Setting up E-commerce Analytics

What e-commerce events should be transmitted?

The set of events depends on the store's functionality, but the standard GA4 ecommerce model already includes recommended names for key actions. Using standard events simplifies reporting and makes the data structure clearer for specialists who will maintain analytics later.

The event should fire at the moment that corresponds to a real user action. For example, purchase is sent after a confirmed purchase, and begin_checkout is associated with the start of the checkout process. Simply sending all events on page load will result in an incorrect funnel.

In addition to the event name, the parameters and the items array are checked. Without the correct product, ID, price, or currency, some reports will be incomplete, even if the event itself is displayed in GA4.

View products

To view a product list, use view_item_list. This event can be triggered when opening a category, a recommendation block, or another list where the user sees multiple items. Parameters within the items array allow you to determine which products were shown in a specific list.

select_item records the selection of a product from the list. After clicking the product card, view_item is typically passed, which describes the viewing of a specific item. This sequence helps evaluate the path from the product's display in the catalog to further interaction.

If a store has multiple types of product blocks, it's useful to pass list information. This allows you to compare the standard catalog, recommendations, popular product blocks, and other elements without creating separate custom events.

Working with the basket

add_to_cart is sent when an item is actually added to the cart. The event should contain the selected item, price, quantity, and other available parameters. If the user changes the quantity of an item after it's already in the cart, the event routing logic should be agreed upon with the developer in advance.

remove_from_cart is used when deleting an item, and view_cart records the cart view. These actions help understand how many users interact with the cart before moving on to the next step.

Errors often occur on sites where adding items is done via AJAX or components without a page reload. In such cases, the event should be bound to a successful change in the cart state, not to the button click itself.

Placing an order

The begin_checkout event marks the start of checkout. It makes sense to dispatch it after the user proceeds to checkout or when the first full step of the form is opened. The specific rule depends on the site structure.

add_shipping_info records the addition of shipping information, and add_payment_info is associated with the selected payment method. These events allow you to build a detailed funnel within the checkout and see which step is most often abandoned.

If the checkout is located on a single page, events can be dispatched upon successful completion of the corresponding form step. For a multi-step form, the logic is usually tied to the transitions between steps.

Purchase and return

Purchase is considered the primary transaction event. Along with it, the transaction_id, value, currency, and items array are passed. Depending on the implementation, tax, shipping, and promo code may also be passed.

A unique transaction_id is especially important to protect data from being counted twice for the same purchase. If the user refreshes the thank-you page and the event is sent again without proper deduplication, the revenue report may be inflated.

Refunds are used for returns. The structure of the transfer depends on whether a full or partial refund is being processed. The logic should be aligned with the actual order processing process in the store.

What product and transaction parameters need to be passed?

Parameters add specific values to an event. The system understands that a purchase occurred via the purchase key, and the parameters explain what transaction took place, the amount involved, and the items included in the order.

For e-commerce, identifier stability is especially important. The same product should be transmitted identically during product view, add-to-cart, and purchase. If the item_id changes between stages, subsequent analysis becomes less reliable.

Before launch, a mapping table of site fields and GA4 parameters is typically created. It records the source of each value, its format, and the sending conditions.

Transaction parameters

transaction_id stores a unique order identifier. It helps distinguish purchases from each other and is used when processing resubmissions. The value must match the actual transaction identifier in the store's system.

Value conveys the event cost, and Currency specifies the currency. For accurate analytics, these values must comply with the project rules. Tax, shipping, and coupon can also be used.

Before launching, it's a good idea to test several orders with different cart contents. This will make it easier to spot errors that don't appear when testing a single product without a discount or shipping.

Product parameters

Product parameters are passed within the items array. Basic parameters include item_id, item_name, price, and quantity. Item_brand, item_category, item_variant, and other recommended fields are also available.

A stable item_id simplifies matching actions for a single product. item_name should convey a clear name, and item_category helps create catalog slices. For variable products, it's useful to use item_variant.

The items array is checked at every important event. If a product is present in the add_to_cart but is missing in the purchase, the total revenue can be accounted for separately from the product details.

How can I set up e-commerce?

Ecommerce setup depends on the platform and the current website architecture. On some projects, the data is already available in the dataLayer, so most of the work is done in GTM. On others, the website will need some additional work, as the browser doesn't receive the information needed to trigger the correct event.

