Home › Resources › Insights › Operations › Why audits find the wrong problem
Doubling traffic into a 0.8% conversion rate doubles your acquisition cost and teaches you nothing. This is the twelve-point sequence we run on every new account — what each point measures, what clears it, and the three constraints the audit surfaces most often.
The Dazzle Commerce Team
eCommerce Management & Marketing
Updated
5 September 2026
Reading time
7 min read
Every engagement opens with a diagnosis, not a media plan.
Every new account we take on starts the same way. Nobody touches a bid, a listing, a template or a line of ad copy until we can answer one question in a sentence: what is the binding constraint on this business right now?
It is a deliberately narrow question. Most stores have twenty things that could be improved and two that are genuinely holding revenue back. The audit exists to tell those apart, because the cost of working on the wrong one is not zero — it is a quarter.
KEY TAKEAWAY
An audit is not a list of everything that is wrong. It is a ranked answer to one question: which single constraint, if removed, releases the most revenue for the least spend?
Checklists are easy to find. The order they run in is what makes them useful. Each point in this audit assumes the ones above it have cleared, because a finding at point nine is meaningless if point one is broken.
If your analytics double-counts returning customers, every conversion rate, every channel ROAS and every cohort chart downstream of it is wrong — and you will spend three months optimising toward a number that does not exist. Measurement integrity comes first for that reason alone, not because it is interesting.
The same logic applies further down. There is little point auditing creative volume for a brand whose contribution margin after returns and carriage is four per cent. The constraint is the margin model, and more creative will only help it lose money faster.
Each point has a threshold. It either clears, or it becomes a candidate for the binding constraint. We deliberately do not score them out of ten — a scored audit lets everyone feel moderately good about a store that is failing badly on one thing.
Audit point
Layer
What clears it
1. Measurement integrity
Foundations
Store, analytics and ad-platform revenue agree within 5%
2. Margin model
Foundations
Contribution margin known per SKU after returns, carriage and fees
3. Traffic composition
Demand
No single channel above 60% of sessions; branded and non-branded split known
4. Conversion funnel
Demand
Session-to-cart, cart-to-checkout and checkout-to-order tracked separately
5. Product page quality
Conversion
Top 20 SKUs by revenue reviewed against a written page standard
6. Catalogue and feed health
Conversion
Under 2% feed disapprovals; titles, identifiers and attributes complete
7. Marketplace account health
Channels
Account health, buy-box share and suppression all inside policy
8. Ad account structure
Channels
Campaigns segmented by intent and margin, not by product folder
9. Creative volume
Channels
Enough new assets each month to refresh the top ad set before fatigue
10. Lifecycle coverage
Retention
Welcome, browse, cart, post-purchase and winback flows all live
11. Technical performance
Foundations
Mobile LCP under 2.5s on template pages, not just the home page
12. Competitive position
Demand
Price, delivery promise and review volume benchmarked against three rivals
Thresholds are starting points, not universal truths — a 40-SKU brand and a 40,000-SKU distributor clear point six very differently.
The checks group into four layers, and the layers are what we actually report on. A store rarely has one broken check. It has one weak layer, and the checks inside it fail together.
1
Foundations — measurement, margin and technical performance. Nothing above this layer can be trusted while it is broken, and almost none of it is visible in a dashboard.
2
Demand — where traffic comes from, what it costs and how concentrated it is. This is where fragility hides: a store doing well on one channel is one policy change away from a bad quarter.
3
Conversion — product pages, catalogue data and the checkout path. The cheapest revenue in most accounts sits here, because releasing it does not require buying anything.
4
Retention — lifecycle coverage and repeat rate. It is rarely the binding constraint in month one, but it sets the ceiling on what acquisition is allowed to pay.
The audit is a working session rather than a document handover: findings are argued through with the client team before anything is ranked.
We have run this enough times to have a boringly consistent answer. Roughly seven in ten audits surface the same three culprits.
of audits flag the same three issues
Attribution, product pages, marketplace catalogues
constraints, ranked
Not a backlog of forty recommendations
typical catalogue neglect
Since anyone last owned marketplace listing data
Figures reflect Dazzle Commerce engagements and are indicative of what we see, not published industry benchmarks.
The first is attribution that cannot support a spending decision — not because the setup is exotic, but because nobody has reconciled platform revenue against the store’s own numbers since tracking was installed. The second is product pages that were never rebuilt after the store scaled: templates designed for forty SKUs, still carrying four thousand. The third is marketplace catalogues nobody has owned in eighteen months, quietly losing the buy box on the listings that used to carry the account.
None of them are glamorous. All of them are cheaper to fix than buying more traffic.
EXPERT VIEW
“The most expensive number in eCommerce is a ROAS figure nobody has reconciled. It gives a team enough confidence to keep spending and not enough information to know whether the spending is working.”
Dazzle Commerce · eCommerce Management team
The full framework — every check, every threshold and where each threshold comes from — is published on our eCommerce Growth Assessment. This article is the argument for running it in a fixed order; that page is the framework itself.
Most of it does not need an agency. With a spreadsheet, admin access and a fortnight of evenings you can get to about eighty per cent of the value. The sequence below is the one we use.
1
Reconcile three revenue figures for the same 30-day window: store admin, analytics, and the sum of your ad platforms. Write the variance down before you explain it away.
2
Build a contribution margin column for your top 50 SKUs by revenue, deducting returns, carriage, payment fees and platform commission. Sort ascending and read the top of that list slowly.
3
Split sessions by channel, and search by branded versus non-branded. Note any channel above 60%.
4
Walk the funnel on your own site, on a phone, buying a real product with a real card. Time each step.
5
Open the twenty highest-revenue product pages side by side and score each against your own written standard rather than against a competitor.
6
Pull marketplace account health and feed diagnostics. Count disapprovals and suppressed listings before you look at anything else.
At the end you will have a short list. The point is not the list — it is being able to say, in one sentence, which item on it is holding the others back.
An audit that produces a forty-item backlog has failed. The output should be three things, in this order:
BEFORE YOU COMMISSION ANOTHER AUDIT
If your last audit produced a document rather than a decision, it was a report. The test is simple: could the person who commissioned it say what to stop doing on Monday?
Three services that pick up the constraints this audit tends to surface.
The engagement model this audit opens: one team owning storefront, channels, catalogue and reporting.
The testing programme behind points four to six, from funnel instrumentation to product page standards.
Account health, catalogue ownership and buy-box recovery for points six and seven.
WRITTEN BY
eCommerce Management & Marketing
Dazzle Commerce runs storefronts, marketplaces, paid media, SEO and retention for eCommerce brands across ten markets. This article reflects the audit process we open every management engagement with, written by the team that runs it.
More from the operations and conversion side of the practice.
ROAS ignores returns, carriage and platform fees. How to rebuild the report around contribution margin.
When an in-house team is the right answer, when an agency is, and when neither will fix the problem.
The elements that reliably move conversion, the four that rarely do, and how to tell them apart.
We open every management engagement with this audit, and you keep the findings whether or not we work together afterwards.
