Ecommerce Growth Audit: What to Expect and How to Run One

A commercial diagnosis of your whole growth system, not another channel report, and how to tell a real audit from a screenshot dump.

Updated on: 2026-09-02

Most growth audits I get handed to review make the same mistake in the first ten pages. They open with a channel. "Meta ROAS is down." "Search is underperforming." "The site converts below benchmark." Then they spend forty slides fixing whatever the person writing the audit happens to be good at.

That is not a growth audit. That is a specialist looking at the business through their own tool.

A proper ecommerce growth audit does not start by asking which channel is broken. It starts by asking whether the numbers can be trusted, whether the growth you already have is profitable, and which single constraint is holding the business back. Economics, measurement, acquisition, conversion, retention, or execution. Only one of those is usually the binding problem at any given time. The rest are symptoms or distractions.

If you run a brand somewhere between $2M and $20M in revenue, with a team, a budget, and a plateau nobody can quite explain, this is the diagnostic you actually need.

What a growth audit is, and what it is not

A growth audit is a commercial diagnosis of the entire system that produces revenue. It sits above channel work. It should tell you where profitable growth is stuck and what to do about it, in priority order, with owners attached.

Here is how it differs from the narrower audits most agencies sell:

The last row is the one that matters. A growth audit can point you at the constraint. It cannot prove causation on its own. That comes later, through testing. Anyone who promises certainty before running a single experiment is selling confidence, not diagnosis.

What a properly run audit produces

The output of a real audit is a decision system, not a backlog. If you finish reading it and you still do not know what to do next week, it failed.

At minimum, expect these deliverables:

An executive diagnosis. One primary constraint. The evidence behind it. What was ruled out. And the commercial cost of leaving it alone. That last part gets skipped constantly, and it is the part that gets budget approved.

A measurement-confidence assessment. Which numbers come from the commerce platform, which from finance, which from GA4, which from ad platforms, and which are modeled estimates. Where they disagree, and by how much. A good auditor explains discrepancies instead of quietly picking whichever platform looks best.

A funnel and customer analysis. Sessions, product views, add-to-cart, checkout starts, payment completion, conversion rate, AOV, refunds, discounts, repeat purchase rate. Segmented, not blended into one flattering average.

A channel assessment by role, including how much each channel actually contributes versus what the platform claims, and how much of that is new customers versus people who were going to buy anyway.

A prioritized action plan with fewer than ten high-confidence moves, each with an owner, expected effect, effort, and a way to measure whether it worked.

An operating rhythm. A review cadence, agreed definitions, one commercial target, and clear owners for media, creative, site, retention, analytics, and finance.

A report made of screenshots and generic best practices is not any of this.

Run it in this sequence

The order matters more than people expect. Skip the early phases and you end up optimizing numbers that were never real.

Phase one: define the commercial question

Before opening a single dashboard, write down the decision the audit has to inform. Not "audit everything." Something you can answer.

• Can we increase paid acquisition by 30% without cutting contribution margin?
• Why has revenue grown while cash generation has weakened?
• Is conversion, traffic quality, creative supply, or retention the real ceiling?
• Should we keep or replace the agency?

A broad scope is fine. A scope with no decision at the end is not.

Phase two: establish whether the numbers are trustworthy

This is where most audits should spend their first week and almost none do. Build a source-of-truth table.

Then check the plumbing. Does the purchase event fire once or twice? Are transaction IDs present? Are refunds imported? Are tax, shipping, and discounts handled consistently? Does the date range hide a site migration, a stockout, or a big promotion?

Google's ecommerce measurement guidance recommends tracking the full shopping sequence including item views, add-to-cart, checkout, purchase, and refunds, and validating the purchase event in DebugView before trusting reports. Its set-up-a-purchase-event documentation also notes purchase data can take around 24 hours to appear. Enhanced conversions, per Google's own documentation, improve matching by sending hashed first-party data, but they improve modeled measurement rather than revealing the true cause of every sale.

Do not benchmark an unreliable conversion rate. Fix the measurement or flag it as low-confidence first.

Phase three: analyze the economics

Now that the numbers mean something, work out the allowable acquisition cost from real margin and customer value. A few distinctions that keep audits honest:

• Revenue ROAS vs contribution ROAS. The first is what platforms report. The second is what your bank account cares about.
• Blended CAC vs new-customer CAC. If you are only tracking blended, you are probably flattering yourself.
• Payback period. How long contribution profit takes to recover acquisition cost.

The diagnostic patterns I keep seeing:

• Revenue rises, contribution profit falls. Look at discounting, returns, fulfilment, and product mix.
• ROAS improves while new-customer share drops. You are harvesting existing demand and calling it growth.
• CAC looks stable but payback worsens. Usually lower AOV, slower reorders, or margin compression.
• Email revenue climbs while total revenue is flat. Attribution is reallocating credit, not creating demand.

The real buying question is not "can this person improve ROAS." It is "can this person tell me the point at which more growth stops being profitable." That number is the entire game.

Phase four: inspect acquisition by channel role

Judge each channel by what it is supposed to do, not by a single ROAS figure compared across platforms that all count conversions differently.