Before choosing a method, it's helpful to check the existing implementation. Sometimes a site runs a CMS plugin, an old Google Tag Manager container, and additional developer code simultaneously. This setup creates duplicates and complicates troubleshooting.

Once you've chosen your approach, you need to maintain consistent rules for events and parameters. Field names cannot be changed from page to page if they describe the same action or entity.

Via Google Tag Manager and dataLayer

Google Tag Manager is often used as an intermediary layer between a website and GA4. The developer passes data to the dataLayer, after which GTM reads the required values through variables and sends an event to Google Analytics 4.

This approach is convenient when ecommerce data is already prepared in a structured format. Setting up Google Analytics ecommerce via GTM typically involves tags, triggers, variables, and separate validation of each event.

GTM itself doesn't create missing order or product data. If the dataLayer doesn't contain the required values, you'll need to modify your website or use a different data source.

Via gtag.js

Direct configuration via Google Tag and gtag.js is suitable for projects where it's more convenient to send events directly from the website code. The developer triggers the desired event when the user performs an action and passes parameters along with it.

This approach requires careful maintenance during subsequent changes to the checkout, catalog, or frontend. If some events are handled through GTM and others through direct code, this architecture must be documented and any duplication checked.

When choosing a method, consider the current project stack and the team that will be responsible for future analytics. There is no single approach that suits every store.

Through CMS and platform integrations

Shopify, WooCommerce, Magento, OpenCart, and other platforms offer ready-made integrations, plugins, and extensions for transferring e-commerce data. These can significantly reduce the amount of manual development required for a standard store structure.

After installing the integration, testing is still required. You need to check events, the items array, transaction_id, value, currency, product quantity, and the absence of duplicates. The CMS version, theme, third-party checkout, and additional plugins can affect the results.

If the standard integration does not cover the required funnel or additional parameters, it is supplemented through GTM, code, or a separate technical solution.

Setting up Google Analytics reports for e-commerce

Google Analytics report setup begins after e-commerce events are correctly transmitted. The report won't correct an invalid purchase or an empty items array, so the source data is first verified, and then views for analysis are built based on it.

GA4 includes standard reports on monetization and ecommerce, while Explorations is used for custom tasks. It's best to choose a report set based on your business needs to avoid creating dozens of tables without a clear purpose.

For regular work, it's helpful to define a few key metrics in advance: sales, products, channels, devices, and funnel. This is usually sufficient for basic store monitoring.

What standard GA4 reports should I use?

Standard reports provide quick access to business metrics without building a separate analytical framework. Once Google Analytics eCommerce is properly configured, they are populated with data from recommended events and parameters.

Each report answers its own set of questions. Some show overall monetization dynamics, while others reveal specific products or the user journey between purchase stages.

The set of available reports and the layout of sections may change along with the GA4 interface, so when working, focus on the purpose of the report and the dimensions it contains.

Monetization overview

The Monetization overview is suitable for general monitoring of commercial metrics. It allows you to see revenue trends and related monetization metrics, and then drill down to more detailed reports.

This overview is useful for initial diagnostics. If overall ecommerce revenue has changed significantly, the next step is to analyze products, channels, or individual stages of the funnel.

A single overview page is usually insufficient for a management report. It's used as an entry point, after which the reasons for changes are verified using more detailed data.

Ecommerce purchases

Ecommerce purchases reveals results at the product level. The report analyzes item views, adds to cart, purchases, items purchased, and item revenue based on available metrics.

This report helps you identify items with high interest and low purchase volume. It also helps you identify products that generate a significant share of revenue with relatively few views.

Comparing products solely by revenue can hide funnel issues. Therefore, it's best to consider the data alongside views and pre-purchase actions.

Purchase journey

The purchase journey shows the sequence of key purchase stages. A typical diagram is built around session_start, view_item, add_to_cart, begin_checkout, and purchase.

The report helps quickly see how many users complete each stage and where the greatest loss occurs. For in-depth analysis, segments or user funnel exploration are then used.

The standard sequence works well for many online stores, but complex user journeys sometimes require an additional, custom funnel.

Checkout journey

The checkout journey focuses on the checkout process. The funnel may include begin_checkout, add_shipping_info, add_payment_info, and purchase.

This report is useful for diagnosing issues with the order form, delivery, and payment. If users reach checkout but abandon the order en masse after selecting delivery, the direction for further investigation becomes clearer.

For accurate reporting, all relevant events must be transmitted sequentially and reflect the actual user action.

