Knowledge / Digital Commerce
Digital Commerce Analytics
Commerce performance is a system, not a dashboard.
A digital-commerce business can generate more traffic and still perform worse.
Conversion can improve while margin deteriorates.
A product can attract strong engagement but fail because the offer, availability or transaction path creates friction.
A channel can appear unprofitable while assisting demand that converts elsewhere.
A strong analytics system therefore needs to explain more than what happened.
It needs to connect:
Acquisition → Discovery → Product Evaluation → Intent → Transaction → Retention → Commercial Outcome
Digital Commerce Analytics is a framework for analyzing those stages as one connected commercial system.
The objective is not to maximize every metric.
The objective is to understand where customer behavior is creating value, where value is being lost and which decision should change next.
- 01Acquisition
- 02Discovery
- 03Product Evaluation
- 04Intent
- 05Transaction
- 06Retention
- 07Commercial Outcome
Why ecommerce metrics are easy to misread
Digital commerce produces an enormous amount of measurable activity.
Sessions · Clicks · Product views · Searches · Add-to-carts · Checkout starts · Purchases · Revenue · Returns · Repeat orders · Advertising cost
These metrics are useful.
They are also easy to interpret incorrectly when viewed in isolation.
A lower conversion rate can result from weaker traffic.
It can also result from stronger top-of-funnel reach bringing more exploratory visitors into the site.
A high product-view rate may indicate interest.
It may also indicate that customers are struggling to understand category navigation.
A strong ROAS may reflect effective advertising.
It may also reflect customers who were already likely to purchase.
A low-performing product may have a demand problem.
Or a stock problem.
Or a pricing problem.
Or weak imagery.
Or incomplete product information.
Or the wrong traffic source.
Analytics becomes useful when the measurement is connected to the commercial system around it.
Stage 1 — Acquisition
Acquisition asks:
Who entered the commerce system, and why?
Useful dimensions may include:
- channel
- campaign
- source
- medium
- geography
- device
- landing page
- new versus returning customer
- branded versus non-branded demand
- paid versus organic traffic
- audience type
- campaign intent
Traffic volume alone is rarely enough.
The quality of acquired traffic matters.
A useful acquisition analysis asks:
- Which sources bring customers who continue deeper into the commerce journey?
- Which channels generate product discovery?
- Which sources generate immediate transactions?
- Which channels primarily assist later conversion?
- Are customers landing on the right page for their intent?
- Are high-cost visitors behaving differently from lower-cost visitors?
- Does the business understand the difference between creating demand and harvesting existing demand?
Acquisition should therefore be evaluated by what happens after the click, not only by the click itself.
Stage 2 — Discovery
Once customers arrive, they need to find something relevant.
Discovery can happen through:
- navigation
- category pages
- internal search
- filters
- recommendations
- collection pages
- merchandising
- landing pages
- direct product links
Useful signals may include category engagement, search usage, filter usage, product-list clicks, search refinements, zero-result searches, navigation paths, and the percentage of sessions reaching product detail pages.
Discovery analysis asks:
- Can customers find relevant products efficiently?
- Are important categories receiving visibility?
- Do internal searches reveal unmet demand?
- Are filters helping customers narrow choices?
- Are certain products overexposed while others remain effectively invisible?
- Are campaign landing pages aligned with the promise of the acquisition source?
Discovery sits between traffic and product evaluation.
If this layer fails, demand can exist without ever reaching the right product.
Stage 3 — Product Evaluation
Product evaluation is where the customer begins deciding whether an item deserves further consideration.
Useful signals may include:
- product views
- media engagement
- specification interaction
- variant selection
- size selection
- availability
- reviews
- certification views
- product comparison
- policy interaction
- time on product detail
- product-page exits
- related-product clicks
The purpose is not to assume that more engagement is always better.
Long time on a product page may indicate careful evaluation.
It may also indicate confusion.
A high exit rate may reflect weak product evidence.
It may also reflect an intentionally informative page serving customers early in their decision process.
This stage should therefore be interpreted alongside product context.
Questions include:
- Does the customer have enough information to evaluate the product?
- Are commercially important attributes visible?
- Are images matched correctly?
- Is the offer understandable?
- Are relevant policies accessible?
- Does availability create friction?
- Are customers repeatedly seeking information that should already be present?
This stage connects directly to the Product Data Spine and the Trust Architecture Review.
