GA4 and the store report different sales
Duplicates, consent, script blocking, currencies, cancellations and refunds change the transaction picture.
E-commerce analytics · revenue · margin · funnel · retention
Connect sales, customers and marketing. See what makes e‑commerce profitable.
I build analytics that shows not only how much the store sold, but why the result changed, where customers drop out, which acquisition channels pay back and what drives repeat purchase.
GA4, Piwik PRO, GTM, BigQuery, Looker Studio, Power BI and advertising platforms are delivery tools. We first agree the decisions, store economics, systems of record and KPI definitions.
The order as the reference pointstatus, payment, refund, discount, cost and net revenue
The customer and pre-purchase funnelsource, device, product, basket, checkout and return
Performance after marketing costROAS in the context of margin, CAC, retention and attribution limits
Signals that tracking is not enough
The data is spread across systems, yet the store result is still hard to explain.
The usual problem is not the absence of another dashboard. It is the lack of an agreed sales definition, a known source for each number and a way to connect customer behaviour to the actual order result.
ROAS looks good, margin does not
Campaign reports omit discounts, product cost, delivery, refunds and category-level profitability.
The funnel drops without a clear cause
It is unclear whether the issue is the device, product, traffic source, delivery or a specific checkout step.
Every purchase looks the same
New, returning and promotion-driven customers are merged into one result without cohorts or retention.
Campaigns lose their context
UTMs, click identifiers and campaign names are inconsistent, while owned channels distort acquisition source.
Reporting remains manual
Marketing, e-commerce and finance merge exports and recreate the same definitions in spreadsheets every week.
E-commerce analytics model
One path from source to decision, with a clearly named system of record.
I do not force every tool to show an identical result. I define source roles, reconciliation rules and tolerances so differences are understood and controlled.
Store, payments, refunds, CRM, media and web analytics
Each system has its own refresh rhythm, identifiers and limitations.
Order, line item, customer, session, channel and calendar
Systems of record, keys, statuses, currencies, costs and KPI definitions.
Commercial result, funnel, channel, product and customer
Decision-level reporting with a path from a variance to its likely cause.
Budget, offer, checkout, retention and test priority
A decision owner, response threshold, next step and outcome check.
The store or ERP reconciles ordersweb analytics explains behaviour, not the sales ledger
Definitions come before the dashboardrevenue, customer and channel cannot change between reports
Reconcile first, optimise seconda decision should not be based on an implementation error
Six decision lenses
Reporting scope follows store economics and management practice.
Not every store needs every area on day one. We prioritise by expected business impact and the quality of available data.
- 01Decision: what truly creates the result
Commercial performance
Orders, gross and net revenue, discounts, cancellations, refunds, average order value, product cost and margin.
- 02Decision: where to increase or reduce budget
Acquisition and profitability
Campaign cost, CAC, ROAS, new-customer share, profitability after media cost and attribution limits.
- 03Decision: which stage needs diagnosis or testing
Funnel and checkout
Product list, product page, basket, checkout start, payment, errors, device, source and delivery option.
- 04Decision: what to promote, improve or retire
Product and merchandising
Visibility, interest, basket additions, conversion, discount, refund, availability and category profitability.
- 05Decision: whom to acquire and how to retain them
Customer, retention and LTV
New and returning customers, cohorts, time to next purchase, frequency, customer value and segments.
- 06Decision: whether the data is trustworthy
Quality and governance
Transaction completeness, identifier consistency, freshness, KPI owners, documentation and change process.
Example cockpit
The report starts with the outcome, then leads to cause and action.
This mock-up shows how information can be organised. It is an illustration, not client data or a promised implementation result.
E-commerce performance cockpitLast 30 days
Market: UKCustomer: allChannel: all
Purchase funnelchange vs previous period
Product sessions214,800+6%
Add to basket43,260+4%
Checkout started24,910-9%
Paid order5,886+3%
Channel and outcomerevenue is not profit
ChannelROASMargin after media
Paid Search6.218.4%
Paid Social4.86.1%
Email / CRM12.728.9%
Organicn/a33.2%
Signal to investigateThe checkout-start decline is concentrated on mobile and products with delivery over three days.
Next step: review availability and delivery-cost communication, then test an improved version for this product group.
Data reconciliation
Every number has a definition, a source and a point when it is safe to use.
GA4 does not need to match the store transaction by transaction. The difference, its scale and the source accountable for each decision must be clear.
AreaSystem of recordPurposeControls
Order and revenueStore or ERPcommercial outcome and fulfilment statusduplicates, currency, discounts, cancellations
Payment and refundPSP, store or ERPsettled revenue and adjustmentsstatus, date and partial refund
Behaviour and funnelGA4 or Piwik PROjourney, context and segmentationconsent, events, parameters and data loss
Campaign costAdvertising platformspend, clicks and deliverycurrency, time zone and campaign structure
Customer and retentionCRM, store or CDPcustomer status, cohorts and valueidentity, consent and change history
Engagement options
The work can start with measurement repair or a complete profitability model.