When are custom reports needed?

Custom reports are needed when a standard view is insufficient for a specific business need. For example, a store might want to compare funnels across different categories, differentiate new customers from repeat customers, or analyze checkouts only for a specific source.

Explorations offer more freedom in choosing steps, segments, and dimensions. However, a complex report doesn't compensate for source data errors, so it's built after technical verification of ecommerce tracking.

If a report is used regularly, it's a good idea to document the filtering and segmentation rules. This helps the team interpret the figures consistently over several months.

Funnel exploration

Funnel exploration is used for a custom sequence of steps. You can define funnel stages, choose an open or closed scenario, add segments, and view transitions between actions.

For example, a Google Analytics sales funnel setup might include category views, product details, shopping cart, checkout start, and purchase. For another project, the start of the funnel might be linked to an advertising landing page.

The main requirement is related to the logic of the steps. The funnel must correspond to the user's actual journey and answer a specific question.

User segments

Segmentation helps break down the overall picture into groups with different behaviors. Often compared are mobile and desktop, new and returning users, organic and paid traffic, different regions, or product categories.

This allows you to see a problem that's lost in the average metric. For example, the overall conversion rate remains stable, but the mobile version gradually deteriorates, while the desktop version compensates for the decline.

Segments are selected based on the task. The more conditions used simultaneously, the more difficult it is to interpret the result and verify its stability.

Additional options

Standard ecommerce parameters are sufficient for most basic reports, but businesses sometimes need additional features. This could include an internal product type, loyalty program, delivery format, or another characteristic not included in the standard set.

Such data is passed as additional parameters and, if necessary, registered as custom dimensions or metrics. Before adding, you should check whether the parameter will actually be used in the analysis.

Too many custom dimensions complicate maintenance and increase the risk of different interpretations of data within the team.

How to set up a sales funnel in Google Analytics 4?

A sales funnel shows how users progress through successive stages and what share of the audience remains after each step. For e-commerce, the basic chain is linked to the product, cart, checkout, and purchase.

When analyzing, event consistency is important. If view_item is sent multiple times per load or purchases are duplicated, the interstage coefficients become meaningless.

It's best to use the funnel in conjunction with segments. The overall metric provides a benchmark, while segmenting by device, source, or product helps pinpoint the problem.

Basic ecommerce funnel

A typical ecommerce funnel begins with product viewing. The user then adds the item to the cart, proceeds to checkout, and completes the purchase. In GA4, these stages are conveniently associated with view_item, add_to_cart, begin_checkout, and purchase.

Intermediate events help us understand where interest is waning. If many users are viewing the card but the cart remains nearly empty, check the offer and product page. If the problem starts after begin_checkout, examine the order form.

The funnel is assessed dynamically. A single day with low traffic levels rarely provides a sufficient basis for drawing serious conclusions.

A hypothetical example of dynamics:

This funnel chart is used only as an example structure. Actual store metrics should be calculated based on the store's data and the same methodology.

Order funnel

Within the checkout process, the user can go through several steps: beginning the checkout process, specifying delivery, choosing payment, and confirming the order. The begin_checkout, add_shipping_info, add_payment_info, and purchase parameters are used for analysis.

The more complex the form, the more useful it is to see the transitions between individual steps. If the store uses a single-page checkout, events are tied to successfully completed user actions within the form.

During diagnostics, they check for technical errors, form usability, required fields, and delivery and payment methods. Analytics reveals the location of the loss, and the cause is then confirmed with additional data.

What does funnel analysis show?

Key funnel metrics relate to the number of users at each step, the conversion rate between stages, and the abandonment rate. These values indicate how successfully the audience completes the purchase journey.

It's useful to compare identical stages across periods. If, after changing the checkout, the share of conversions from payment to purchase has increased, the results can be further verified using comparable traffic.

Retention rates and repeat purchases are part of a broader analysis of customer behavior. Separate reports and segments are used for these, as a single transaction funnel does not capture the customer's subsequent return.

How to segment a funnel?

The overall funnel conceals differences between user groups. Therefore, after testing the basic scenario, it is broken down by device, channel, product, and audience type.

Segmentation helps determine whether the problem is localized. If the decline is noticeable across all groups, the cause may lie in a common website element. If the performance has deteriorated only for a specific channel, the quality and relevance of that traffic is checked.