Stage 4 — Intent
Intent represents behavior suggesting that the customer is moving from evaluation toward commitment.
Possible intent signals include:
- add to cart
- wishlist
- shortlist
- inquiry
- appointment request
- financing interaction
- size selection
- shipping estimate
- product comparison
- return to the same product
- repeated category visits
- contact initiation
- save-for-later activity
Not all commerce businesses express intent through the same action.
A low-value commodity purchase may move directly from product view to cart.
A high-value or customized product may generate an inquiry long before a transaction.
This means intent should be defined according to the commercial model.
A useful analysis asks:
- Which behaviors reliably precede purchase?
- Which actions are merely engagement?
- Where does the customer hesitate after expressing intent?
- Which products generate strong intent but weak transactions?
- Does the site support the appropriate next step for the type of purchase?
Intent is often where analytics begins to reveal the difference between product interest and commercial readiness.
Stage 5 — Transaction
Transaction analysis examines what happens once the customer attempts to complete the purchase.
Useful signals may include:
- cart creation
- checkout start
- checkout progression
- payment method
- shipping selection
- coupon usage
- tax calculation
- payment failure
- abandonment
- completed order
- average order value
- units per transaction
- discount depth
A transaction funnel can identify where customers disappear.
But disappearance does not explain why.
Checkout abandonment may result from unexpected shipping, delivery timing, tax, payment failure, policy uncertainty, technical error, required account creation, discount expectations, or simple customer hesitation.
The metric identifies the location of the problem.
Diagnosis requires additional evidence.
This is why Digital Commerce Analytics should connect transaction data with customer behavior, technical performance, commercial terms, service questions, and qualitative evidence.
Stage 6 — Retention
The commercial relationship continues after the first transaction.
Retention may be observed through:
- repeat purchase
- time to second order
- reorder behavior
- category migration
- customer lifetime activity
- email engagement
- service interaction
- product replenishment
- referrals
- returns
- cancellations
Retention should not be reduced to a single repeat-purchase percentage.
Different product categories naturally produce different purchase frequencies.
A customer buying a frequently consumed product behaves differently from a customer buying a high-value durable product.
A useful retention analysis asks:
- What behavior is reasonable for this category?
- Which first purchases are associated with later customer value?
- Which channels acquire customers who return?
- Do customers move into additional categories?
- Are service failures damaging future purchase behavior?
- Are returns part of healthy commerce behavior or evidence of product mismatch?
Retention needs category context.
Without that context, the same metric can be interpreted very differently.
Stage 7 — Commercial Outcome
The final stage asks whether customer activity created economically useful outcomes.
Revenue is important.
It is not the entire outcome.
Commercial analysis may also require:
- gross margin
- contribution margin
- acquisition cost
- discount cost
- shipping cost
- return cost
- payment fees
- inventory exposure
- average order value
- repeat value
- assisted sales
- fulfillment cost
A campaign can generate revenue while destroying contribution.
A promotion can increase conversion while training customers to wait for discounts.
A high-volume product can consume working capital while producing weak margin.
A lower-volume product may contribute disproportionate profit.
The correct financial depth depends on what data is available.
Digital Commerce Analytics should never pretend to know margin if only revenue is available.
A decision-first diagnostic model
When performance changes, avoid jumping directly from metric to tactic.
Use a simple diagnostic sequence:
- 01ObservationWhat changed?
- 02LocationWhere in the commerce journey did it change?
- 03SegmentWhich customers, products, channels or devices are affected?
- 04Plausible causesWhat could reasonably explain the change?
- 05EvidenceWhat supports or weakens each explanation?
- 06DecisionWhat should change?
- 07VerificationDid the change produce the expected result?
This diagnostic sequence connects Digital Commerce Analytics with the broader Marketing Analytics framework.
Product-level analysis
Site-wide averages can hide important product behavior.
Commerce analysis should often move to the product level.
For each important product or product family, useful questions may include:
- How much qualified traffic reaches it?
- How often is it discovered?
- How often is it viewed?
- Does it generate intent?
- Does intent convert?
- Is it available?
- Is the customer selecting the correct variation?
- Is pricing competitive within the business’s positioning?
- Are returns unusually high?
- Does the product attract new customers or existing customers?
- Does it contribute margin?
- Does it assist other purchases?
DiscoveryEvaluationIntentTransactionEconomicsProduct performance
A product can underperform at any one of these stages.
