E-commerce
April 8, 2026
E-commerce analytics refers to the systematic measurement, collection, and interpretation of data related to your online store: traffic, behavior, conversion, cart, revenue, retention, and acquisition costs. It differs from the word “analysis” in the broad sense: analysis can be occasional or qualitative; analytics relies on stable definitions, traceable events, and repeatable dashboards.
This Analytics & Insights guide sets out the vocabulary, indicator families, and frequent pitfalls. It complements the article on conversion in Google Analytics, conversion rate definitions, and e-commerce business foundations.
Tool and screen names evolve: check the official documentation (Google Analytics, your CMS) for your current configuration.
Qstomy produces conversational signals (frequent questions, intentions, escalations) that enrich analytics when they are exported or tagged in your data stack.
Executives should require teams to provide a documented gap between objectives and results each month, with possible causes (seasonality, stock, competition, tracking bug) rather than a vague narrative.
Finance teams reconcile cash receipts and recognized revenue: marketing analytics does not replace accounting but should move toward explainable gaps.
Product teams use analytics to prioritize the roadmap (funnel features) and pages to enrich based on measured friction.
Legal teams validate that internal reports do not contain unnecessary personal data and that access is logged.
In monitoring, follow release notes for analytics tools and advertising APIs: a schema change can break your reports overnight without sufficient internal communication.
Data committees bringing together marketing, IT, finance, and legal make it possible to arbitrate access, retention, and new use cases without creating incompatible silos between countries or subsidiaries.
Retail teams that combine web and store must unify definitions of “sale” and “customer” to avoid two competing truths in the same executive slides.
Fast-growing startups often move from Excel to a cloud warehouse: plan for the human cost of migration and documentation, not just the license cost.
International scale-ups must harmonize time zones, currencies, and fiscal calendars in consolidated reports: a currency conversion or time-scope error can distort costly strategic decisions.
Logistics teams add indicators for availability and preparation time that explain conversion variations even when no marketing change has occurred: crossing web analytics and ops is essential.
Investors often request standardized metrics (GMV, take rate, cohorts): preparing definitions aligned with your internal analytics avoids last-minute restatements before due diligence.
Statistical tests do not replace business judgment, but they prevent treating noise as a trend: set sample size, acceptable risk, and stopping criteria in advance to validate a change in journey or pricing.
Weekly review meetings should start with significant gaps versus a baseline, then dig into root causes: the reverse order produces long meetings with no decision and no owner assigned for next steps.
Summary
Analytics vs analysis: useful internal definitions
In meetings, clarify: analytics = a continuous measurement framework with rules; analysis = interpretation of a situation (e.g., a drop in conversion on a landing page). Without shared definitions, we mix up impressions, sessions, and users.
Indicator vs metric
Often used as synonyms; specify whether you are talking about a raw count or a ratio (rate, average basket value).
Dimensions and measures
In tools like GA4, a dimension is an attribute (country, channel); a measure is an aggregated quantity (revenue, events).
Business objectives
Analytics should support decisions: pricing, assortment, ad budget, customer service hiring.
Internal documentation
A glossary of “what we call paid conversion” avoids endless debates.
Industry benchmark
Compare with public benchmarks cautiously: basket size, country, and channel mix differ.
OKRs and analytics
Link quarterly objectives to measurable indicators: this avoids discussions without a success criterion.
Retrospective
After each seasonal peak, analyze what diverged from forecasts and update assumptions for the following year.
Data literacy
Train managers to read a confidence interval and avoid hasty conclusions from small samples.
E-commerce funnel: acquisition, behavior, conversion, retention
Classic framework: acquisition (traffic sources), activation (first value), revenue, retention, referral according to the AARRR model. For e-commerce, we often connect product sessions, add-to-cart, checkout, purchase, then repeat purchase.
Micro-conversions
Newsletter signup, account creation, click on “in-store availability”: useful steps before the sale.
Seasonality
Compare period to comparable period: a spike can mask a structural drift.
Segments
New vs returning, mobile vs desktop, country: global averages often lie.
Link with marketing
See the e-commerce conversion funnel to structure the measured steps.
Touchpoint mapping
Cross-functional workshop to list data sources and owners: the basis of a measurement plan.
Order of steps
A theoretical funnel may differ from the actual journey observed in session recordings.
Real journeys
Session recordings and heatmaps complement theoretical funnels to reveal loops and dead ends.
Mobile vs desktop
Separate the funnels: the overall average conversion rate can hide a broken mobile checkout.
Systems mapping
Diagram of CRM, analytics, ads, email flows: identify duplicates and sources of truth for each KPI.
B2B journey
Long cycles: combine web analytics with CRM and sales pipeline to attribute value to content.
KPIs: traffic, engagement, conversion, value and profitability
Common indicators: sessions or visits, bounce rate or engagement, conversion rate, average order value (AOV), revenue, margin after variable costs, CAC, LTV. Each KPI answers a specific question; stacking them without hierarchy drowns readability.
Conversion rate
Define the numerator and denominator: see definitions.