Segments that are too small yield insufficient data. For evaluation, it's better to select groups with a sufficient number of users.

By devices

Comparing mobile, desktop, and tablet helps identify technical or UX issues specific to a particular device type. Checkout analysis is especially useful, as long forms and payment elements may work differently on mobile devices.

If mobile traffic generates a lot of view_item, but the begin_checkout stage is noticeably slower, it's worth checking the product page, the shopping cart, and the checkout process itself. Further analysis should be supplemented with technical tests.

You should compare the same events and comparable periods. Otherwise, differences in traffic structure may distort the conclusion.

By traffic sources

Splitting by source/medium helps compare the commercial behavior of audiences across organic search, paid search, social, email, referral, and other channels. One channel may yield a high cart rate but a low purchase completion rate.

The reason is sometimes related to user expectations after an ad message. If the landing page or offer doesn't match the ad, churn may begin before checkout.

It is advisable to evaluate channels together with revenue, number of transactions, and acquisition cost when cost data is available and verified.

By category and product

A funnel for individual categories helps you see where high interest isn't converting into sales. A product might receive a lot of view_item, but rarely makes it to add_to_cart due to price, terms, lack of a desired option, or product page features.

Comparisons within a single category are often more informative than a store's overall rating. Products may have different prices, seasonality, and purchasing decisions.

For such analysis, stable item_id, item_category and related product parameters are especially important at all stages.

By user type

New and repeat customers typically experience different paths to purchase. Returning customers are already familiar with the store, brand, and terms, so their behavior can't always be directly compared to their first visit.

Segmentation helps to separately evaluate first-time and repeat purchases. Customer lifetime value and retention are also used for long-term analysis, provided the project's data structure allows for reliable metrics.

Audience segmentation becomes more useful when it is applied regularly and according to the same rules.

What e-commerce metrics should you track?

The number of metrics depends on the store's objectives, but it's best to keep the basic set small and clear. A regular report typically includes sufficient metrics for sales, products, funnel, and marketing sources.

Each metric should answer a specific question. If a team regularly collects dozens of metrics but doesn't use them in decision-making, the report gradually turns into a repository of numbers.

Before setting up a dashboard, it's helpful to identify key business questions and tie metrics to them.

Sales and revenue

Key commercial metrics include revenue, transactions, purchases, average order value, and number of items sold. These are compared across time periods, channels, devices, and other key segments.

The average order value helps explain revenue fluctuations when the number of transactions remains stable. The number of units sold complements the analysis when customers purchase multiple items in a single order.

If there are discrepancies between GA4 and CMS, first check the methodology, time zone, currency, returns, and technical implementation of events.

Product indicators

Product analytics includes item views, add-to-cart, checkout, items purchased, and item revenue. These metrics help compare product interest with actual sales results.

High viewership performance doesn't necessarily translate to high sales. Therefore, it's best to evaluate your rankings through several sequential actions and the resulting revenue.

Similar logic can be used for categories if item_category is transmitted equally at all events.

Conversion funnel

Conversion rate is calculated for the entire purchase and for individual clicks. It's useful to look at the product preview to cart, cart to checkout, and checkout process separately.

This separation quickly reveals the area where the change occurred. Overall conversion may decrease by a few percent, but the cause is often concentrated in just one click.

With a small amount of data, it's better to draw conclusions over a longer period so that random purchases don't change the picture too much.

Marketing effectiveness

For marketing, analyze traffic source, source/medium, campaign, purchases, revenue, and conversion rate. If expenses are involved, ROAS and other advertising performance metrics can be calculated.

GA4 also evaluates channel contributions based on available attribution. When compared with advertising accounts, discrepancies may occur, as different systems use their own accounting rules.

Full CAC requires costs that are often outside of Google Analytics. If these costs aren't reported, it's impossible to calculate the metric using GA4 data alone.

Customer behavior

Customer behavior includes new and repeat visits, repeat purchases, audience segmentation, customer lifetime value, and retention. These metrics help you see beyond a single transaction.

For an online store, repeat purchases can be a significant part of revenue, so a separate analysis of returning customers provides a more complete picture. The period and segment criteria should be consistent across all reports.

It is useful to combine data with product and marketing analytics when such a connection is truly needed for the task.

What mistakes are encountered when setting up e-commerce?

Most issues are related to incorrect event timing, missing parameters, or duplicates. An event may still appear in the GA4 interface, even though its data is no longer sufficient for a proper ecommerce report.