That makes “low sales” a result, not a diagnosis.
Traffic quality and experience quality should be separated
One of the most common analytical mistakes is blaming the site for traffic problems—or blaming marketing for experience problems.
Suppose conversion falls.
- Explanation ATraffic quality weakened.
- Explanation BThe customer experience weakened.
- Explanation CThe product mix changed.
- Explanation DAvailability deteriorated.
- Explanation EA promotion ended.
- Explanation FMeasurement changed.
A useful diagnostic therefore separates:
Who arrived? from What happened after they arrived?
Traffic quality
Intent · Source · Audience · Landing-page fit · Device · Geography
Experience quality
Discovery · Product evidence · Offer · Trust · Availability · Checkout · Service
Commercial result
Transaction · Revenue · Margin context · Retention
Assisted conversion matters
Customers do not always purchase in the same session or through the same channel that created demand.
A customer may discover the business through social media, return through organic search, research several products, visit again directly, contact the business, and complete the purchase later.
A last-click view may credit only the final interaction.
That does not mean earlier interactions had no value.
Digital Commerce Analytics should distinguish between direct conversion, assisted behavior, discovery activity, and closing activity.
This does not require pretending that attribution can perfectly reconstruct customer causality.
It requires recognizing that commercial journeys can span multiple sessions and channels.
Merchandising is measurable
Merchandising decisions shape what customers have the opportunity to discover.
Analytics can examine:
- product position
- category placement
- sorting
- collection exposure
- recommendation visibility
- search ranking
- promotional placement
- homepage exposure
- inventory availability
If one product receives five times the visibility of another, sales comparisons between them should be interpreted carefully.
A product may appear to outperform because it was given more opportunities to be seen.
This creates an important analytical distinction:
Demand versus Exposure
The two are related but not identical.
A strong commerce analysis asks:
- What did customers choose?
- What were customers actually given the opportunity to choose from?
Conversion rate needs context
Conversion rate is useful because it compresses customer behavior into a simple relationship between visits and transactions.
It is also one of the easiest commerce metrics to misuse.
Conversion can vary because of:
- traffic source
- device
- geography
- new versus returning customers
- brand demand
- product availability
- pricing
- promotion
- product category
- season
- purchase complexity
- page speed
- checkout friction
- measurement changes
A site-wide conversion rate therefore answers only a broad question.
More useful analysis often asks:
- Conversion for whom?
- Conversion from what source?
- Conversion on which product?
- Conversion under what offer?
- Conversion during what availability condition?
- Conversion compared with what period?
Context turns conversion rate from a score into a diagnostic signal.
A worked digital-commerce example
Consider an ecommerce business where revenue falls 15% month over month.
A superficial conclusion might be: “We need more traffic.”
The commerce system analysis begins differently.
- Acquisition→ traffic declined only 2%
- Discovery→ category-to-product progression remained stable
- Product Evaluation→ product views remained stable
- Intent→ add-to-cart rate declined materially on several high-volume products
- Transaction→ checkout completion remained approximately unchanged
- Retention→ repeat behavior remained stable
- Commercial Outcome→ revenue declined because fewer customers reached purchase intent
The analysis then moves to product-level evidence.
Several high-volume products became unavailable in popular variations.
Customers continued reaching the product pages.
They simply could not select the variation they wanted.
The problem was not primarily traffic.
It was availability affecting intent.
The decision might therefore involve inventory, merchandising, alternative recommendations, availability communication, or acquisition routing.
The example illustrates the purpose of Digital Commerce Analytics:
identify where performance changed before deciding how to respond.
A Digital Commerce Analytics review
A practical review can begin with sixteen questions.
- 01Which acquisition sources are bringing customers into the commerce system?
- 02Which sources bring customers who continue beyond the landing page?
- 03Can customers efficiently discover relevant products?
- 04Which searches, categories or filters reveal demand?
- 05Which products receive the most exposure?
- 06Which products generate the strongest evaluation behavior?
- 07Which behaviors represent meaningful commercial intent?
- 08Where does intent fail to become transaction?
- 09Are checkout losses concentrated by device, payment method, geography or offer?
- 10Which products generate revenue but weak commercial value?
- 11Which customer groups return or purchase again?
- 12Which acquisition sources generate stronger downstream customer value?
- 13Are merchandising decisions affecting product exposure?
- 14Are analytics comparing demand fairly when exposure differs?
