Growth initiatives often start with a single measure: more Amazon Ads, a Shopify relaunch, internationalization or AI automation. The essential groundwork gets skipped: is the problem even properly understood? Can the product range serve additional demand profitably? Are data and responsibilities reliable?
The following 20 questions form a compact audit framework. Every answer should be backed by evidence, not just opinions: reports, unit calculations, process descriptions, account data and clearly named owners.
A. Product range and product
1. Which products generate revenue, and which generate contribution margin?
An ABC analysis by revenue is not enough. Add contribution margin, capital tied up, returns and operational complexity. Bestsellers can be unprofitable; small products can carry the healthiest growth.
2. Which specific customer problem does each core product solve?
The answer must be understandable in one sentence and show up in search terms, reviews and purchase arguments. Unclear positioning makes content and advertising more expensive.
3. Where can the differentiation be proven?
Check material, function, set composition, design, service, brand and price. "Better quality" only holds up once it is visible, explainable or measurable.
4. Which products should be expanded, reworked or discontinued?
Define clear thresholds for expansion, relaunch and delisting. These include margin, demand, reviews, replacement risk and strategic role in the product range.
B. Margin and unit economics
5. Is the full unit calculation per channel up to date?
Include purchasing, freight, customs, packaging, payment fees, marketplace costs, fulfillment, returns, discounts and advertising. Averages hide differences between ASINs, countries and variants.
6. What is the contribution margin before and after marketing?
Only this shows how much a product can carry for new customer acquisition or scaling. A target ROAS with no link to margin is not profitability management.
7. Which price limits apply, and who may change them?
Define regular price, promotional price, minimum price and the approval process. Uncoordinated discounts can damage margin, Buy Box and brand perception all at once.
8. Which costs do not grow linearly with revenue?
Support, returns, quality problems, storage, pre-financing and team effort can rise disproportionately with scale. An audit makes these bottlenecks visible before more demand is generated.
C. Channels and demand
9. What role does each channel have?
Amazon, Shopify, retail and B2B do not have to do the same job. For each channel, define target group, role in the product range, pricing logic, service promise and economic goal.
10. Where is demand created, and where is it only closed?
Separate demand creation from demand capture. Branded search, organic rankings, social creatives and marketplace advertising contribute differently to the customer journey.
11. Which acquisition measures are incremental?
Not every attributed sale would have been lost without an ad. Look at tests, brand and non-brand shares, new-customer metrics and the development of total revenue instead of platform attribution alone.
12. Is the sales surface "retail ready"?
Before adding traffic, price, availability, main image, reviews, product information, variants and mobile display must be right. Otherwise you are buying visitors for an unclear detail page.
D. Data and steering
13. Which five metrics does the management team really steer by?
A good selection combines revenue, contribution margin, new customers, inventory and cash. Every team member should use the same definitions and the same reporting period.
14. Are metric definitions consistent across channels?
Revenue gross or net? Returns in the order month or the refund month? Ad revenue by which attribution window? Documented definitions prevent false precision.
15. Which data is missing for important decisions?
Do not list all conceivable data, but the decisions that are blocked today. Then clarify which minimum dataset reduces the uncertainty.
16. Where are there automatic alerts instead of late reports?
Inventory risks, budget outages, price deviations, conversion drops and Account Health problems should be noticed early. Automation is often worth more here than another dashboard.
E. Processes, systems and responsibility
17. Which recurring processes take up a disproportionate amount of time?
Measure frequency, minutes per run, error rate and required approvals. This baseline shows whether standardization, delegation or AI automation makes sense.
18. Is there a clear owner for critical processes?
"The team" is not an owner. Price, inventory, campaigns, listings, product data and compliance each need a named person with decision authority.
19. Which systems are the source of truth?
Product data, costs, inventory and tasks must not live in conflicting spreadsheets. Define one leading system per data type and a traceable change process.
20. Which initiative takes priority in the next 90 days, and what is deliberately not done?
Prioritizing means giving things up. Rate initiatives by impact on results, effort, risk, dependencies and speed of learning. The chosen initiative gets an owner, budget, target value and review date.
From audit to 90-day roadmap
After the audit, findings are sorted by impact, not by department. A pragmatic model uses four categories:
- Secure immediately: risks to availability, compliance, margin or Account Health.