Therefore, the test is built around a real user scenario. The specialist walks the user through the entire process, from catalog to purchase, verifying events and values at each stage.

After corrections, the script is tested again, including non-standard order options.

Events are not sent at all

If the event is missing, first check the site action itself and the dataLayer status. Then check the GTM trigger, variables, and the GA4 send tag.

This could be due to a frontend change, an incorrect selector, AJAX logic, a JavaScript error, or a conflict between multiple integrations. On SPAs, the component lifecycle is additionally checked.

It's best to run diagnostics sequentially from the data source to GA4. This will quickly identify at which level the event is being lost.

The event is sent without required parameters.

An event may successfully appear in DebugView but still return an empty item, a missing currency, or an invalid product value. For e-commerce, this result cannot be considered a valid setup.

Parameters are validated based on the user's actual action. The product price, quantity, and identifier must match the cart or order status at that moment.

After testing, it is useful to check several different products, since the error may only affect a certain card type or variation.

The purchase event is duplicated

Duplicate purchases lead to inflated revenue and transaction counts. A common cause is a repeated confirmation page load, two tags running simultaneously, or an integration error.

A unique transaction_id must be passed for each transaction. Additionally, they check whether the same tag is fired multiple times within GTM or the site's code.

After the correction, carry out several test orders and reopen the confirmation page to ensure the logic is stable.

The order cost is being transmitted incorrectly

The value parameter must comply with the accepted cost accounting rule. The project pre-defines whether shipping, tax, and other components are included in the value passed.

Errors occur with discounts, promo codes, multiple currencies, and changes in product quantities. Therefore, testing a single simple order doesn't cover all scenarios.

When checking, the GA4 value is compared with the order in the CMS or other system that stores the primary transaction data.

Items array errors

The items array contains data about the products within the event. If the item_id, name, price, or quantity are passed incorrectly, product reports will be incomplete or contain duplicate items.

Particular attention is paid to identifiers. A single product shouldn't suddenly have different item_ids on the product card, in the cart, and after purchase.

After launching, it's helpful to test a few popular products and variations to ensure consistency in structure.

Payment service breaks attribution

Some payment flow scenarios redirect users to an external domain, after which they return to the store's website. Without proper configuration, this redirect can impact the session source and purchase attribution.

For such projects, cross-domain tracking and referral source behavior are checked. The exact approach depends on the payment service and the user's journey.

The test should be conducted using a real click-through sequence. Simply submitting a purchase on the thank-you page doesn't reveal a potential attribution issue.

Analytics diverges from CMS

GA4 and the store's administrative system may use different accounting rules. These may include differences in order entry timing, returns, currencies, test transactions, and cancellation processing.

Minor discrepancies should be assessed within the context of the methodology. Significant discrepancies require review of events, transaction_id, cost, and report filters.

It is useful to take a specific period and a few real orders, then trace their path through both systems.

Consent Mode and ad blockers affect data collection

Consent Mode, cookie restrictions, and blockers can affect the amount of available data. Some users limit or disable certain types of storage and tracking.

The setup must take into account the actual consent system on the website. Ecommerce tracking quality cannot be assessed solely by the overall order match with the CMS without taking into account browser limitations and consent.

When diagnosing, technical errors are distinguished from data losses related to privacy rules.

What does a business get after setting up E-commerce Analytics?

Once configured, the store has a connected system of data on products, purchases, and customer journey stages. The team can see what actions occur before a transaction and where audience behavior changes.

This data can be used for regular analysis of marketing, product range, and checkout. Decisions are tested against the same events and metrics, making it easier to correlate website changes with results.

Below are the main directions that become available after correct configuration.

A clear sales funnel

The funnel shows the number of users at each stage and the transition between actions. You can see the share of the audience that, after viewing the product, adds it to the cart, initiates checkout, and completes the purchase.

If a change occurs in a single step, the team receives a specific direction to investigate. For example, a decrease in the transition to payment requires a checkout analysis, while a drop in add_to_cart is associated with an earlier step.

The funnel can be segmented by devices, channels, and other characteristics if the data volume is sufficient.

Product data

GA4 collects information on views, cart additions, purchases, and item revenue. This helps compare products based on audience interest and commercial performance.

Positions with high views and low conversion rates can be analyzed separately. Products with stable sales can also be easily compared by traffic source and category.

