eCommerce Performance Analytics: What to Track and How to Turn Data Into Growth

Most eCommerce businesses rely on performance data to guide decisions around marketing spend, conversion optimisation, and growth. Metrics such as revenue, sessions, and conversion rate are used to evaluate what is working and where improvements are needed.

However, having access to data does not automatically lead to better decisions. In many cases, teams review metrics individually but struggle to connect them into a clear explanation of what is happening across the business. As a result, effort is often directed toward the wrong areas, or improvements fail to deliver meaningful impact.

eCommerce performance analytics is valuable because it provides a structured way to interpret data. It helps you identify where performance is breaking down, understand what is causing it, and prioritise the actions that will have the greatest commercial impact.

This guide outlines how to approach eCommerce performance analytics in practice, helping you move from raw performance data to clear, prioritised decisions that improve revenue and efficiency. It is based on On Tap’s real-world experience in monitoring and optimising eCommerce stores through a data-driven approach.

eCommerce performance analytics is the process of using data to understand how your business is performing and what is influencing that performance.

The three core layers of eCommerce performance analytics

In practice, eCommerce analytics can be understood through three core layers:

• Measurement - What is happening?
  This involves tracking key metrics such as traffic, orders, and revenue to have a clear view of what is changing and where performance differs from expectations. This typically involves comparing current performance against previous periods, targets, or benchmarks.

• Diagnosis - Why is it happening?
  This layer focuses on explaining the causes behind the observed performance. It involves analysing how different factors contribute to the outcome, such as traffic quality, product performance, conversion behaviour, or customer trends. Rather than looking at metrics individually, this layer connects them to identify what is driving the change.

• Action - What should you do next?
  This layer focuses on how you use insights from data to make decisions. Once you understand the causes, you can decide what to optimise and prioritise based on your business goals and constraints.

A common mistake to avoid

Many businesses focus heavily on measurement, but stop short of understanding why performance changes or what actions to take.

As a result:

• Performance changes are visible, but it is unclear what is actually driving them

• Time and budget are spent on improvements, but results are inconsistent or difficult to explain

• Growth relies on trial and error, rather than clear, data-driven decisions

To get real value from analytics, you need to move beyond tracking metrics and develop a structured approach to interpreting them. The next sections outline the key principles, metrics, and steps that help you do this effectively.

Performance data informs decisions that directly impact revenue and investment. Without a clear way to interpret it, teams risk prioritising the wrong opportunities and acting on misleading signals. The following principles define how to analyse performance in a way that keeps decisions focused, commercially relevant, and actionable.

• Metrics must be tied to business goals: Metrics only become meaningful when viewed in the context of what the business is trying to achieve. Without that anchor, analysis quickly becomes unfocused, with different teams prioritising different numbers.
  For instance, a growth-focused business will read performance through traffic and conversion, while a profitability-focused business will look more closely at margins and order value.

• Understand the difference between a metric and a KPI: A metric describes what is happening, while a KPI is deliberately chosen because it reflects a specific business objective. Without this distinction, teams may track many data points but still lack a clear measure of success.
  For example, if the goal is to improve customer retention, then repeat purchase rate or customer lifetime value would be the KPI. Metrics such as traffic or new customer acquisition may still be monitored, but they do not indicate whether retention is improving. If the team focuses on acquiring more new customers while the repeat purchase rate remains low, activity increases, but the core objective is not being achieved.

• Always read numbers in context: Metrics are only meaningful when interpreted against a reference point, such as historical trends, targets, or comparisons across segments. Without this context, it is difficult to determine whether a number reflects good or poor performance.
  For instance, revenue may remain broadly flat compared to the previous month, which can suggest stable performance at first glance. However, if the business had planned for growth during that period due to seasonality or increased marketing investment, flat revenue indicates underperformance relative to expectations. Without this context, the business may assume performance is acceptable and delay corrective action, allowing the gap to widen over time.

