Ecommerce Marketing Analytics: The Metrics, Models, and Tools That Matter
Most stores are not short on data. They have Shopify reports, GA4, three or four ad dashboards, Klaviyo, and a spreadsheet someone built two years ago. What they are short on is a way to make those numbers agree long enough to decide where next week's budget goes.
Ecommerce marketing analytics is the practice of connecting ad spend to orders and, ultimately, to profit. It answers three questions: is marketing making the business money, which channels and campaigns are doing the work, and what should change.
This guide covers the metrics worth tracking, where each number comes from, why the sources disagree, and how to run a weekly review that ends in decisions instead of debates.
What is ecommerce marketing analytics?
Ecommerce marketing analytics is the measurement of how marketing activity turns into revenue, new customers, and contribution margin for an online store. It sits between your ad platforms, which report what they think they caused, and your store, which reports what actually sold.
The distinction matters because those two sides rarely match. An ad platform sees impressions and clicks. Your store sees orders. Analytics is the work of joining them with enough accuracy that you can move money with confidence.
A useful working definition: if a number does not help you decide to spend more, spend less, or change something, it is reporting, not analytics.
Which ecommerce KPIs actually matter?
The ecommerce KPIs that matter fall into three tiers: business metrics that tell you whether marketing is profitable, efficiency metrics that tell you which spend is working, and diagnostic metrics that tell you why. Read them top down. A great click-through rate means nothing if contribution margin is shrinking.
| Tier | Metrics | Question it answers | Review cadence |
|---|---|---|---|
| Business | Revenue, MER, contribution margin or profit, new vs returning customer revenue | Is marketing growing the business profitably? | Weekly and monthly |
| Efficiency | ROAS, POAS, CAC, new customer CAC (nCAC), LTV:CAC | Which channels and campaigns earn their budget? | Weekly, daily for large spenders |
| Diagnostic | Conversion rate (CVR), average order value (AOV), click-through rate (CTR) | Why did an efficiency metric move? | As needed when something shifts |
Business metrics
MER, or marketing efficiency ratio, is total revenue divided by total ad spend. It ignores attribution entirely, which is its strength. Say a store spends $10,000 on ads in a week and books $42,000 in revenue. MER is 4.2. No platform can inflate that number, because it uses store revenue and invoiced spend. Our MER guide for ecommerce CEOs covers how to set a MER target by margin.
MER in one line: store revenue over invoiced spend
$42,000
store revenue for the week
$10,000
total invoiced ad spend
4.2
MER, with no attribution involved
Contribution margin is what is left after cost of goods, shipping, payment fees, returns, and ad spend. It is the metric that actually pays salaries. Split revenue by new and returning customers as well, because a business can grow MER for months by harvesting its existing list while acquisition quietly stalls.
Efficiency metrics
ROAS is attributed revenue divided by spend for a channel or campaign. POAS swaps revenue for gross profit, so a campaign selling low-margin products does not look as good as one selling high-margin ones at the same ROAS. If you discount heavily in Q4, the difference gets large quickly; see POAS vs ROAS for how to switch your thinking.
CAC is spend divided by customers acquired. nCAC counts only first-time buyers, which is the version that matters for prospecting budgets. Pair it with lifetime value: a healthy LTV to CAC ratio tells you how much you can afford to pay for a customer without needing them to be profitable on the first order.
One trap here: blended CAC, which divides all spend by all new customers, can look fine while one channel is badly overpaying. The breakdown in blended CAC vs channel CAC shows why you need both views.
Diagnostic metrics
CVR, AOV, and CTR explain movement in the tiers above. If ROAS drops, check whether conversion rate fell (a site or offer problem), AOV fell (a merchandising problem), or CTR fell (a creative problem). These are where conversion rate optimization work starts. Do not set budgets from them directly.
Where do your numbers come from, and why don't they agree?
Your numbers disagree because each source uses a different attribution model, a different lookback window, and a different view of which touchpoints it can see. None of them is broken. They are answering different questions.
