The Ecommerce Profit Leak Audit

• True contribution margin calculated AFTER COGS, inbound freight, pick-pack-ship, payment fees, and returns (not gross margin)
• Blended margin recalculated per product tier, not one company-wide average (your hero SKU and your loss-leader have different math)
• Discount leakage measured: average order discount as a percent of revenue over the last 90 days
• Return and refund rate pulled from actuals, with the COGS you do not recover on returns counted as a loss
• Shipping subsidy quantified: what you charge for shipping minus what it actually costs you, per order
• Payment and platform fees confirmed against the processor statement, not assumed (Shopify Payments, PayPal, BNPL, and Amazon Pay all differ)
• Free-shipping threshold checked against AOV so it lifts basket size instead of just eroding margin
• Break-even ROAS for the business recalculated from real contribution margin and posted where the media buyer can see it

• Branded search spend isolated and discounted for cannibalisation (how much would have converted organically anyway)
• Brand terms excluded from PMax and non-brand campaigns so they don't quietly absorb cheap branded conversions and inflate ROAS
• Search terms report swept for the last 90 days; irrelevant and broad-match junk added as negatives
• PMax channel distribution checked (is it just running cheap Display and Gmail placements and claiming the credit?)
• Shopping/PMax spend concentration checked: is 80% of spend on 20% of SKUs, and are those the high-margin ones?
• Duplicate or overlapping campaigns bidding against each other identified and consolidated
• Device, geo, and time-of-day performance reviewed for segments burning spend with no conversions
• Bid strategy targets set above true break-even ROAS, not above the inflated dashboard ROAS
• Asset coverage and ad strength checked so you are not losing impression share to thin creative

• Conversion tracking reconciled: Google Ads conversions vs GA4 vs the actual order count in Shopify for the same window (they should be within ~5-10%)
• Server-side tracking (Conversions API / server-side GTM) live, not just the browser pixel, so iOS and ad-blockers don't erase 20-40% of conversions
• Consent Mode v2 configured so modelled conversions recover what the cookie banner blocks
• Enhanced conversions for leads/purchases enabled and verified as receiving hashed data
• Conversion value passing actual order value (and ideally margin), not a static placeholder
• Deduplication confirmed so one purchase isn't counted by both pixel and server
• No test or internal traffic polluting conversion data
• Attribution window and model documented so month-over-month comparisons are apples to apples

• Zero unresolved disapprovals or account-level warnings in Merchant Center
• GTIN, brand, and identifier-exists set correctly so Shopping doesn't suppress products
• Titles front-load the attributes shoppers search (brand + product type + key spec) within the first 70 characters
• Product types and Google product categories fully populated for better matching
• Sale price and price accuracy synced so 'price mismatch' suspensions never trigger
• Custom labels in place for margin tier, bestseller, seasonality, and stock so you can bid by profitability
• Out-of-stock and low-inventory products excluded or down-weighted, not wasting spend
• Product ratings/reviews feed connected so stars show in Shopping (a free CTR lift)
• Image quality and supplemental feed checked for the SKUs that drive the most spend

• Largest Contentful Paint under 2.5s on mobile for the top landing pages (every second of delay cuts conversion)
• Product page has trust signals above the fold: reviews, guarantee, shipping/returns clarity
• Cart and checkout abandonment rates pulled; one-page or accelerated checkout enabled
• Mobile checkout tested end to end on a real phone, including wallet payments
• Email/SMS capture live with a welcome flow and an abandoned-cart flow actually sending
• Post-purchase flow drives the second order (the cheapest revenue you will ever get)
• Subscription or replenishment offered where the product justifies it
• Site search and out-of-stock handling don't dead-end paid traffic

The nine most common leaks and what they typically cost

Ranges from Ad-Lab audits across 350+ ecom accounts. Stack two or three of these and a 'profitable' account is quietly losing money.

How to run it

1. 01

   Pull the actuals first

   P&L, processor statement, refund report, and a 90-day Google Ads + GA4 + Shopify export. The audit is only as honest as the data you feed it.

2. 02

   Work top to bottom, score each line

   Mark every item pass / fail / unsure. Unsure counts as fail until proven. Tally fails per section.

3. 03

   Quantify the top three leaks

   Use the table to put a dollar range on your worst three. That is your prioritised fix list and usually pays for a quarter of agency fees on its own.

External resources

Authoritative references we link to alongside the template. Read them before running the audit.

• Google Ads, conversion lift and incrementalityMeasure causal lift instead of trusting attributed conversions.
• Google Merchant Center, product data specificationThe authoritative spec for feed attributes that affect serving.