendations: "tighten audiences," "refresh ad copy," "add negative keywords." We ran 50 audits on ecommerce accounts between Q2 2025 and Q1 2026. The recommendations weren't generic. They were the same six problems, repeated almost verbatim, across brands in furniture, pet food, supplements, skincare, drinkware, apparel, and B2B hospitality.

    Total recoverable waste across the 50 audits: £1.2M per quarter. Median per account: £22k. Largest single-account finding: £186k.

    This isn't a marketing post. It's the pattern.

    Our methodology

    Every account we audited met three criteria:

    1. Monthly Google Ads spend of £10k or more
    2. An ecommerce revenue model with variable product margins (i.e. not every SKU has the same contribution margin)
    3. At least six months of account history to analyse

    We treated each audit as a P&L exercise, not a platform exercise. That means we didn't rate accounts against Google's Optimisation Score. We rated them against a brand's contribution margin, SKU-level profitability, and cash conversion cycle. Waste, in our definition, is spend a CFO would cut if they saw what it was actually doing.

    For each account we pulled:

    • 12 months of conversion data, segmented by SKU and campaign type
    • Feed quality metrics (disapprovals, missing attributes, incorrect pricing)
    • Search term reports for all Performance Max asset groups, where visible
    • Bid strategy performance versus realised contribution margin
    • Budget allocation versus first-party demand signals

    Everything in this post is an anonymised composite of those audits. The specific numbers below are medians across the 50 accounts.

    The six patterns we found

    1

    Performance Max is funding your low-margin SKUs

    43% of PMax spend on bottom-quintile margin SKUs

    2

    Branded search is hiding inside Performance Max

    37 of 50 accounts serving against own trademark

    3

    Broad match keywords live too long after they stop converting

    Median £4,200/mo wasted per account

    4

    Smart Shopping migrations were never fully reconciled

    29 of 50 accounts on inherited 2023 settings

    5

    Feed exclusions are never pruned

    Median 34 saleable SKUs hidden per account

    6

    Target ROAS is set to a number someone made up in 2021

    18 of 50 accounts loss-making at current margins

    1. Performance Max is funding your low-margin SKUs

    On average, 43% of Performance Max spend was going to SKUs in the bottom two margin quintiles.

    The mechanism is simple. Performance Max optimises for whichever signal it can attribute most conversions to. Low-priced SKUs have higher conversion rates. Higher conversion rates tell PMax to bid more. More bidding means more spend. The SKU with 9% contribution margin ends up getting four times the spend of the SKU with 38% contribution margin, because it converts more frequently per click.

    This is not a bug in Performance Max. It's a feature that only pays off if you've told PMax which SKUs are actually worth prioritising. In 46 of 50 accounts, nobody had.

    What the waste looks like in your P&L

    Revenue grows, unit economics quietly deteriorate, and ROAS looks fine because you're still selling product. Contribution margin tells the real story, and it usually isn't being reported to the board until quarterly close.

    How to check

    Pull your last 90 days of PMax conversions by product. Rank products by spend share. Cross-reference with your margin data. If the top 20% of spend products don't overlap with your top 20% of margin-weighted revenue products, you have this problem.

    2. Branded search is hiding inside Performance Max

    In 37 of 50 accounts, Performance Max was serving against the brand's own trademark.

    Google does not let you fully exclude brand terms from Performance Max without submitting a brand list through your rep, and even then enforcement is patchy. The result: PMax gets credited with conversions that would have happened via organic search anyway, and your blended ROAS looks better than it should.

    On average, removing brand traffic from PMax ROAS calculations halved the apparent ROAS on those campaigns. Not 20%. Not 30%. Halved.

    Why it matters commercially: your CFO is approving budget increases on a lie. The incremental click-through-to-profit isn't coming from PMax discovering new customers. It's coming from Google charging you for people who would have typed your URL directly.

    How to check

    In Google Ads > Performance Max campaign > Insights > Search Terms (if available) or Asset Group Insights. Count brand mentions. If brand queries are more than 5% of impressions, you're subsidising Google for traffic you already owned.

    3. Broad match keywords live too long after they stop converting

    Across all 50 accounts, the median account had 14 active broad match keywords that had received spend in the last 30 days with zero conversions in the last 90. Median monthly spend on these keywords: £4,200 per account.

    Broad match is not inherently bad. It's the pruning that fails. Account managers review search term reports weekly in theory, and quarterly in practice. A broad match keyword that started out well can degrade as Google expands the match set. If nobody's looking, spend continues.

    We found keywords in three accounts that had consumed over £20k each in the previous 12 months while converting fewer than two times.

    How to check

    Filter your keyword report to broad match type, last 90 days, conversions < 2, cost > £500. Anything that appears here should be paused or renegotiated tomorrow morning.

    4. Smart Shopping migrations were never fully reconciled

    Google retired Smart Shopping in August 2023. Accounts were auto-migrated into Performance Max.

    In 29 of 50 accounts, the migrated PMax campaigns still had the original Smart Shopping structure: one catch-all asset group, no product feed segmentation, no SKU exclusions, no geographic tuning.

    These campaigns were running on 30-month-old settings. Product catalogues had turned over three or four times. Seasonal products were still enabled in January. Discontinued SKUs were still attracting clicks.

    The migration was technically complete. The campaign was operationally abandoned.

    How to check

    Look at each PMax campaign's creation date. If it's older than August 2023 and you (or the previous agency) haven't restructured it since, assume it's running on inherited assumptions that no longer match your catalogue.

    5. Feed exclusions are never pruned

    Every ecommerce account has product exclusions in its Merchant Center feed. Excluded because they're out of stock, discontinued, regional, seasonal, policy-flagged, or low-margin.

