Learning Objectives
By the end of this lesson, the learner should be able to:
- Explain the purpose of web analytics in e-commerce.
- Distinguish between users, sessions, events, dimensions and metrics.
- Explain how digital customer journeys can be measured.
- Analyse acquisition, engagement and conversion behaviour.
- Explain funnel analysis and customer journey analysis.
- Apply segmentation to web and customer analytics.
- Interpret behavioural patterns across devices and channels.
- Explain cohort and retention analysis.
- Identify limitations of web analytics data.
- Use analytical findings to recommend e-commerce improvements.
8.2.1 Introduction to Web Analytics
Web analytics is the systematic collection, measurement, analysis and interpretation of data about website or application activity.
In e-commerce, web analytics helps organizations understand:
- who visits;
- how visitors arrive;
- what they view;
- what they search for;
- what they click;
- where they leave;
- what they purchase;
- which products perform well;
- how customers move through the buying journey.
Modern analytics platforms can combine website and application information to provide a broader view of customer behaviour. Google Analytics, for example, provides reporting that can be used to monitor traffic, investigate user activity and understand business performance.
8.2.2 The Digital Customer Journey
A simplified e-commerce customer journey may be represented as:
Awareness → Acquisition → Engagement → Consideration → Cart → Checkout → Purchase → Retention → Advocacy
Each stage produces different analytical questions.
Awareness
How did potential customers discover the business?
Acquisition
Which channels brought them to the website?
Engagement
What did they do after arriving?
Consideration
Which products or categories did they examine?
Cart
Which products were added?
Checkout
Where did customers abandon the transaction?
Purchase
Which customers completed transactions?
Retention
Did customers return and purchase again?
Advocacy
Did customers recommend, review or share the business?
Web analytics allows managers to measure these stages rather than viewing the customer journey as a single event.
8.2.3 Users, Sessions and Events
Three fundamental concepts are important.
Users
A user represents an individual or device/account identity as defined by the analytics system’s measurement and identity rules.
Sessions
A session represents a period of user interaction with a website or application under the analytics platform’s session definition.
Events
An event records a specific interaction.
Examples include:
- product view;
- search;
- add to cart;
- begin checkout;
- purchase;
- refund;
- login;
- sign-up.
Modern Google Analytics e-commerce measurement uses events and associated parameters to measure customer shopping behaviour.
8.2.4 Dimensions and Metrics
Analytics systems generally distinguish between dimensions and metrics.
Dimension
A descriptive attribute used to categorize data.
Examples:
- country;
- device category;
- product;
- traffic source;
- campaign;
- browser;
- customer type.
Metric
A numerical measurement.
Examples:
- users;
- sessions;
- purchases;
- revenue;
- item quantity;
- engagement rate.
Example
Suppose management wants to know:
How much revenue came from mobile users?
Dimension:
Device category = Mobile
Metric:
Revenue
The combination creates a meaningful analytical view.
8.2.5 Acquisition Analytics
Acquisition analytics examines how users arrive at an e-commerce platform.
Common acquisition sources include:
- organic search;
- paid search;
- social media;
- email;
- direct traffic;
- referral websites;
- affiliate channels;
- display advertising.
The important question is not merely:
Which channel produces the most traffic?
It is:
Which channel produces valuable customers?
A channel may generate large volumes of visitors but few purchases.
Another channel may generate fewer visitors but substantially higher customer value.
8.2.6 Engagement Analytics
Engagement analysis examines what customers do after arriving.
Examples include:
- product views;
- search activity;
- category browsing;
- video interaction;
- content engagement;
- add-to-cart activity;
- return visits.
High engagement does not necessarily mean high commercial performance.
For example:
Customers may spend considerable time reading product information because they are interested.
But they may also spend considerable time because:
The website is difficult to navigate.
Managers therefore need to interpret engagement in context.
8.2.7 Product Analytics
Product-level analytics allows management to determine which products are:
- frequently viewed;
- frequently added to cart;
- frequently purchased;
- generating high revenue;
- generating high margins;
- generating high returns.
Consider four products:
|
Product |
Views |
Purchases |
Revenue |
|
A |
100,000 |
5,000 |
KSh 10M |
|
B |
20,000 |
4,000 |
KSh 12M |
|
C |
80,000 |
800 |
KSh 4M |
|
D |
10,000 |
2,500 |
KSh 7M |
Product A has the highest number of views.
But Product D has a much stronger relationship between views and purchases.
Product B also generates substantial revenue despite having much lower traffic.
This illustrates why product views alone are insufficient.
8.2.8 Funnel Analytics
A conversion funnel represents the sequence of steps customers follow toward a desired outcome.
A simplified e-commerce funnel might be:
Product View → Add to Cart → Checkout → Payment → Purchase
Suppose:
- 100,000 product views;
- 20,000 add-to-carts;
- 10,000 checkouts;
- 7,000 purchases.
The largest potential problem may occur at a particular transition.
