A Practical Guide to E-Commerce Analytics

E-commerce analytics is useful when it connects store behavior to revenue decisions. It becomes noisy when every click, pixel, and customer trait is collected simply because a tool can collect it.

A privacy-friendly store can still answer the important questions: where shoppers come from, which products attract interest, where checkout breaks, which campaigns pay back, and which pages need improvement.

Most of this only becomes worth measuring once the store exists, of course. If you are still at the stage of working out how to set up an e-commerce business, decide on your platform and payment flow first, then come back and instrument it properly rather than bolting analytics on later.

The Metrics That Matter Most

Conversion rate

Conversion rate is the share of visitors or sessions that complete a purchase. Track it by traffic source, campaign, device category, landing page, and product category.

Do not compare conversion rates without context. Paid search traffic with strong purchase intent may convert very differently from blog traffic. Mobile conversion may be lower because users research on phones and buy later on desktop.

Revenue per visitor

Revenue per visitor combines traffic quality, conversion rate, and order value. It is more useful than conversion rate alone.

Formula:

revenue per visitor = total revenue / visitors

If conversion rate goes down but average order value rises enough, revenue per visitor may still improve. This helps avoid optimizing only for cheap purchases.

Average order value

Average order value shows how much customers spend per transaction.

Formula:

average order value = revenue / orders

Use AOV to evaluate bundles, free-shipping thresholds, product recommendations, and merchandising. Be careful with discount campaigns: a discount can increase conversion while reducing margin.

Cart and checkout abandonment

Track the funnel:

• Product view
• Add to cart
• Cart view
• Checkout start
• Shipping step
• Payment step
• Purchase

A high cart abandonment rate can indicate unexpected shipping costs, forced account creation, limited payment methods, slow checkout, poor mobile UX, or trust concerns. You do not need to identify the individual shopper to see where the funnel leaks.

Where checkout breaks

Unexpected shipping costsRevealed too late in checkout

Forced account creationAn extra step before payment

Limited payment methodsThe shopper's preferred option is missing

Poor mobile UXCheckout is harder to finish on a phone

Product performance

Measure:

• Product detail views
• Add-to-cart rate
• Purchase rate
• Revenue
• Refunds or returns, if available
• Search terms that lead to product pages

For privacy, avoid sending product-level events to advertising platforms unless you have a valid consent and a clear reason. Aggregate store analytics is usually enough for merchandising decisions.

Acquisition Metrics

E-commerce teams overspend when they trust ad-platform dashboards without independent measurement.

Use UTMs for every campaign. Google's URL builder documentation explains standard parameters such as utm_source, utm_medium, utm_campaign, utm_id, and utm_content (Google Analytics URL builder). These parameters work in privacy-first analytics because they are passed in the landing URL.

Track:

• Sessions by source and campaign
• Conversion rate by campaign
• Revenue by campaign
• Revenue per visitor by campaign
• Assisted conversions where your analytics supports them
• New vs returning customer revenue, if your commerce platform provides it

For paid ads, combine analytics with platform spend data to estimate return:

ROAS = attributed revenue / ad spend

Do not pretend attribution is perfect. Browser privacy features, consent rejection, cross-device shopping, and delayed purchases all create gaps. Use attribution as directional evidence, not absolute truth.

Privacy-Friendly Measurement Design

A store does not need to track shoppers across the web to improve sales.

A lean setup can use:

• First-party page and event analytics
• Aggregated product and funnel events
• UTMs for campaign source
• Order IDs stored in the commerce backend, not analytics profiles
• Country or region instead of exact location
• Short retention for raw events
• Consent-gated ad pixels only where necessary

Avoid:

• Sending email addresses or phone numbers to analytics
• Recording checkout sessions by default
• Loading retargeting pixels before consent
• Storing raw IP addresses longer than needed
• Combining analytics events with broker-enriched profiles

The strongest ecommerce analytics stack separates operational order data from website behavior data. Your store platform needs customer details to fulfill an order. Your website analytics usually does not.

Two systems, two jobs

Store platform

• Customer details needed to fulfill orders
• Order IDs kept in the commerce backend

Website analytics

• First-party page and event data
• Aggregated product and funnel signals

How to Prioritize Improvements

Use metrics to find the highest-leverage issue:

• High product views, low add-to-cart: improve price clarity, images, reviews, sizing, or availability.
• High add-to-cart, low checkout start: review cart UX, shipping estimates, coupon distractions, and trust signals.
• High checkout start, low purchase: test payment methods, form errors, address validation, and mobile usability.
• High traffic, low revenue per visitor: review campaign intent and landing page message match.
• High conversion, low AOV: test bundles, thresholds, and product recommendations.