• Demand capture: Google Search, Shopping, marketplaces.
• Demand creation: Meta, TikTok, Pinterest, creators.
• Owned demand: email, SMS, loyalty.
• Organic discovery: SEO, content, product search.

For each, look at spend, new-customer percentage, CAC, contribution margin, creative volume and fatigue, audience overlap, and any incrementality evidence. GA4 now offers data-driven, paid-and-organic last click, and Google-paid last click models. The older first-click, linear, and position-based models are gone. Data-driven attribution is a model built on your account's converting and non-converting paths. It is a useful estimate, not a recording of what caused each sale.

Phase five: walk the customer journey

From ad click to second purchase, look for the breaks. Message continuity from ad to landing page. Total cost shown early. Shipping timing. Returns clarity. Mobile usability. Guest checkout. Post-purchase education.

Baymard's checkout research puts average cart abandonment at 70.19%, with checkout design and difficulty as recurring causes, and reports that 23% of US shoppers abandoned an order because they were not shown the total cost upfront. Directional evidence, not proof that checkout is your biggest constraint. Your own funnel data decides that.

Phase six: technical performance and SEO

Keep this commercial. A list of 200 warnings helps nobody. Check mobile performance on key templates, product indexability, feed quality, structured data, and price consistency between site and Merchant Center.

Core Web Vitals targets at the 75th percentile are LCP at or under 2.5 seconds, INP at or under 200 milliseconds, and CLS at or under 0.1, measured on real users and split by device. Google also distinguishes product snippets from merchant listings, the latter reserved for pages where customers can actually buy.

Phase seven: retention and lifecycle

Look at cohorts, not open rates. What share of first-time buyers place a second order, and how long does it take? Does discounting bring in customers who never return? Are replenishment reminders timed to real consumption?

Definitions matter here because people use them loosely. Shopify defines repeat purchase rate as customers with more than one purchase divided by total customers, and retention as (end customers minus new customers) divided by starting customers. State which one you are using and over what window, or the number is meaningless.

Phase eight: team, agency, and execution

A correct diagnosis still fails through execution. So check who owns the growth number, who has decision rights, whether the agency optimizes to platform metrics or to profit, and whether creative and development capacity actually exists.

Ask each responsible person the same four questions: What is the current growth target? What is the biggest constraint? What changed last month? What would make you stop or scale this activity? Conflicting answers point to an operating-model problem, not a bad hire.

How to prioritize what you find

The audit should end with fewer than ten top-priority actions, scored on impact, evidence strength, effort, and reversibility.

A 50-item backlog with no sequencing is a data dump. If everything is a priority, nothing is.

Choosing who runs it

The strongest auditor is not the one with the longest checklist or the deepest single-channel specialty. It is the one who can connect reliable measurement to real economics, isolate the binding constraint, and hand you a small, owned, testable plan.

Be wary of a provider who promises a specific revenue lift before seeing your data, treats platform ROAS as profit, gives recommendations without evidence, audits one channel while calling it a growth audit, or recommends firing your agency without first reviewing that agency's brief, budget, and commercial target.

This is roughly where Miguel Casteleiro's work fits. He targets established brands in the $2M to $20M range where there is traction, a team, and real spend, but growth has become hard to direct. His approach covers commercial targets, attribution, paid acquisition, conversion, retention, and agency capability, working above execution rather than replacing capable teams. Two formats exist: a three-month advisory where your existing team and agencies stay in place while the constraints get diagnosed and a review rhythm gets built, and a fractional Head of Growth arrangement where he takes direct responsibility for goals, budgets, and accountability.

One honest note. The site does not sell a fixed, packaged "growth audit" with public pricing. The diagnostic work lives inside those engagements, and capacity is capped at five companies at a time. Track-record claims (seven years leading growth at a performance agency, 130 brands, companies from roughly $1.5M to over $250M) are self-reported. If you approach this as an audit purchase, ask for the exact scope, data access, deliverables, timeline, fee, and definition of success in writing first. That is true of any provider, not just this one.

FAQ

How long does a proper ecommerce growth audit take?

Usually a few weeks, not a few days. The data-reconciliation phase alone can take a week if finance, GA4, the ecommerce platform, and the ad accounts disagree, which they almost always do. Anyone offering a same-week audit is skipping the part that makes the rest trustworthy.

Do I need to fix my tracking before an audit?

No, and you should be skeptical of anyone who says otherwise. Assessing your measurement is part of the audit. What you should not do is act on conversion rates or ROAS figures before knowing whether they are reliable. Fix or flag the measurement, then decide.

Should the audit recommend replacing my agency?

Only after evaluating the agency's actual brief, budget, creative inputs, and commercial target. Plenty of "underperforming" agencies are executing a vague brief against an unrealistic target with no clear owner. Replacing them changes nothing if the operating model is the real problem.

Are conversion-rate benchmarks useful?

For generating questions, yes. As targets, no. Shopify's June 2026 category figures run from around 0.63% for luxury and jewelry to 5.70% for pet care, with a cited global average near 2.66%. Those vary by category, geography, traffic intent, device, and price point. Compare against your own trend first, segmented by customer type and channel, before you compare against anyone else.

What is the single most important thing an audit should tell me?

The point at which additional acquisition spend stops being profitable. Everything else is downstream of that number.