The Ecommerce Data Cheat Sheet: Key Metrics, Where They Live, and Why They Never Agree

Ecommerce teams usually have no shortage of metrics.

What they have is a shortage of agreement.

Shopify says one thing. Meta says another. Google Ads says it drove everything. Finance has a number nobody in growth recognizes. Then somebody asks for a quick answer in a Monday meeting, and everyone ends up defending their own dashboard instead of making a decision.

That is why a cheat sheet like this matters.

Not because the formulas are mysterious. Because the same metric gets calculated for different purposes, in different systems, with different exclusions, windows, and assumptions. If you do not know which version you are looking at, you end up arguing about a number that was never designed to answer your question in the first place.

The Operating Rule Before We Start

Here is the rule I wish more ecommerce teams adopted:

A metric is only useful when you know all three of these things:

1. where the number lives
2. what the system is actually counting
3. which decision that version is fit to support

Without that, “revenue” is not a metric. It is a fight waiting to happen.

The Ecommerce Metrics Cheat Sheet

Ecommerce Metrics Compared

Here is the quick-reference version before we unpack each metric in detail:

The cost of getting these definitions wrong is not small. NRF and Happy Returns’ 2024 Consumer Returns in the Retail Industry report says retailers expect 16.9% of annual sales to be returned in 2024, which is one reason ecommerce teams cannot afford to treat gross revenue and profitability as interchangeable.

1. Revenue

Where it usually lives

• Shopify or your commerce platform
• ERP / finance system
• ad platforms with conversion values
• warehouse models and executive dashboards

Why the numbers disagree

Revenue is the easiest metric to say and one of the easiest to misuse.

The conflicts usually come from:

• gross vs. net revenue — Shopify may show gross order value before refunds, while finance cares about net after returns and adjustments
• recognized vs. booked revenue — finance may defer or recognize revenue on a different timeline than operations or growth
• order date vs. capture date vs. settlement date — each system may anchor the same dollar amount to a different date
• taxes, shipping, discounts, and gift cards — some systems include them, others strip them out
• platform credit inflation — ad tools often claim conversion value based on attribution windows, not a clean system-of-record transaction table

Which version to use for which decision

• Shopify gross revenue: useful for day-to-day commerce monitoring
• net revenue after returns/discount effects: better for operator decisions
• recognized finance revenue: right for board, accounting, and formal financial reporting
• warehouse-modeled revenue: best when you need one explicit definition tied to business decisions

Practical rule

If the question is “did sales happen,” Shopify is usually good enough.

If the question is “did we make money,” Shopify alone is usually not enough.

2. Average Order Value (AOV)

Where it usually lives

• Shopify dashboards
• GA4 ecommerce reports
• ad platform conversion reporting
• finance or BI dashboards

Why the numbers disagree

AOV sounds simple: revenue divided by orders.

But even here, teams quietly mix different numerators and denominators:

• one system counts all orders, another only attributed orders
• one includes discounted orders, another normalizes list price or excludes certain transactions
• one uses gross sales, another uses net sales after refunds
• one includes subscription renewals, another isolates first purchases only

This is why paid social can appear to have a heroic AOV inside one dashboard and an ordinary AOV everywhere else.

Which version to use for which decision

• platform-specific attributed AOV: useful for channel directional reads, carefully
• storewide gross AOV: useful for merchandising trend checks
• net AOV after refunds/discounts: more useful for profitability and planning
• first-order vs. repeat-order AOV: critical when acquisition and retention have different economics

Practical rule

Never compare a platform-attributed AOV to a finance-grade storewide AOV as if they are the same metric. They are answering different questions.

3. Customer Acquisition Cost (CAC)

Where it usually lives

• ad platforms
• blended spend spreadsheets
• CRM / revenue reporting
• warehouse or finance-adjacent dashboards

Why the numbers disagree

CAC breaks because the denominator changes constantly.

Common versions include:

• ad spend divided by platform-reported conversions
• spend divided by new customers in Shopify
• spend divided by qualified customers after a business rule filter
• fully loaded CAC including agency fees, creative, tools, or payroll overhead

That means two teams can both say “CAC” and be off by 30-50% without either one making a calculation error. For a deeper look at the denominator and inclusion choices that cause the most disagreement, How to Calculate True Customer Acquisition Cost breaks down the versions worth separating.

Which version to use for which decision

• platform CAC: useful for in-platform optimization, but not enough for budget truth
• blended paid CAC: better for channel allocation
• fully loaded CAC: best for executive decisions and profitability conversations
• cohort CAC by segment: useful when customer quality varies sharply by channel or offer

Practical rule

If a team says CAC is improving while margin is getting worse, the first thing to check is what costs they excluded.

4. Return Rate

Where it usually lives

• Shopify or the returns platform
• 3PL / fulfillment tooling
• finance reports
• warehouse models by product, channel, and cohort

Why the numbers disagree

Return rate is almost always too blunt in native dashboards.