The accuracy of such analysis directly depends on stable identifiers and a correct items array.

Evaluation of marketing channels

The team gains the ability to compare channels based on purchases and revenue. Traffic is evaluated alongside commercial results, so sources with the same number of sessions may appear completely different.

For paid advertising, the data is supplemented with cost data if the integration is configured and the figures are verified. Then, ROAS and related metrics can be analyzed.

When comparing advertising accounts with GA4, differences in attribution and system methodology are taken into account.

The basis for CRO

CRO hypotheses are best built around a specific problem in the data. If users frequently open the cart but rarely proceed to checkout, testing should focus on this area.

After the change, the same set of events is used to evaluate the outcome. With sufficient data, periods, segments, or the results of a controlled experiment can be compared.

Analytics alone cannot explain every reason for behavior, so quantitative data is supplemented with UX research when necessary.

Data for management decisions

Ecommerce data helps discuss sales at the level of specific products, channels, and funnel stages. Managers see not just one overall revenue metric, but the factors that influence its change.

For regular reporting, you can create a dashboard in Looker Studio or use BigQuery export if your project requires deeper data management.

The set of indicators is selected based on the company's objectives so that the report remains understandable and suitable for regular monitoring.

Which websites require e-commerce setup?

Google Analytics eCommerce is most often used by online stores, but eCommerce events are also suitable for other projects with a clear online transaction. The main criteria are product availability, price, and the sequence of actions leading up to purchase.

The eCommerce Analytics setting can be applied to services with online payment, subscriptions, and other models where the user completes a commercial action on the site.

Before implementation, it is necessary to check how well the standard e-commerce model corresponds to the actual project process.

Online stores

For an online store, ecommerce tracking covers the catalog, product cards, cart, checkout, purchase, and refund. For a large product matrix, detailing by item_id, category, and revenue is especially useful.

Setting up e-commerce in Google Analytics helps link product behavior to customer sources. A store can analyze which channels drive sales for specific categories and where customers most often abandon checkout.

To achieve a high-quality result, a unified data structure is required at all stages.

Websites with online payment for services

If a user selects a service and pays for it directly on the website, the ecommerce model can also be applied. The service can be transferred as an item with a clear identifier, name, and price.

This model is suitable for projects where there is a full transaction and the purchase is recorded. If the site only collects orders without payment, standard e-commerce may not correspond to the actual process.

In this case, it is better to build an analytical model around lead events and qualification stages.

Subscription services

For a service with pricing plans and online payment, you can record product selection, checkout initiation, and the transaction. The structure depends on how the initial payment, renewal, and subsequent charges are structured.

Repeat payments require special attention, as not every scenario occurs in the user's browser. Server processes and payment systems may store some data outside of standard client tracking.

Before setting it up, you need to determine what operations GA4 can actually reliably capture.

Projects with a complex purchase funnel

If a user goes through several significant steps before payment, standard ecommerce events can be supplemented with other events. The main rule is a clear flow and the absence of unnecessary actions without analytical value.

Such projects especially benefit from funnel exploration and segmentation. You can see the transitions between stages and compare scenarios for different user groups.

Before implementation, it is advisable to describe a complex funnel in a tracking plan so that developers and analysts have a common understanding of each step.

Marketing effectiveness analysis

Marketing channels are conveniently compared based on commercial metrics rather than the number of visits. In GA4, you can link source/medium, campaign, and available acquisition parameters to purchases, transactions, and revenue. This allows you to see which channel is driving customers and the volume of sales generated by those clicks.

One source may generate high traffic and low conversion rates, while another may generate fewer users but result in higher orders per thousand visits. When analyzing paid advertising, advertising costs are also taken into account if they are correctly imported into the analytics system.

ROAS is calculated only with reliable revenue and cost data. However, profit, margin, and total CAC require additional data that the standard GA4 installation may not include.

Finding problems in the funnel

A funnel shows the number of users at successive stages of the purchase process. For an online store, the basic chain typically begins with viewing a product, then adding it to the cart, proceeding to checkout, and completing the order.

Comparing stages helps identify the area with the highest abandonment rate. If the issue recurs over several periods and is noticeable in a specific segment, it can be analyzed separately. For example, churn on mobile devices alone requires a different analysis than a general decline in conversion across all groups.

After making changes, use the same funnel for retesting. It's best to compare the same periods and comparable traffic sources to ensure seasonality or advertising activity doesn't distort the results.