• Data must be reliable before you analyse it: Analysis is only as reliable as the data behind it. Inconsistent tracking or unclear metric definitions can distort performance signals, leading to incorrect conclusions.
  For instance, if add-to-cart events are not captured properly, reported engagement with product pages may appear lower than it actually is. This can lead teams to adjust pricing or redesign product pages, while the underlying issue lies in tracking rather than performance. As a result, time and budget are spent on changes that do not address the real problem.

• Analytics exists to drive decisions, not to generate reports: The purpose of analytics is to inform what should be done next. When reporting becomes the focus, insights are often observed but not translated into action, which limits their value.
  In practice, teams may regularly review performance dashboards and identify issues such as declining conversion rates or underperforming products. However, without clear ownership or follow-up actions, these insights do not lead to change. As a result, the same problems persist over time, opportunities are missed, and performance improvements become slow and inconsistent.

Once you understand how to approach analytics, the next step is to define what to track.

The metrics below give you a complete view of your business performance, from demand generation to conversion and retention. Together, they help you evaluate performance, align teams around the right KPIs, and support more effective analysis in the next steps.

Where to get your data

In practice, these metrics are not stored in a single place. You will typically need to combine data from multiple sources to get a complete view of performance.

Common sources include:

• Website analytics tools (e.g. GA4): Provide visibility into user behaviour, traffic sources, and funnel performance

• eCommerce platform dashboards: Show orders, revenue, product performance, and customer data

• Marketing platforms (ads, email, CRM): Help you understand acquisition performance and campaign impact

Important note: Each source provides only part of the picture. To analyse performance accurately, data often needs to be combined and interpreted together rather than in isolation.

Once you have defined the right metrics, the next step is to interpret them and turn them into clear decisions.

The framework below is designed to move from performance signals → problem identification → root cause → action, so that each step produces a clear output you can act on.

To make this framework more practical, we will walk through a hypothetical scenario and apply each step using real-like data.

Scenario (hypothetical): Glow Store is a Shopify-based skincare brand running Facebook and Google Ads. This month, revenue missed the target despite an increased ad budget. You need to identify why performance declined and what to do next.

Before looking at any metrics, take a step back and clarify your business's context. Many analysis issues start here. Teams apply the same set of metrics across very different business models, or focus heavily on channels they can measure, rather than the ones that actually drive revenue.

Focus on three key areas:

1. How customers typically buy in your industry and business model

How customers buy depends on both the type of product you sell and the context in which the purchase is made. This directly affects how you interpret performance metrics.

Product type influences how quickly decisions are made. Lower-priced, repeat-purchase products (such as cosmetics or FMCG) tend to rely on frequency and retention, while higher-value products (such as furniture or electronics) usually involve longer consideration and comparison.

The buying context also varies depending on your business model. In B2C, purchases are typically made by individuals and completed quickly. In B2B, purchases often involve multiple stakeholders, longer cycles, and more structured decision-making.

Because of these differences, the same metric, such as conversion rate, can represent very different levels of performance depending on how customers buy in your specific context.

2. Where your sales come from

Different channels provide different levels of data visibility and control.

For instance:

• Your website offers full visibility into the customer journey

• Marketplaces often limit what data you can access

• Social commerce platforms may have more complex attribution

As a result, metrics should be interpreted in the context of how customers typically buy in your industry, rather than against fixed expectations.

3. Your business goals and priorities

Your current goal determines how you interpret performance and where you focus your analysis. The same data can lead to very different conclusions depending on what the business is trying to achieve at that time.

For example:

• Growth: The priority is to increase revenue by expanding demand and acquiring new customers. This is often the focus for early-stage or fast-growing businesses investing in acquisitions and scaling their reach,  with attention on metrics such as revenue growth and conversion.

• Profitability: The priority is to improve efficiency and maximise return from existing demand.  This is more common in established businesses where growth is stable, and the focus shifts to cost control and margin improvement, often assessed through metrics such as average order value and revenue per session.