Here is how the main sources compare:
| Source | How it credits orders | What it sees | Main blind spot |
|---|---|---|---|
| Shopify marketing reports | Last non-direct click by default | Every order in the store | Only the touchpoints it can tie to a session |
| GA4 | Data-driven or last click | Site sessions and paths | Consent, cookie loss, revenue mismatches with the store |
| Ad platforms | Their own windows, often including views | Their own impressions and clicks | Every other channel, so totals overlap |
| Independent click-only attribution | Your choice of model, across channels | Verified clicks stitched to store orders | Does not measure incrementality on its own |
Shopify's marketing reports default to the last 30 days using a last non-direct click model, with first click, last click, any click, and linear also available. That is accurate for orders, but it only sees the journeys it can stitch. Our breakdown of why Shopify data is accurate but isolated goes deeper.
GA4 has its own view. Its key events attribution paths report shows which channels initiate, assist, and close conversions, with paths up to 20 touchpoints. But first click, linear, time decay, and position-based models are no longer available in GA4 as of November 2023, and its revenue often will not match the store. The common causes are in GA4 vs Shopify revenue.
Then the ad platforms. Each one reports the conversions it believes it influenced, inside its own window, and many count view-throughs. Add up Meta, Google, and TikTok and the total usually exceeds the orders your store actually took. That gap is the core of platform-reported ROAS vs actual ROAS.
The practical answer is not to pick a winner and ignore the rest. It is to choose one source for budget decisions and treat the others as context.
What does a good ecommerce measurement stack look like?
A good measurement stack has four layers, each with one job: store analytics for the order record, ad platforms for in-platform optimization, attribution for channel and campaign credit, and marketing mix modeling for long-range budget planning.
Four layers, one job each
- Store analytics (Shopify, BigCommerce, WooCommerce). The system of record for orders, refunds, and customers. Every other number should reconcile back to it.
- Ad platforms. Use their reporting to optimize creative, audiences, and bids within the platform. Their algorithms need their own signal.
- First-party attribution. An independent layer that credits verified clicks to real store orders, across every channel, using one model. This is what you use to move money between channels.
- MMM. A model that estimates channel contribution from spend and outcome history, including channels that clicks do not capture well. Google's Meridian is free and open source and was built to answer historical contribution by channel, response curves, and future budget allocation, according to Google's Meridian introduction.
Each layer answers a different question, and the comparison in MMM vs attribution vs incrementality explains when to lean on which.
Multi-channel journeys are why the attribution layer exists at all. McKinsey reports that more than half of B2C customers engage with three to five channels each time they make a purchase. A single platform's view of that journey will always be partial.
See it on your own account.
See the gap channel by channel.
AdBeacon shows click-verified revenue next to what each ad platform reports, so you can see the gap channel by channel instead of guessing at it. Book a live AdBeacon demo.
Which attribution model should you use for analytics?
Use the model that matches the decision you are making. There is no single correct model, and switching between two or three is often the fastest way to understand a channel.
- First click shows which channels start journeys. Useful for judging prospecting.
- Last click shows which channels close. Useful for retargeting and email, misleading for top of funnel.
- Linear spreads credit evenly across every click. A fair default for reporting.
- Linear excluding direct and paid linear remove touchpoints you did not pay for, which keeps owned and direct traffic from diluting paid channel credit.
- Lighthouse, AdBeacon's optimization model, works like linear but credits only the platform you are optimizing. It is built for in-platform decisions rather than omnichannel reporting.
In AdBeacon you can apply any of these with lookback windows up to 180 days, which matters for higher-priced products where the gap between first click and purchase runs into weeks.
AdBeacon uses click-only attribution. It credits clicks it can verify and leaves view-through conversions out. That makes the numbers more conservative than most platform dashboards, and more auditable, because every credited order traces back to a real click.
How should you run a weekly marketing analytics review?
A weekly review should take under an hour and end with a short list of budget and creative changes. If it ends with “let's keep an eye on it,” it was a reporting meeting.
- Start with business metrics. Revenue, MER, contribution margin, and the new vs returning split, week over week and against the same week last year.
- Check reconciliation. Do attributed orders roughly match store orders after removing canceled ones? AdBeacon added an “Exclude canceled orders from attribution” setting in October 2026 for exactly this reason.
- Review channel efficiency. ROAS, POAS, and nCAC by channel from your attribution source, with platform-reported numbers alongside for context.
