E-commerce ads audit checklist: what I check in 7 working days

Why a Google Ads or Meta Ads audit alone is not enough

Most audits stop at the ad account. Useful, but the money often leaks somewhere else.

Take Museo de la Imaginación, an illusion museum in Málaga (September–October 2024). From Meta ads, 608 visits reached ticket selection and only 8 orders followed. The drop was between choosing a ticket and paying. That was a website problem. No bid change would have fixed it.

So I audit the full funnel, from the first impression to the repeat purchase, in the order below. You can run through it yourself.

1. Tracking and data quality: can you trust the numbers?

Start here. Google and Meta optimise towards the purchases you send them. If that data is wrong, every other finding in your PPC audit is built on sand.

The gaps can be big. At the museum, the Meta pixel counted 25 purchases in a month, web analytics 8 orders from Meta: more than three times apart for the same month. Decide which number you trust first. What this looks like in an online store is in my article on break-even ROAS.

Before me, the museum ran a Google smart campaign with no tracking. With analytics rebuilt from scratch, the first two weeks of October 2024 showed 16 sales worth €420 on €90.61 of Google Ads.

• Each purchase is counted once. With both the Meta pixel and the Conversions API (CAPI), the same event needs the same event_id; Meta's documentation says it deduplicates within 48 hours.
• In Google Ads the primary conversion is the purchase, not add-to-cart or a page view. For sales, Google recommends counting every conversion.
• Order value and currency are passed with each purchase. Without them, value-based bidding cannot tell a €30 order from a €300 one.
• Platform numbers match store orders for the same dates, or the gap is explained.
• GA4 sees every step from product page to payment, and your consent setup does not cut more data than it has to (check the legal side with a lawyer).

2. Account structure: where the budget really goes

Graft Collab sells premium wine glassware in Dubai (March–June 2025). In March, Google Ads spent AED 8,418 for AED 2,005 in sales: 0.24 per dirham. In May, Performance Max started pulling junk traffic: AED 3,863 spent, AED 2,510 in sales. We did not feed it more money. We narrowed it to one brand (Riedel) and one city (Dubai). In June Google Ads reported AED 14,120 in sales on AED 2,663, a return of 5.3. The order sheet shows 3.6, and I report both.

• Campaign objectives match money: if you expect orders, optimise for sales, not reach or traffic.
• Paused and removed campaigns and the change history are reviewed too. In Google Ads, removed campaigns can be hidden by the status filter, so check the filter first: telling mistakes often sit there.
• Search terms and placements: what you pay for without sales, especially in Performance Max and broad match.
• Prospecting and retargeting are separate and do not overlap.
• The product feed in Google Merchant Center: no disapproved products, clear titles, correct prices and stock. At Villa Cotton, a US home-textile brand, cleaning up Merchant Center and reworking product titles and descriptions was a separate part of the work.
• Breakdowns by country, device and product: an average campaign often hides both money-burners and winners.

3. Budget vs margin: is the advertising actually profitable?

A ROAS of 4 on its own tells you little. ROAS is ad revenue divided by ad spend. It says nothing about profit until you know your margin.

The quick maths: break-even ROAS = 1 ÷ your margin per order, after product cost, shipping, payment fees and returns. At a 40% margin, break-even ROAS is 2.5. That is an example; your number will be different. Anything below it loses money, however nice the report looks.

• Margin is calculated by product category, not as a store-wide average.
• Each campaign has a threshold: above it you scale, below it you fix or pause.
• You track blended numbers too: total ad spend against total store revenue, not only the return inside each platform.

4. Creatives and offer

With a weak offer, even a tidy account ends up buying expensive clicks.

Villa Cotton ran a weekly cycle: a new promotion (a category, a collection or a US holiday), fresh Meta and Google creatives and a newsletter around it.

• How often new creatives go live, and whether frequency shows the same people seeing the same ads.
• The promise in the ad matches what people see after the click.
• There is a reason to buy now: a promotion, a season, a limited batch, a clear returns policy.

5. Landing pages, site speed and checkout

This is where the problem often hides. At the museum, analytics showed people picking tickets and leaving. A one-or-two-section website did not convey the place. The next step was a new website, not new campaigns.

• Product pages answer the buyer's questions: price, shipping, delivery time, sizes, returns.
• You have gone from ad to payment yourself, on a phone.
• Product pages and cart are tested in PageSpeed Insights, not only the homepage.
• Checkout: how many steps, whether an account is required, when shipping costs appear.

6. Repeat sales and email

Ads buy the first order. The second order from the same person should cost almost nothing. Without email flows, you keep paying for the same customer.

At Villa Cotton, email and SMS in Klaviyo cost 4.7% of the sales they brought in July 2025, against 65% for Meta in the same month.

• Abandoned cart and abandoned browse flows.
• Post-purchase emails and cross-sells.
• The share of revenue from returning customers, calculated at least once.

7. Reporting in revenue and margin

I do not believe in pretty reports. If a report has no orders, revenue and margin, you cannot make decisions with it.

When I took over Museo de la Imaginación, there was not a single report on sales from ads. At Graft Collab, a summary went out every week: spend, clicks, orders, order value, cost per order and payback. Decisions were made on orders, not clicks.

• Weekly: spend, orders, order value, cost per order and payback for each channel.
• Overall ad cost share: all ad spend against store revenue.
• Each report ends with a decision: what to scale, what to pause, what to test.

What I check in my 7-day e-commerce audit

Every project starts with a diagnostic: for the first 7 working days I go through the funnel above on your real data. Your ad accounts and data stay yours.

• Goals, margin, average order value, store orders against ad platforms and analytics.
• Ad accounts (Google Ads, Meta and others), product feed, search terms, creatives.
• Site and checkout, offer, repeat sales and email.
• At the end: the loss map with priorities — where the money leaks and how much.

What you get after the 7 days

You get a map of where money leaks, how much and what to fix first. Then the work moves to growing revenue along that map. I run every project myself, no more than three at a time.

I work in writing and with recorded video walkthroughs instead of calls. The contract runs 3 months, then month to month. If the diagnostic does not convince you, tell me within 3 days and I refund the month's fee.