After a short qualification, I recommend the smallest useful scope. Not every problem requires a warehouse, server-side tracking or a tool migration.
Audit and remediation plan
Assessment of implementation, sources, discrepancies, definitions and reports.
- problem and risk map
- impact-based priorities
- testing and delivery plan
Behaviour and sales measurement
Measurement plan, dataLayer, GA4 or Piwik PRO, tags, consent and QA.
- funnel and e-commerce events
- identifiers and business parameters
- documentation and acceptance
Integration and trading cockpit
Store, campaign and behaviour data, optionally enriched with CRM or cost.
- data model and KPI layer
- dashboards and analysis
- automation and monitoring
Ongoing support and development
Quality control, performance analysis, backlog and collaboration with marketing and IT.
- recurring KPI review
- diagnoses and recommendations
- measurement that evolves with the store
Working process
Decisions and economics first, technology second.
- 01Outcome: decision brief and KPI dictionary
Decisions, KPIs and store model
Audiences, operating rhythm, order, margin, customer and channel definitions, and required cuts.
- 02Outcome: data and gap map
Source inventory
Store, payments, ERP, CRM, consent, web analytics, advertising platforms, identifiers and quality.
- 03Outcome: implementation specification
Measurement and data-model design
Events, dataLayer, keys, statuses, integrations, systems of record, architecture and test plan.
- 04Outcome: working data flow
Implementation and integration
Agreed tool configuration, transformations, cost imports, dashboards and automation.
- 05Outcome: acceptance record and known differences
Validation and reconciliation
Scenario tests, deduplication, order and value checks, refunds, consent, campaigns and tolerances.
- 06Outcome: analytics embedded in the team
Operational adoption
Training, owners, alerts, review agenda, analysis backlog and change-management rules.
Acceptance criteria
The implementation is ready when the team understands data quality and can use the result.
Specific thresholds are agreed before delivery. Platforms and consent create natural differences, so we accept both verified consistency and documented limitations.
Honest limitations
Good measurement reduces uncertainty. It does not remove it.
- GA4 and Piwik PRO do not replace transaction systems.Store, payment or ERP data is needed to reconcile sales, refunds and margin.
- ROAS is not profitability.Campaign revenue must be set against margin, discount, product cost, refunds and acquisition cost.
- Attribution is not proof of causality.A model organises channel contact but does not automatically measure incremental advertising impact.
- Not every journey can be connected.Consent, ad blocking, cross-device use, authentication and platform limits fragment data.
- Server-side and CAPI do not bypass privacy.They may improve control, but still require an appropriate basis, consent and configuration.
- A dashboard does not make the decision.It needs an owner, response threshold, business context and an outcome check.
Business, measurement and the data layer
I connect e-commerce, marketing, implementation and reporting perspectives.
I lead the work from management questions and order economics, through measurement design and integrations, to dashboard acceptance and team adoption. Technology remains a means, not the goal.
20+years of experience
150+analytics and data projects
50+clients
Figures refer to the broader analytics and data practice, not only e-commerce implementations.
Before we talk
Frequently asked questions.
Why does GA4 revenue differ from the store?
GA4 measures browser or app behaviour and depends on consent, script blocking, purchase-event correctness and deduplication. The store knows order status, later cancellations and refunds. The store or ERP normally reconciles performance, while GA4 explains the journey and marketing context.
GA4 or Piwik PRO for e-commerce?
The choice depends on privacy, hosting, advertising integrations, reporting and the organisation's way of working. I do not recommend migration simply because one tool is more popular. Requirements and switching cost come first.
Do we need server-side tagging or Meta CAPI?
Not always. They can improve data-flow control, integration stability and event delivery, but introduce implementation and maintenance cost and do not remove privacy obligations.
Can the store, campaigns and CRM be connected?
Yes, when sources expose stable identifiers and suitable detail. The work needs matching rules, a customer model, access controls and documentation of journeys that cannot be linked.
Can the report include margin and LTV?
Yes, but GA4 alone does not know product cost, refunds or full customer history. Margin and LTV require store, ERP or CRM data and agreed definitions, horizons and cost-allocation rules.
Can we start with an audit only?
Yes. It is often the right first step for a complex implementation or material discrepancies. The audit ends with an impact assessment, priorities and remediation plan without committing to a full implementation.
First step
Start with three decisions and the sources that currently give different answers.
Describe your commerce platform, current measurement, payment system, marketing channels and the reports the team does not trust or still lacks.
Calculate and organise