- Drill into movers. For any channel that shifted more than your normal noise band, check CVR, AOV, and CTR to find the cause.
- Decide. Write down what moves: budget up, budget down, creative refresh, test to launch.
The 10 attribution reports for a CMO weekly review list gives you the specific views to pull. For a single screen that holds most of this, see the ecommerce ad performance dashboards guide.
What are the most common ecommerce analytics mistakes?
The most common mistake is optimizing budget against platform-reported ROAS. Each platform credits itself generously, so every channel looks like it deserves more money, and the total never matches the bank account.
Other mistakes show up repeatedly in audits:
- Mixing models in one report. Comparing Meta's numbers with Google's numbers compares two different rulebooks.
- Ignoring margin. A 4.0 ROAS on a 25% margin product loses money once fees and returns land.
- Counting canceled and refunded orders. Attribution that includes them overstates every channel.
- Reading daily ROAS as signal. Daily numbers swing with lag and small samples. Weekly is the minimum useful window for most stores.
- Blending new and returning customers. Retargeting and email look brilliant on blended numbers because they reach people who were already going to buy.
Why a 4.0 ROAS can lose money at a 25% margin
Fees and returns come out of that $0.00, which turns it negative.
None of these need new tools. They need agreed definitions.
How does AdBeacon fit into ecommerce marketing analytics?
AdBeacon is the attribution layer in the stack. It collects first-party click data, stitches it to orders from Shopify, BigCommerce, WooCommerce, and (in beta) Magento and Adobe Commerce, and reports each channel's revenue using the model you choose. Platform-reported numbers sit beside AdBeacon-measured numbers, so the gap is visible rather than hidden.
It connects to Meta, Google, TikTok, Pinterest, Snapchat, Microsoft, Amazon Ads, and Reddit Ads (beta), plus Klaviyo for email and SMS. The free Google Meridian MMM integration covers the planning layer. Luma, AdBeacon's AI assistant, flags anomalies and makes recommendations, and data flows out through a Google Sheets add-on, an API, and an MCP server for AI tools like Claude. You can explore the reporting side on the report accurately page.
On average, clients see 15 to 20% revenue growth and about 25% ROAS growth after 90 days. Individual results vary; Poppy & Pine, working with an agency partner, saw revenue rise 60% and ROAS rise 19% over 90 days.
What to do next
- Write down your three business metrics (revenue, MER, contribution margin) and the target for each this quarter.
- Pick one source of truth for channel budget decisions, and document which model and lookback window it uses.
- Reconcile last month's attributed orders against store orders, excluding canceled and refunded ones.
- Split every efficiency metric by new and returning customers.
- Book a recurring 45-minute weekly review with a fixed agenda and a decisions log.
- Once you have six months or more of clean spend history, add an MMM layer to check long-range allocation.
Good ecommerce marketing analytics is less about more dashboards and more about fewer arguments. Choose one trusted source for budget, keep platform data for in-platform optimization, and review the same metrics in the same order every week. When the numbers disagree, the disagreement itself becomes useful information. If you want to see how your current channel numbers compare once views are stripped out and orders are matched to real clicks, walk through your own data with our team.
Frequently Asked Questions
What is ecommerce marketing analytics?
Ecommerce marketing analytics is the measurement of how marketing spend turns into orders, customers, and profit for an online store. It joins ad platform data with store data so you can decide where budget should go.
What are the most important ecommerce KPIs?
Start with revenue, MER, and contribution margin, split by new and returning customers. Then use ROAS, POAS, nCAC, and LTV:CAC to judge channels, and CVR, AOV, and CTR to diagnose changes.
Why don't Shopify, GA4, and my ad platforms show the same numbers?
Each uses a different attribution model, lookback window, and set of visible touchpoints. Shopify defaults to last non-direct click, GA4 uses its own models, and ad platforms credit themselves inside their own windows, so their totals overlap.
Is MER better than ROAS?
They answer different questions. MER tells you whether total marketing is efficient and cannot be inflated by attribution, while ROAS tells you which channel or campaign is earning its budget.
How often should I review marketing analytics?
Weekly is the right cadence for most stores, with monthly reviews for margin and LTV trends. Daily numbers are too noisy to set budgets from for most accounts.