- 15Could a measurement change explain part of the observed performance change?
- 16What decision should change because of the analysis?
The final question is the most important.
If the analysis does not change a decision, the measurement may still be interesting—but it has not yet become operationally useful.
Common analytical failures
- 01
Optimizing traffic before understanding downstream performance
More sessions are treated as the solution before identifying whether the actual constraint is discovery, product, availability, offer or checkout.
- 02
Treating averages as customers
Site-wide metrics hide differences between channels, devices, products and customer groups.
- 03
Treating correlation as explanation
Two metrics move together and are assumed to have a causal relationship without sufficient evidence.
- 04
Using revenue without economic context
A campaign or product looks strong because revenue increased while discounting, returns or acquisition cost weakened the commercial outcome.
- 05
Ignoring merchandising exposure
Products are compared without accounting for how often customers had the opportunity to see them.
- 06
Assuming the last click created the demand
Closing channels receive all credit while discovery and assisted interactions disappear from the narrative.
- 07
Treating tracking as reality
Analytics data is assumed to be complete even when consent, implementation, blockers, device changes or attribution limitations affect measurement.
- 08
Reporting without deciding
Dashboards become more sophisticated while the business remains unclear about what action should change.
Measurement architecture matters
A commerce analysis is only as reliable as the measurement supporting it.
Useful measurement may require consistent definitions for:
- session
- customer
- product view
- product identifier
- variant
- cart
- checkout
- transaction
- revenue
- discount
- return
- channel
- campaign
- customer status
If different systems interpret those concepts differently, the analysis can become internally inconsistent.
This is particularly important when commerce data crosses website analytics, advertising platforms, commerce platforms, CRM systems, inventory systems, email platforms, and financial records.
Measurement architecture should therefore answer:
- What is being measured?
- Where is it measured?
- Which system owns the definition?
- How is the identifier matched across systems?
- What is known to be incomplete?
Applied context: high-consideration digital commerce
Digital-commerce analytics becomes particularly interesting when customers do not move directly from advertisement to purchase.
High-consideration categories may involve multiple visits, product comparison, education, contact, appointments, financing, customization, availability questions, or offline-assisted conversion.
Luxury jewelry provides one example.
A customer may discover a product through advertising, return through search, review specifications, compare products, ask a question and complete the transaction later.
That journey cannot be understood through a single click metric.
My professional work supporting digital commerce in luxury jewelry (see Nuit D’Or — Selected Work) has reinforced the importance of connecting traffic, product information, customer behavior, trust, transaction, and commercial outcome.
The same analytical principle applies beyond jewelry wherever the purchase requires meaningful evaluation before commitment.
Limitations
Digital Commerce Analytics cannot observe every customer thought.
It cannot prove causality from behavioral data alone.
It cannot reconstruct every cross-device or offline journey.
Attribution remains incomplete.
Consent choices can reduce observable behavior.
Advertising platforms may report data differently from analytics systems.
Commerce systems may apply different revenue definitions.
Returns may occur after the original reporting period.
Margin data may be unavailable.
Customer identity may remain fragmented.
And measurement implementation can itself contain errors.
The framework therefore separates what is measured, what is inferred, and what is decided.
The purpose is not perfect certainty.
The purpose is better commercial decisions from imperfect but structured evidence.
Related frameworks and systems
- Marketing AnalyticsA decision-first framework connecting measurement, evidence, diagnosis, action and verification.
- Product Data SpineA framework connecting source evidence, governed product information, media, customer experience, distribution and measurement.
- Trust Architecture ReviewA method for evaluating whether customers have credible evidence across identity, product, offer, service, transaction and post-purchase experience.
- Nuit D’Or — Selected WorkApplied digital-commerce work involving product architecture, customer experience and measurement within luxury jewelry.
Part of Knowledge · See also Methods, Research and Selected Work.
The useful question is not what the dashboard says.It is what the business should do differently because of it.
Digital commerce produces thousands of measurable interactions.
The analytical challenge is connecting those observations into an explanation of how customers move through the commercial system.
Acquisition creates opportunity.
Discovery creates access.
Product evaluation creates understanding.
Intent signals readiness.
Transaction creates exchange.
Retention extends the relationship.
Commercial outcome determines whether the system is creating durable value.
Digital Commerce Analytics connects those stages so measurement can support decisions rather than simply describe activity.