Ecommerce AnalyticsFewer numbers, better decisions

The problem is not too little data#

Online stores generate more data than any team can read. The store platform, analytics tool, ad platforms, marketplaces, courier dashboards, payment gateway and messaging tools each produce their own reports, often with conflicting numbers. The usual problem is not a shortage of data. It is the absence of a structure that says which numbers matter, which source is right and what decision each metric informs.

This framework provides that structure. It does not require expensive tools. It requires agreement on definitions, a hierarchy of metrics and a routine for reading them.

A dashboard that cannot change a decision is decoration.

The metric hierarchy#

Metrics sit at different levels. At the top are outcomes the business ultimately cares about. Below them are the drivers that produce those outcomes. At the bottom are diagnostics that explain why drivers move. Reading from the top down keeps attention on what matters and stops teams from chasing diagnostic noise.

The revenue equation underpins the hierarchy: sessions multiplied by conversion rate multiplied by average order value gives revenue; subtracting variable costs gives contribution. Each lower-level metric should map to a term in that equation.

Define before you measure#

Many disagreements about numbers are disagreements about definitions. Is revenue gross or net of discounts? Does it include GST? Is a customer new if they bought once on a marketplace? Is conversion rate per session or per user? Write the definitions down once, and use them everywhere.

The data stack#

Every ecommerce business has a data stack, whether it was designed or accumulated. Naming the layers clarifies which system answers which question and where the source of truth sits.

For small stores, the ‘warehouse’ can be a well-structured spreadsheet updated weekly. As the business grows, a proper data warehouse and Looker Studio or similar dashboards become worthwhile. The principle is the same: join orders with costs and marketing so contribution can be calculated.

Event tracking that serves ecommerce#

Behavioural analytics should record the key steps of the shopping journey. In GA4, the recommended ecommerce events include view_item, add_to_cart, begin_checkout and purchase, each carrying item and value parameters. See GA4 events and measurement plan for set-up.

• Pass item IDs that match your catalogue, so behaviour can be joined with product data.
• Record payment method on purchase, so prepaid and COD behaviour can be compared.
• Respect consent requirements and check current consent mode guidance; see consent mode.
• Consider server-side tracking where browser-based tracking is unreliable.

Reconciling the sources#

The store, the analytics tool and the ad platforms will never agree exactly. They count at different times, use different attribution rules and are affected differently by consent and ad blockers. Trying to make them match is wasted effort. Instead, choose roles.

Attribution deserves humility. Last-click, platform-reported and data-driven models each tell a partial story. For large budgets, incrementality testing and marketing mix modelling offer more reliable answers. See marketing attribution.

Cohorts: the view most stores miss#

Monthly totals mix new and returning customers, and older and newer cohorts, in ways that hide what is happening. Cohort analysis groups customers by the month of their first order and follows their behaviour over time. It answers questions totals cannot: are newer customers less loyal than older ones? Do festive cohorts repeat less? Which channel brings customers who come back?

See cohort analysis for method. Even a simple cohort table, updated monthly, changes the quality of growth decisions.

Joining marketing to delivery outcomes#

In COD-heavy businesses the most important join in the whole stack is between marketing source and delivery outcome. Without it, every channel is judged on orders placed, and channels that attract refusals look better than they are. With it, the business can see cost per delivered order by campaign, creative and region.

The join requires an order ID that travels from the store to the courier data, and a record of the marketing source on the order. Most platforms can store UTM parameters or a source field against an order; see UTM parameters. Once joined, the data can also be sent back to ad platforms as offline conversions, so automated bidding learns from delivered orders rather than placed ones.

Common analytics mistakes#

• Reporting gross revenue including tax, cancellations and returns as if it were income.
• Adding up each ad platform’s attributed revenue and treating the total as real.
• Judging retention after too short a window for the product’s natural cycle.
• Comparing conversion rates across periods without separating device and channel mix.
• Building dashboards before agreeing definitions, then arguing about the numbers in every meeting.
• Ignoring marketplace and quick-commerce data because it lives outside the main stack.

A review rhythm#

• Daily: orders, stock-outs, site errors, payment failures. Act on problems, do not analyse trends.
• Weekly: spend versus new customers, conversion by device and channel, top products, RTO and returns.
• Monthly: contribution by route and channel, cohort repeat rates, CAC trend, promotion results.
• Quarterly: strategy review against the metric hierarchy, measurement plan updates, incrementality findings.

Published by Fabulous.Media, a network of specialist marketing agencies. Updated 9 October 2026. Platform features change often; check current official documentation before acting on platform-specific detail.