Product and category
Revenue per SKU, add-to-cart rate per product page, products frequently viewed together.
Costs
Ad ROAS without product margin can celebrate loss-making sales.
Traffic quality
High traffic with massive bounce on a landing page may signal an off-target advertising promise.
Contribution margin
After product cost and variable logistics: a better compass than gross revenue alone.
Share of voice
In competitive categories, relative organic traffic complements internal KPI tracking.
Average order value and bundles
Measure the effect of bundled offers on net margin, not only on basket size.
Discounts
Share of revenue sold under promotion: an indicator of dependence on sales periods.
Price elasticity
Structured price tests and measurement of impact on volume and margin: analytics must isolate external factors.
Category seasonality
Compare YoY by product family to distinguish market trend from site performance.
Google Analytics 4, e-commerce events and limits
GA4 is based on an event model; enhanced ecommerce (or equivalent) makes it possible to track view_item, add_to_cart, purchase, etc. Implementation must be validated through testing (debug mode, real-time reports). Documentation: Google Analytics Help.
Sampling and thresholds
Some reports may be approximate at high volume or under privacy constraints.
Cross-device identity
User-ID, signed-in users: improve continuity but require respect for privacy.
Differences with the store
Compare GA revenue vs back office: differences related to refunds, time zones, bot filters.
Data retention
Configurable durations: impact on long-term analysis.
BigQuery
Export GA4 to a warehouse for advanced joins: SQL skills required.
Enhanced measurement
Useful automatic events but should be audited for noise (scroll, outbound).
DebugView
Validate in real time the triggering of e-commerce events during production deployments.
Reporting API
Automate extractions to spreadsheets or BI with controlled quotas and pagination.
Cross-domain
Consistent measurement between checkout and marketing domains: link configuration and internal exclusion.
Subdomains
Blog or help center on a subdomain: inclusion in the same property or controlled linking.
Attribution, multi-touch paths, and incrementality
The last click often overestimates the closing channel; data-driven models distribute value across multiple touchpoints. Incrementality tests (geo holdout, ad cutbacks) help measure the real effect beyond correlations.
Organic / paid cannibalization
Turn off brand terms on Search Ads to observe SEO: a controlled experiment.
Post-view and post-click
Advertising platforms have their own window rules: align them with your internal reporting.
Offline
Physical store: coupon, QR, loyalty program to connect with digital.
Finance alignment
Same definition of “attributed sale” between marketing and management.
MMM
Marketing mix modeling: useful at scale, requires historical data and statistical skills.
Geographic lift
Compare regions with and without campaign: external controls (weather, local competition).
Walled gardens
Each platform optimizes its own metrics: external cross-checking is essential for overall budget decisions.
Incrementality lift tests
Controlled experiments to estimate causality rather than advertising correlation.
Budget calibration
Allocate spending by channel according to estimated elasticity and creative constraints, not only according to fixed historical performance.
First-party data, consent and quality
Choices around consent (CMP) affect data completeness. Server-side tracking and conversion APIs reduce browser signal loss. Quality comes first: poorly named SKUs, duplicate transactions, and duplicated events distort every dashboard.
Governance
Who can create a new event or a new UTM campaign name?
PII
Avoid sending plaintext emails to unauthorized tools.
Data testing
SQL queries or regular exports to detect anomalies (impossible spikes).
GDPR
Purpose, duration, legal basis: alignment with analytics and CRM.
Dark traffic
Unattributed traffic: strengthen tagging and trackable links.
Consent mode
Advanced configuration for modeling when analytics consent is denied.
Access management
Principle of least privilege for warehouses and reports containing personal data.
Anonymization
Aggregates for analysts without the need for a direct customer identifier.
Legal retention
Different durations for raw logs, marketing aggregates, and CRM data: documented deletion schedule.
Cohorts, LTV and retention analyses
Cohorts group users by acquisition date or first order to track repeat purchases over time. LTV links customer lifetime value and CAC. See retention and LTV.
Cohort by channel
Compare the quality of customers coming from social networks versus SEO.
Churn
Subscriptions: cancellation rate and reasons, if available.
Recurring products
Replenishment and consumables: expected purchase frequency.
RFM segments
Recency, frequency, monetary value: classic scoring for campaigns.
Predictive
Purchase probability models: the quality of training data is decisive.
Privacy sandbox
Evolution of advertising identifiers: impact on cross-site measurement.
Device cohorts
First purchase in app vs web: differentiated re-engagement journeys.
Time between orders
Interpurchase distribution to calibrate email follow-ups and inventory.
Star product vs long tail
Analyze profitability by curve: avoid over-investing in acquisition for low-margin SKUs.
Dashboards, reports, and data culture
A good dashboard answers a decision-making question on a single page: not thirty decorative charts. Weekly meetings should start from gaps vs. targets, not from the list of available numbers.
North Star metric
One synthetic indicator per team, evolving with the growth phase.
Self-service
Train countries or categories to extract their own reports without centralizing everything.