• Retention: The priority is to increase the value of existing customers over time.  This is more relevant for mature businesses. Repeat purchases and long-term value drive growth, reflected in metrics such as repeat purchase rate and customer lifetime value.

Without a clearly defined priority, analysis can become unfocused, with different teams interpreting the same data differently.

Output for this step: A clear context for interpreting your data, including your business model, primary goal, and key revenue drivers.

Output example (Glow Store scenario): Glow Store operates in the skincare industry, where customers often compare products and take time to build trust before purchasing. This leads to longer decision cycles and lower immediate conversion rates.

The business relies on paid acquisition (Facebook and Google Ads), and the current priority is revenue growth.

Performance should therefore be evaluated:

• In the context of longer purchase journeys and lower baseline conversion rates

• Based on how efficiently paid traffic converts into revenue, rather than traffic volume alone

Before analysing performance, confirm that your data accurately represents what is happening in your business. If the data is incorrect or inconsistent, any conclusions drawn from it will be misleading.

How to check if your data is accurate and what to do if issues are found

Output for this step: A dataset you can trust, with no major gaps or inconsistencies.

Output example (Glow Store scenario): Tracking is consistent across sessions, add-to-cart, and purchase events. Revenue and order data match Shopify. The increase in traffic aligns with increased Facebook Ads spend. This confirms that the data is reliable and the issue is not caused by tracking errors.

Start with a high-level view to understand whether your business is performing as expected.

What you focus on here should depend on the goal defined in Step 1.

For example:

• If your priority is growth, focus on revenue, orders, and customer acquisition

• If your priority is profitability, focus on margins, order value, and cost efficiency

• If your priority is retention, focus on repeat purchase rate and customer lifetime value

Compare performance against:

• Previous periods

• Your targets or expectations

Output for this step: A clear statement of whether performance is on track for your primary goal.

Output example (Glow Store scenario): Sessions increased by 17%, but revenue declined, and the AOV also decreased.

This indicates that performance is not meeting the growth objective. Increased traffic is not translating into revenue.

Once you have identified the gap, the next step is to determine its source.

Most performance issues are not spread across the entire funnel. They are usually concentrated in a specific stage, channel, or product. The more precisely you define where the issue occurs, the easier it becomes to identify the root cause in the next step.

Common ways to break this down include:

• By channel: Does the issue affect all traffic sources, or only specific ones?
  A problem isolated to certain channels suggests the issue is linked to acquisition or traffic quality.

• By product or category: Is performance consistent across your catalogue, or concentrated in certain products?
  If only some products underperform, the issue is likely product-specific rather than site-wide.

• By stage in the journey: Where do users stop progressing?
  For example, users may reach product pages but not add items to the cart, or start checkout but not complete it. If the drop is concentrated at a specific stage, the issue is likely occurring there.

• By customer type: Does the issue affect new users, returning customers, or both?
  This helps distinguish between acquisition and retention problems.

Output for this step: A clearly defined problem area.

Output example (Glow Store scenario): A more detailed breakdown shows:

• Facebook Ads primarily drive traffic growth

• Add-to-cart rate dropped from 33% to 25%

• SPF50 sunscreen has high product views but a low conversion rate (0.9%)

• Repeat purchase rate declined from 22% to 15%

These signals suggest that the performance gap is likely concentrated in:

• The product page stage, where fewer users are adding items to their cart

• The returning customer segment, where repeat purchase behaviour has declined

Once you know where the issue is concentrated, the next step is to understand what is causing it.

1. Look for what has changed

Start by checking whether the issue aligns with any recent changes.

For example:

• New campaigns or changes in targeting

• Updates to pricing or promotions

• Changes to product content or positioning

• Adjustments to layout or user flow

If performance drops shortly after a change, that change is often a strong indicator of the cause.

Output at this stage: A shortlist of recent changes that may be linked to the performance issue.