- Quick wins: high impact, low effort, few dependencies.
- Structural levers: data, processes, product range or systems with a medium time horizon.
- Parking lot: interesting ideas without a current business case or the necessary prerequisites.
Every active measure needs six fields: problem, baseline, target value, owner, next concrete step and review date. That turns an inventory into a steering instrument.
Where AI helps in the audit
AI can structure data and documents, flag recurring deviations, cluster customer feedback and prepare first hypotheses. Assessing margin, risk and priority, however, remains a business decision.
For a clean pilot, permitted data, quality criteria, test cases and human approvals are defined in advance. Teams that use AI systems also need adequate AI literacy. That is exactly why training and governance belong in the roadmap, not as an appendix after the technical build.
How to run the audit in practice
Twenty questions sound like a lot of work. In practice it takes two to three days if the approach is structured, and most of that goes into gathering numbers, not into thinking.
Preparation, about half a day. Collect the data exports before answering the first question: revenue per product and channel for the last twelve months, current unit calculations, ad costs per channel, return rates and the list of active products. If one of these sources is missing, that is already a result, and often the most important one.
Solo run-through, about one day. Answer the questions in writing first, without discussion. Where you have no reliable answer, write "unknown" instead of a guess. In the end, the list of unknowns is just as valuable as the answers.
Joint review, half a day. Go through the answers with two or three people who are close to operations. In our experience, assessments differ significantly on a third of the questions, and these differences show exactly where the company has no shared picture.
Prioritization, a few hours. Rate each identified action item by expected impact and required effort, each on a scale from 1 to 5. Everything with high impact and low effort belongs in the first 30 days.
A note on sequence: deliberately start the audit with product range and margin, not with channels and advertising. The reason is practical: if the unit economics are unclear, channel performance cannot be discussed meaningfully, because the yardstick is missing. Conversely, reliable contribution margins answer some of the later questions on their own. If you start with the channels, you almost always end up in a discussion about ad budgets that cannot be decided without margin data.
Five answer patterns that almost always point to a problem
Regardless of industry and size, the same answer patterns keep appearing in audits. If you recognize one of them in your own business, it is worth a closer look:
- "Only [one person] knows that." Critical knowledge sits in one head. That is not a staffing risk but an operational one, and it only becomes obvious when the person is absent.
- "The number comes from different systems, depending." If the same metric can have different values, you are not steering by it, you are arguing about it.
- "We have done it this way for years." Processes that were never questioned are the most reliable source of saved time, precisely because they are inconspicuous.
- "The product just sells." Revenue without a known contribution margin is an assumption, not an insight. Often the top-selling items are the ones with the weakest margins.
- "We haven't had time for that yet." If this sentence repeats for the same task over several quarters, either it is not important enough, in which case it should come off the list, or clear responsibility is missing.
None of these patterns is dramatic on its own. If several occur together, it is a reliable sign that the next growth initiative would be built on a foundation that cannot support it.
What should happen in the first 30 days after the audit
The value of an audit comes not from the analysis but from what happens in the four weeks after it. A tight, unspectacular routine has proven itself.
Week 1: decide, don't discuss. Choose no more than three initiatives from the prioritized list. Three is not an arbitrary number: with more parallel projects, completion rates in teams under ten people drop significantly, in our experience. Everything else goes onto a visible waiting list, not into the wastebasket, but not into current work either.
Week 2: set responsibility and a measuring point. Each of the three initiatives gets a named owner, an end date and exactly one metric by which success will show. If one of these three is missing, the initiative is not ready to start.
Weeks 3 and 4: implement and make obstacles visible. A short weekly meeting of 20 minutes is enough if it asks the right question: not "What's the status?" but "What is blocking right now?" Blockers that come up twice in a row belong at the level that can resolve them.
After 30 days comes an honest interim review. If an initiative has not moved, the cause is almost always one of three: it was cut too large, the owner had no real capacity, or a decision was missing that only management can make. All three are solvable, but only if they are named instead of letting the initiative quietly continue.
Conclusion
An e-commerce audit is not meant to prove how complex the company is. It should make visible the few bottlenecks that limit the next growth. If you examine product range, unit economics, demand, data and responsibility together, you invest more precisely and recognize early which initiative is not yet ready to start.