Differences usually come from:

• units returned vs. orders returned
• revenue refunded vs. item count
• returns created vs. returns fully processed
• same-period returns vs. lagged returns against original order cohorts

A product launch can look amazing in month one and terrible in month two when the returns finally land. If you report both months off transaction date alone, you will misread what happened.

Which version to use for which decision

• same-period operational return rate: useful for customer service and ops management
• cohort-based return rate: better for merchandising and channel truth
• refund-adjusted revenue rate: important for finance and profitability analysis

Practical rule

If you want to know whether a campaign drove good customers, tie return behavior back to acquisition source and product mix. Storewide return rate hides the interesting part.

5. Lifetime Value (LTV)

Where it usually lives

• spreadsheets and executive decks
• subscription tooling
• retention tools
• warehouse models

Why the numbers disagree

LTV is where a lot of teams stop measuring and start storytelling.

Different versions use:

• historical realized revenue only
• predictive models
• gross revenue vs. gross margin contribution
• customer lifespan assumptions that were never revalidated
• customer segments blended together even when repeat behavior is wildly different

That is how you get an LTV:CAC ratio that looks healthy in a slide deck and useless in real operations.

Which version to use for which decision

• realized historical LTV: useful for grounded retrospective analysis
• gross-margin-adjusted LTV: better for budget decisions than revenue-only LTV
• predictive LTV: useful if the model is validated and the operating team trusts it
• segment-specific LTV: essential when products, channels, or cohorts behave differently

Practical rule

If your LTV number has not been revisited in six months, it is probably being used more as comfort than as truth.

6. Contribution Margin

Where it usually lives

• finance workbooks
• custom BI models
• warehouse tables
• almost never in one native marketing tool

Why the numbers disagree

This is the metric that most directly explains whether growth is actually worth having.

It is also the one least likely to exist cleanly in a single system because it depends on connecting:

• revenue
• discounts
• returns
• cost of goods sold
• shipping and fulfillment
• channel spend
• sometimes payment processing or support cost allocations

If those live in six places, contribution margin will not show up by accident. The Ecommerce Data Playbook walks through how to connect Shopify, ad platforms, and fulfillment data into a single view where this kind of metric can actually be built.

Which version to use for which decision

• high-level contribution margin: useful for overall business health
• channel-level contribution margin: critical for media allocation
• product/category margin: necessary for merchandising decisions
• segment margin: useful when different customer groups behave differently after acquisition

Practical rule

If you can only see ROAS and revenue, you are still missing the part that tells you whether the growth story survives contact with reality.

Why Ad Platforms and Finance Rarely Match

This deserves its own section because teams keep treating it like a bug.

It is not always a bug.

Ad platforms are designed to answer a question like: “How much conversion value can this platform credibly claim influence over?”

Finance is designed to answer a question like: “What money actually happened, when, and under what accounting rules?”

Those are not the same product requirements.

So no, Meta and finance should not be expected to match line for line.

What should happen is:

• the differences are documented
• the system-of-record number is explicit
• the team knows which version is used for optimization versus reporting versus profitability decisions

That is what mature measurement looks like.

The Minimum Viable Source-of-Truth Stack

For most ecommerce teams, the first sane version of metric trust looks like this:

• commerce platform for operational transaction visibility
• ad platforms for directional optimization signals
• finance / ERP for formal revenue treatment
• warehouse + modeled definitions for cross-functional decision-making

You do not need infinite tooling.

You do need one place where the business definitions are made explicit and defended. If you are in SaaS rather than ecommerce and want to see what that source-of-truth stack actually looks like at a mid-size company, the Data Stack Teardown is a useful reference point.

What to Fix First

If your ecommerce metrics are constantly fighting each other, do not try to standardize everything at once.

Start with these four questions:

1. Which number creates the most expensive disagreement right now?
2. Which systems contribute to that number?
3. Where do the definitions fork?
4. Which version should govern which decision?

That sequence gets you out of metric philosophy and back into usable operating decisions.

Download the Ecommerce Cheat Sheet

The PDF version is a one-page operator reference: the core metrics, where they live, why they drift, and which version to trust for which decision.

Bottom Line

The goal is not to make every dashboard match perfectly.

The goal is to know why the numbers differ, which version is fit for which decision, and where your team should stop pretending a platform screenshot is the same thing as decision-grade reporting.

If your team needs that level of clarity, start with Show Me the Margin. If the problem is broader than one ecommerce profitability lens, Revenue Analytics is where the larger measurement work starts.

For a concrete example of what better ecommerce measurement makes possible, see how a DTC ecommerce brand cut wasted ad spend 35% with true channel-level ROAS or how an ecommerce SaaS team replaced a $120K/year vendor tool with a warehouse-native approach.