Data lineage
Track where each BI column comes from (warehouse, API, file).
Alerting
Thresholds for conversion drops or payment error spikes: fast reaction.
CSV exports
Check decimal separators and time zones before international consolidation.
OKR data
Objectives specific to analytics maturity: event coverage, report availability lead time.
Living documentation
Internal wiki of definitions: reduces tickets like “which number is the source of truth?”
Data warehouse
Star or vault model: architecture choice impacts report freshness and maintenance cost.
Metrics governance
Catalog of official metrics with owner and formula: reduces discrepancies between BI tools.
Shopify, ERP, and BI tool stack
Platforms like Shopify provide native reports and export to BI (Looker, Power BI, etc.). The ERP often holds the accounting source of truth; marketing analytics lives in GA and ad managers: monthly reconciliation is necessary.
Latency
Delay between an order and its appearance in the data warehouse: critical during peak periods.
Tool costs
BI licenses, connectors, storage: a recurring budget to plan for.
Single source of truth
Designate the reference source for official revenue figures.
Automation
Scheduled exports: to be linked to your e-commerce automation strategy (see the dedicated blog guide).
Statistical significance
Do not conclude too quickly from low-traffic A/B tests: use sample size calculators.
Overinterpretation
One-day spike: check technical incidents before concluding about a campaign.
Team silos
Marketing KPIs that conflict with finance margin: reconciliation is mandatory.
Selection bias
Sample of customers volunteering for a survey: do not generalize to the entire customer base without correction.
Common mistakes: vanity metrics and cognitive biases
Hunt down vanity metrics: followers without engagement, sessions without conversion when the goal is sales. Watch out for averages that hide double-peaked distributions, and for spurious correlations (seasonality, inventory).
Optimizing the wrong KPI
High CTR with poor product targeting: low-qualified traffic.
Poorly calibrated comparison
Partial week vs full week, different public holidays between years.
New product effect
A launch artificially inflates a category: normalize the interpretation.
Confirmation bias
Looking for numbers that validate an intuition: challenge it with counter-indicators.
Ops seasonality
Black Friday: plan near real-time dashboards and alert thresholds for payment errors.
NPS and quantitative data
Post-purchase surveys linked to analytics segments to explain scores.
Support SLA
Resolution time by issue type: a service quality indicator coupled with the funnel.
Fulfillment
Link promised and actual delivery times to repurchase rates: cross-reference with logistics indicators (e-commerce fulfillment blog guide).
Link with catalog, orders, and ops
Product analytics combines the catalog (views, listings) and order management (lead times, cancellations: see the blog article on e-commerce OMS). Logistics anomalies appear in cancellation rates or lead times, not only in satisfaction.
Stockouts
Correlate views on the “out-of-stock” page and abandonment rate.
Returns
Return reasons by SKU: feeds product pages and photo shoots.
Customer service
Ticket volume by topic: prioritize site or product fixes.
Experimentation
A/B tests on displayed price or shipping fees: rigorous statistical methodology.
Intent vs page view
Chat reveals questions absent from on-site searches: a signal for SEO content and FAQ.
Conversation exports
Categorize intents to feed the product backlog and self-service content.
Personal data
Pseudonymization in the analytics warehouse if logs contain identifiers.
Product loop
Prioritize catalog fixes when chat questions repeat the same misunderstandings about product pages or policy.
Qstomy, FAQ, summary and sources
Chat logs reveal intentions not covered by pages, checkout friction, and repetition about the return policy. By aggregating these signals with traditional web and transactional analytics, you can prioritize FAQ content and journeys.
Deflection rate
Share of conversations escalating to a human: a KPI for bot relevance.
Conversation satisfaction
Post-chat surveys correlated with journeys and products.
Consistency
Bot responses must reflect the same figures as the site (lead times, stock).
Support
Reducing repetitive tickets frees up human capacity for complex cases.
For the SEO framework and organic measurement, cross-reference with the e-commerce SEO blog guide and landing-page performance.
30-day plan
Week 1: tracking audit and revenue gaps; week 2: document definitions; week 3: decision-making dashboards; week 4: review and alerts: simple milestones to structure data maturity.
To go further on the purchase journey, connect these indicators to funnel design (blog guide) and user testing.
In operational conclusion, treat analytics as a cross-functional function with budget, owner, and a continuous roadmap, not as an occasional Excel export before the executive committee, otherwise decisions are made blindly or too late to correct the trajectory.
Is analytics enough to make decisions? No: it informs; strategy includes supply constraints, brand, and risks.
Should everything be measured? No: prioritize events aligned with decisions; avoid the “instrumentation tax.”
Where to start? Make purchase and revenue reliable, then the checkout funnel, then segments.
Sources
Cited Qstomy articles: Google Analytics and conversion, conversion-rate definitions, e-commerce foundations, conversion funnel, retention, Shopify, catalog.
In summary, e-commerce analytics is a common language between marketing, product, and ops: investing in definitions, data quality, and review cadence delivers more than multiplying tools without governance.
Enzo Garcia
April 8, 2026