Product-view-to-cart rate
20,000100,000×100=20%\frac{20,000}{100,000}\times100=20\%
Checkout-to-purchase rate
7,00010,000×100=70%\frac{7,000}{10,000}\times100=70\%
The organization should investigate the stages where disproportionate customer loss occurs.
Google Analytics provides an e-commerce checkout journey report showing the number and percentage of users progressing through checkout steps.
8.2.9 Open and Closed Funnels
A closed funnel requires users to enter at a specified first stage.
An open funnel can allow users to enter at different stages.
This distinction matters because different funnel designs answer different questions.
For example:
Closed funnel
Of users who started checkout, how many completed payment?
Open funnel
How do users who enter at different points eventually reach purchase?
Analysts should select the funnel design based on the business question.
8.2.10 Customer Journey Analysis
Customer journey analysis examines how customers move across multiple interactions and channels.
For example:
Instagram advertisement → mobile website → product page → exit → Google search → website return → email → purchase
A simplistic analysis might attribute the sale to the final email.
A broader journey analysis recognizes that several touchpoints may have influenced the purchase.
This introduces the challenge of attribution.
8.2.11 Attribution
Attribution is the process of assigning credit for a conversion to one or more marketing interactions.
Common approaches include:
First-touch attribution
Credit is assigned to the first known interaction.
Last-touch attribution
Credit is assigned to the final interaction.
Linear attribution
Credit is distributed relatively evenly across selected interactions.
Position-based approaches
Greater weight is assigned to particular stages, such as first and last interaction.
Data-driven attribution
Statistical or algorithmic methods are used to estimate the contribution of different interactions.
No attribution model perfectly describes customer decision-making.
Therefore, management should avoid treating attribution outputs as absolute truth.
8.2.12 Customer Segmentation
Segmentation divides customers or users into meaningful groups.
Possible segments include:
- new customers;
- returning customers;
- high-value customers;
- low-value customers;
- mobile users;
- desktop users;
- geographic markets;
- product-interest groups;
- acquisition channels;
- frequency-based customer groups.
Segmentation can reveal patterns hidden in aggregate data.
For example:
Overall conversion = 2.5%
But:
- returning customer conversion = 6.5%;
- new customer conversion = 1.1%.
This suggests different strategic priorities for the two groups.
8.2.13 Behavioural Segmentation
Behavioural segmentation focuses on actions rather than simply demographic characteristics.
Examples:
High-intent users
- repeated product views;
- product comparisons;
- add-to-cart activity;
- checkout initiation.
Browsers
- many product views;
- low cart activity;
- repeated category exploration.
Loyal customers
- frequent purchases;
- high order frequency;
- long-term engagement.
At-risk customers
- previously active;
- declining purchases;
- reduced engagement.
Behavioural segmentation is particularly useful for personalization and retention strategies.
8.2.14 Cohort Analysis
A cohort is a group of users sharing a defined characteristic, often a common starting period or event.
For example:
- January customers;
- February customers;
- customers acquired through Campaign A;
- customers making their first purchase during a particular quarter.
Cohort analysis allows management to compare how different customer groups behave over time.
Example
|
Cohort |
Month 1 Retention |
Month 3 Retention |
Month 6 Retention |
|
January |
40% |
25% |
18% |
|
February |
45% |
30% |
24% |
|
March |
50% |
36% |
29% |
The March cohort appears to be showing stronger retention.
Management can then investigate:
What changed in the customer acquisition or onboarding experience?
8.2.15 Retention Analytics
Retention analysis measures the ability of an e-commerce business to keep customers engaged and purchasing.
Useful measures include:
- repeat purchase rate;
- customer retention rate;
- purchase frequency;
- time between purchases;
- churn rate;
- active customer rate.
Retention is strategically important because acquiring an existing customer again can be economically different from acquiring a completely new customer.
8.2.16 Customer Churn
Churn occurs when customers cease engaging with a business according to a defined business rule.
A simple churn rate can be represented as:
Churn Rate=Customers LostCustomers at Start of Period×100Churn\ Rate = \frac{Customers\ Lost}{Customers\ at\ Start\ of\ Period}\times100
However, the definition of “lost” must be appropriate to the business.
For a subscription business, churn may mean cancellation.
For a traditional retailer, churn may mean no purchase for a specified period.
8.2.17 Device and Platform Analytics
Customer behaviour may differ across:
- smartphones;
- tablets;
- desktops;
- mobile applications;
- different operating systems;
- different browsers.
Suppose:
|
Device |
Traffic |
Conversion |
|
Desktop |
30% |
5.2% |
|
Mobile |
65% |
1.9% |
|
Tablet |
5% |
2.7% |
The business receives most traffic from mobile but converts mobile users at a much lower rate.
Potential causes may include:
- slow loading;
- poor navigation;
- difficult forms;
- payment problems;
- unsuitable product presentation;
- intrusive pop-ups;
- technical bugs.
Analytics therefore creates a starting point for investigation.
8.2.18 Geographic Analytics
Geographic analysis can identify performance differences across:
- countries;
- regions;
- cities;
- delivery zones;
- markets.