Make one change at a time where possible and annotate the release date in your analytics.

Ecommerce Event Checklist

Start with a compact event set: product viewed, cart started, checkout started, payment failed, purchase completed, coupon applied, and refund requested. Use safe properties such as product category, price band, currency, campaign, device class, and country.

Do not send names, emails, exact addresses, payment details, order notes, checkout URLs with tokens, or full order IDs to website analytics. Keep revenue truth in the commerce platform and use analytics to explain acquisition, content, and funnel trends.

The Bottom Line

E-commerce analytics should make store decisions easier. Focus on conversion, revenue per visitor, average order value, funnel drop-off, product performance, and acquisition quality. You can measure all of this with a privacy-first approach that respects shoppers and avoids unnecessary third-party tracking.

A privacy-safe event set

A practical store can start with a compact event set: product_viewed, cart_started, checkout_started, payment_failed, purchase_completed, coupon_applied, and refund_requested. Useful properties include product category, price band, currency, campaign, device class, and country. Avoid sending names, email addresses, exact shipping addresses, phone numbers, payment details, or order notes to web analytics.

Keep the order system as the source of truth for revenue. Analytics can report trends and funnel health, but refunds, taxes, discounts, fraud checks, and fulfillment status belong in commerce or finance systems. That separation reduces privacy risk and also prevents marketers from making decisions from incomplete revenue numbers.

Frequently Asked Questions

Why shouldn't I compare conversion rates across channels directly?

Traffic sources carry different intent, so a shared number hides real differences. Paid search visitors often arrive ready to buy, while blog readers are still researching, so a lower blog conversion rate isn't necessarily a problem. Mobile visitors also convert differently since many browse on a phone and finish the purchase later on desktop. Compare conversion rate by source, campaign, device, and landing page instead of looking at one blended figure.

What does revenue per visitor tell me that conversion rate doesn't?

Revenue per visitor combines traffic quality, conversion rate, and order value into one number, so it captures the full picture that conversion rate alone misses. The formula is total revenue divided by visitors. It keeps a team from optimizing for cheap purchases just to push the conversion number up.

Can revenue per visitor improve even if conversion rate drops?

Revenue per visitor can improve if average order value rises enough to offset the fewer purchases. A campaign that brings in fewer but bigger orders can still grow revenue per visitor even as the conversion rate falls. That's why the two metrics need to be read together rather than in isolation.

How do I calculate average order value?

Average order value equals revenue divided by orders. Use it to judge whether bundles, free-shipping thresholds, and product recommendations are working. Watch discount campaigns closely, since a discount can raise conversion rate while cutting into margin.

What usually causes cart or checkout abandonment?

The most common culprits are unexpected shipping costs, a forced account creation step, limited payment methods, a slow checkout flow, weak mobile usability, and general trust concerns. Following the funnel from product view through purchase shows exactly where shoppers drop off. None of this requires identifying the individual shopper, just tracking where the funnel leaks.

What product metrics matter for merchandising decisions?

Track product detail views, add-to-cart rate, purchase rate, revenue, refunds or returns where available, and the search terms that lead shoppers to product pages. Aggregate store analytics is usually enough to make merchandising calls. Avoid sending product-level events to advertising platforms unless there's valid consent and a clear reason for it.

Why bother with UTM parameters if I already have analytics?

UTMs identify source, medium, campaign, ID, and content in the landing URL itself, so they work even in a privacy-first analytics setup that doesn't rely on cross-site tracking. Google's URL builder documentation covers the standard parameters. Tagging every campaign is what lets you measure sessions, conversion rate, and revenue by source instead of trusting an ad platform's own dashboard.

How reliable is ecommerce attribution?

Not perfectly reliable. Browser privacy features, consent rejection, cross-device shopping, and delayed purchases all create gaps between what actually happened and what gets attributed. Treat ROAS and attributed revenue as directional evidence for decisions, not as an exact accounting of every sale.

What data should never go to website analytics?

Keep names, email addresses, exact shipping addresses, phone numbers, payment details, order notes, checkout URLs with tokens, and full order IDs out of website analytics entirely. Stick to safe properties like product category, price band, currency, campaign, device class, and country. Revenue truth, refunds, taxes, and fraud checks belong in the commerce or finance system, not the analytics tool.

Why separate order data from website behavior data?

A store platform needs customer details to fulfill an order, but website analytics usually doesn't need any of that to explain acquisition, content, and funnel trends. Keeping the two systems apart reduces privacy risk and keeps analytics from mixing operational data with behavioral data. It also stops marketers from making calls off incomplete revenue numbers, since the order system stays the single source of truth for revenue.