Output example (Glow Store scenario): Traffic increased following higher Facebook Ads spend, suggesting a shift in traffic mix. No major changes to product pages or retention activities are immediately identified.

2. Compare the experience

If no single change stands out, compare what users see and experience in the problem area versus what is working elsewhere. 

The goal is to answer: What is different in execution that could explain the behaviour?

For example:

• Messaging (Pre-click experience vs product page experience): Paid social campaigns highlight discounts, but the product page does not clearly show the same offer. In contrast, search traffic lands on pages that more closely match user intent.

• Product pages (high vs low performing): High-converting products clearly present pricing, benefits, and reviews, while underperforming products have less detailed information or weaker trust signals.

Output at this stage: Key differences between high- and low-performing experiences.

Output example (Glow Store scenario): Paid social campaigns emphasise product benefits and offers, while product pages for high-traffic items such as SPF50 appear less clear in communicating value. High-performing products show stronger use of benefits, clearer pricing, and stronger trust signals.

3. Form a clear, testable hypothesis

Combine your observations into a specific explanation.

A useful hypothesis should be:

• Specific

• Actionable

• Testable

Output at this stage: Initial hypotheses that explain the observed behaviour.

Output example (Glow Store scenario): Two working hypotheses emerge:

• The drop in add-to-cart rate may be linked to how effectively product pages communicate value, particularly for high-traffic products such as SPF50

• The decline in repeat purchase rate may indicate gaps in post-purchase engagement or customer experience

Once you have clear hypotheses about the root causes, the next step is to decide what to fix and how to execute.

At this stage, the goal is not to solve everything at once, but to focus on the changes that will have the most meaningful impact on performance.

1. Evaluate potential impact

Start by assessing which hypotheses are most important to address first.

Ask:

• Which ones are most likely driving revenue loss or missed opportunities?

• If resolved, where would performance improve most significantly?

Tip: You should prioritise those that affect:

• High-traffic areas

• Key conversion steps

• High-revenue products or channels

This ensures your efforts are focused on the parts of the business that matter most.

Output at this stage: A clear prioritisation of hypotheses based on expected impact.

Output example (Glow Store scenario): The hypothesis related to product page effectiveness should be prioritised. It occurs at a high-traffic stage immediately before purchase, making it likely to directly affect revenue.

The retention-related hypothesis remains relevant but is likely to have a less immediate impact and can be addressed after conversion issues.

2. Decide how to act

Each root cause requires a different approach depending on how confident you are in the explanation and how complex the change is.

• High confidence, low complexity: When the cause is clear and the fix is straightforward, act immediately.
  Examples include missing payment methods, incorrect pricing display, or broken UI elements.

• Lower confidence: When the cause is likely but not certain, validate it through testing. Use A/B testing or controlled experiments to confirm the impact.
  This typically applies to areas such as product page layout, messaging, or pricing perception.

• High complexity: When the cause is clear, but the solution requires significant changes, treat it as a project.
  Examples include redesigning the checkout flow, restructuring pricing, or overhauling the user experience.

Output at this stage: A defined approach for each priority.

Output example (Glow Store scenario): Product page improvements can be implemented directly and monitored, as the issue is clear and relatively simple.

Retention improvements should be tested first, as the underlying cause is less certain and may require validation before scaling.

3. Turn decisions into an action plan

Once priorities are defined, translate them into clear actions.

For each selected root cause, define:

• What needs to be improved

• Who is responsible

• Timeline

• Success metric

Depending on your team’s capacity, multiple improvements can be executed in parallel across different functions such as content, design, and development.

Output at this stage: A structured, execution-ready action plan with clear ownership, timeline, and success metrics.

Output example (Glow Store scenario):

A structured approach to eCommerce performance analysis helps you move from identifying issues to making clear, prioritised decisions. By focusing on where performance breaks down and acting on the most impactful opportunities, you can improve revenue without relying on guesswork or scattered optimisation efforts.