    In every single one of the 50 accounts, the exclusion list was longer than necessary. The median account had 118 permanently excluded SKUs, of which 34 were back in stock, restocked, or otherwise saleable.

    These exclusions were set during a crisis (stockout, policy issue, holiday lull) and never reviewed. The result: products that could be generating revenue are invisible to Shopping and Performance Max.

    This isn't technically waste, it's lost revenue. But the commercial effect is identical.

    How to check

    Cross-reference your Merchant Center exclusion list against your current stock file once a quarter. The pruning takes about 40 minutes. The uplift is immediate.

    6. Target ROAS is set to a number someone made up in 2021

    The bidding strategy on the largest campaign in each of the 50 accounts had been set once, during initial campaign build, and never revisited. The target ROAS in 41 of those 50 cases was a round number (4x, 5x, 6x) that had no relationship to the brand's current contribution margin.

    The problem with a fixed tROAS is that it ignores the most important variable in ecommerce: margin drift. Your product mix changes. Your input costs change. Your shipping costs change. Your return rate changes. Your fixed overheads change. A 4x ROAS target that was profitable in 2021 might be break-even in 2026 and loss-making by 2027.

    In 18 of the 50 accounts, the target ROAS was literally loss-making at current contribution margins. The campaigns were optimising toward a target that destroyed profit on every incremental conversion.

    How to check

    Take your campaign's target ROAS. Calculate your actual blended contribution margin on the products in that campaign. If your target ROAS times your margin doesn't comfortably exceed 100%, you're burning money on every sale.

    What the pattern looks like commercially

    Read those six patterns together and you get a predictable story.

    A brand signs with an agency (theirs or ours) three years ago. Campaigns are built to best practice of the day. Performance Max doesn't yet exist, or it's new. The account manager knows the catalogue. Over time:

    • The account manager changes, or their attention shifts to new client accounts.
    • Google releases Performance Max. Smart Shopping gets migrated over.
    • The catalogue turns over. Margins shift. Input costs rise.
    • The CFO approves a budget increase every year because ROAS stayed flat or improved.
    • Contribution margin slowly deteriorates, but nobody notices because reporting is in ROAS.
    • A recession or cash pinch forces a look at the P&L. The numbers don't add up.

    The audit finds £22k per quarter in recoverable waste. The agency fires back that "ROAS has never been higher." The CFO doesn't care. The conversation becomes adversarial.

    This is the conversation we are hired to prevent.

    What to do tomorrow morning

    You don't need to wait for an audit to run the first three checks. Each takes under an hour:

    Pull PMax product spend by SKU

    Cross-reference with your margin file. Flag any SKU in the top 20% of spend that's in the bottom 40% of margin.

    Filter PMax for brand impressions

    If brand queries are more than 5% of PMax impressions, open a ticket with your Google rep to enforce brand exclusion through a brand list.

    Filter keywords to broad match, 90-day cost > £500, 90-day conversions < 2

    Pause everything that returns.

    If you do those three things and recover meaningful spend, the rest of the audit will tell you exactly where the other £15-20k per quarter is hiding.

    If you'd rather we run the full diagnostic for you, our Commercial Audit takes 5-7 days, comes with a written report, and ends with a findings call where we walk you through the numbers.

    Book a Commercial Audit

    FAQs

    How do I know if my agency has found this waste and is actively fixing it?

    Ask them for a SKU-level spend report. If they can produce it within 24 hours and can explain the commercial logic behind their top 10 spending products, you're in good hands. If they send you a ROAS chart instead, they haven't.

    Is Performance Max a bad product?

    No. It's a powerful campaign type that defaults to optimising for the wrong variable (conversions) unless you tell it otherwise. Used with SKU-level constraints, margin-weighted targets, and brand exclusions, it performs excellently. Used as Google ships it, it quietly destroys margin.

    What if my margins are so tight that I can't be picky about which SKUs get spend?

    Then you almost certainly shouldn't be running Performance Max at all. Tight-margin catalogues need standard Shopping with manual CPC or Maximise Clicks, because they need absolute spend control. PMax's opacity is fine when margins have headroom. It's suicidal when they don't.

    How often should we audit?

    Once a quarter, as part of close. Anything less frequent and the patterns above take hold. If you have in-house Google Ads management, build a 30-minute internal audit into the quarterly close calendar. If you're with an agency, ask them to present their own audit findings at each QBR.

    Do these findings apply to Microsoft Ads and Meta as well?

    The PMax pattern is specific to Google. The others (broad match drift, branded traffic cannibalisation, stale targets, stale exclusions) repeat across every paid platform. We see them in Microsoft Ads accounts at roughly the same frequency.

    Who ran these audits?

    JudeLuxe, a specialist ecommerce Google Ads agency based in Birmingham, UK. The 50 audits referenced were conducted between April 2025 and March 2026 across UK, EU, US, and AU ecommerce brands spending £10k-£350k monthly on Google Ads.

    Methodology & sources

    Findings drawn from 50 commercial audits conducted by JudeLuxe between April 2025 and March 2026 across UK, EU, US and AU ecommerce brands spending £10k-£350k per month on Google Ads. All figures are medians or proportions across the dataset, anonymised at SKU and account level. Each pattern below is also covered in greater depth in a dedicated post:

    Want the full methodology and the template we use for commercial audits? Download the CFO Board Pack — one-page summary of the commercial case, timeline, and objection responses.

    Chris Avery is Founder and Managing Director of JudeLuxe. He writes about profit-led paid media strategy for ecommerce brands.