For example:
A business may discover that:
Country A generates high traffic but low purchases.
while:
Country B generates less traffic but much higher revenue per customer.
Management should then investigate:
- payment options;
- delivery costs;
- localization;
- currency;
- language;
- product relevance;
- customer trust;
- regulations.
8.2.19 Search Analytics
Internal site search can provide valuable information about customer intent.
Suppose customers frequently search for:
“black office chair”
but the store does not provide relevant results.
The business may lose sales even though customers have clearly expressed purchase intent.
Search analytics can identify:
- frequently searched terms;
- searches producing no results;
- searches leading to purchases;
- search refinements;
- product demand;
- content gaps.
This makes internal search data both a customer insight source and a merchandising tool.
8.2.20 Customer Analytics and Personalization
Analytics can support personalization.
Examples include:
- recommended products;
- personalized content;
- relevant promotions;
- product recommendations;
- targeted email campaigns;
- customized onboarding.
However, personalization should be implemented responsibly.
Organizations should consider:
- transparency;
- privacy;
- data minimization;
- security;
- fairness;
- customer expectations.
The objective should be to improve customer relevance without creating unacceptable privacy or ethical risks.
8.2.21 Web Analytics and Data Privacy
Analytics may involve:
- identifiers;
- browsing behaviour;
- purchase history;
- location;
- device information;
- customer preferences.
Organizations therefore need appropriate governance.
Key considerations include:
- What data is collected?
- Why is it collected?
- How long is it retained?
- Who can access it?
- How is it protected?
- Is the use consistent with applicable law and organizational policy?
- Can customers understand the relevant practices?
The NIST Privacy Framework provides a structured approach to identifying and managing privacy risks.
8.2.22 Limitations of Web Analytics
Analytics data should not be treated as a perfect representation of reality.
Potential limitations include:
Tracking gaps
Events may not be implemented correctly.
Ad blockers and browser restrictions
Some user activity may not be measured.
Identity challenges
The same customer may appear as different users across devices.
Attribution limitations
Multiple interactions may influence a purchase.
Bot traffic
Automated activity can distort traffic measurements.
Sampling or platform limitations
Some analytical systems may impose data or reporting constraints.
Privacy restrictions
Privacy-preserving technologies can limit the availability of certain information.
Therefore:
Analytics measures observed behaviour under a measurement system; it does not provide an infallible record of every customer action.
8.2.23 From Analytics to Action
Analytics becomes valuable when findings produce decisions.
Consider:
Mobile traffic = 70%
Mobile conversion = 1.5%
Desktop conversion = 5.0%
A weak analytical conclusion would be:
Mobile conversion is low.
A stronger conclusion is:
Mobile is the organization’s largest traffic source but substantially underperforms desktop in conversion; therefore, management should investigate mobile page speed, navigation, product presentation and checkout usability.
The second statement leads to action.
8.2.24 Case Study: The Mobile Conversion Problem
An international fashion retailer discovers:
- 75% of website sessions are mobile;
- mobile product views are high;
- mobile add-to-cart rate is moderate;
- mobile checkout abandonment is extremely high;
- desktop checkout abandonment is significantly lower.
Further analysis shows that the mobile checkout requires customers to enter extensive information manually.
Management interpretation
The problem may not be customer interest.
The evidence suggests that:
Customers are reaching checkout but experiencing friction during transaction completion.
Possible interventions
- simplify checkout;
- improve mobile forms;
- introduce appropriate digital payment options;
- reduce unnecessary fields;
- improve page speed;
- provide clearer delivery information;
- test alternative checkout designs.
Management should then conduct controlled testing and monitor:
- checkout completion;
- conversion rate;
- transaction value;
- payment failure;
- customer satisfaction.
8.2.25 Best Practices for Web and Customer Analytics
- Define analytical questions before collecting data.
- Establish consistent event definitions.
- Maintain accurate product and transaction data.
- Analyse the complete customer journey.
- Segment important metrics.
- Use funnel analysis to identify customer drop-off.
- Use cohort analysis to understand retention.
- Compare device and channel performance.
- Investigate anomalies rather than assuming their cause.
- Combine quantitative and qualitative evidence.
- Treat attribution as an analytical model rather than absolute truth.
- Validate important tracking implementations.
- Protect customer privacy.
- Convert findings into measurable actions.
- Continuously evaluate whether analytics supports business objectives.
Lesson 8.2 Summary
Web and customer analytics enables e-commerce organizations to understand how customers discover, navigate, evaluate and purchase products.
Important analytical concepts include:
- users;
- sessions;
- events;
- dimensions;
- metrics;
- acquisition;
- engagement;
- funnels;
- customer journeys;
- attribution;
- segmentation;
- cohorts;
- retention;
- churn.
Modern e-commerce analytics can capture detailed interactions such as product views, add-to-cart events, checkout activity, purchases and refunds when appropriate measurement is implemented.
The central management principle is not to ask merely:
“What happened?”
Effective analytics should progress toward:
“Why did it happen?”
and ultimately:
“What should we do about it?”