Ecommerce conversion leaks: where established stores commonly lose customers

When an established online store underperforms, the first prescription is often a new button colour, shorter checkout or larger promotional banner.

Any of those changes might help. None should be the starting assumption.

An ecommerce conversion rate is the output of a connected system. Traffic arrives with a level of intent and an expectation set by the advertisement, search result, email or social post. The customer then has to find a suitable product, understand it, trust the offer, accept the delivery promise, complete payment and receive what was ordered. Inventory, pricing, fulfilment and customer service sit behind the interface.

A leak at any point can reduce sales. Several small leaks can compound. A technically perfect checkout cannot rescue irrelevant traffic or an unavailable product. A persuasive product page cannot compensate for unreliable stock data. A fast website cannot make an unclear returns policy feel safe.

The practical question is therefore not, “Which ecommerce best practice are we missing?” It is:

Where is this store losing commercially valuable customers, what evidence supports that diagnosis and which change is credible enough to test?

The short answer

Review the journey in seven connected stages:

Acquisition: Are the right customers arriving with an accurate expectation?
Product discovery: Can they find and compare suitable products?
Product decision: Can they resolve price, variant, fit, trust and delivery questions?
Cart: Are order rules and total costs clear before commitment?
Checkout: Can they complete details and payment without preventable effort or error?
Fulfilment: Does the operational promise match what the interface sold?
Retention: Does the post-purchase experience support service, repeat purchase and trust?

Performance, accessibility and measurement affect every stage. Diagnose them as horizontal capabilities rather than one isolated page-speed task.

Start by defining the conversion that matters

“Increase conversion” sounds precise but can hide several decisions.

Is the business trying to increase first purchases, repeat orders, subscription starts, wholesale account applications or revenue from a priority range? Is a sale still valuable after discount, freight, payment fees, returns and service cost? Are new and returning customers expected to behave differently?

Choose a primary commercial outcome and the guardrails that must not deteriorate. These may include revenue, margin, average order value, cancellation, return rate, support contact, delivery performance or customer acquisition cost. The correct set depends on the business.

Then validate the denominator. Conversion rates can differ materially by device, channel, geography, customer type, product category and season. An aggregate rate can improve because the traffic mix changed while the experience itself stayed the same.

Before prescribing design work, confirm that product views, cart additions, checkout steps, purchases, values, currency and refunds are being recorded accurately. Google’s current GA4 ecommerce guidance defines events that can measure product-list exposure, product selection, cart activity, checkout and purchase. The existence of a GA4 property does not mean those events are complete or dependable.

Leak 1: the acquisition promise does not match the store

The website may be blamed for traffic that was unlikely to convert.

An advertisement might promote a low price while the landing category contains premium products. A broad search query may indicate research rather than purchase intent. An influencer audience may love the content but sit outside the delivery market. A promotion may attract discount-led customers to products with limited stock.

Review:

Channel, campaign, search term and creative promise
Landing page and product availability
New versus returning customer behaviour
Geography and device
Promotion, price and margin
Downstream customer quality, not only sessions

This does not excuse a weak website. It prevents a user-experience team from trying to solve audience selection with interface decoration.

Leak 2: customers cannot find the right product

Product discovery includes navigation, categories, on-site search, filters, sorting, merchandising and the language used to describe the range.

Common symptoms include:

High use of search followed by weak result engagement
Frequent zero-result or irrelevant-result queries
Filters that reflect internal data rather than customer decisions
Important attributes missing from product cards
Categories that mix incompatible products
Mobile controls that obscure active filters or make comparison difficult
Out-of-stock items dominating the path without a useful alternative

Findability is partly an information problem. If product attributes are inconsistent or incomplete, a better filter interface cannot create reliable results. Taxonomy, product-information management and merchandising ownership may need attention before design.

Baymard’s current product-list and filtering research and ecommerce search research document recurring usability failures in how stores help customers narrow and retrieve products. Their benchmarks are useful context, not a substitute for observing the store’s own customers and catalogue.

Leak 3: the product page leaves important uncertainty unresolved

A product page is a decision environment. Its job is not only to display imagery and an add-to-cart button.

The customer may need to confirm:

What is included and what is optional
Which variant, size, specification or compatibility is right
Current price, promotion and stock status
Delivery location, timing and cost
Returns, warranty and support conditions
Product origin, material, ingredients or care
Evidence such as accurate reviews, demonstrations or technical documentation

The correct content depends on the product and risk. A fashion item needs fit and fabric guidance. A technical component may require dimensions, compatibility and downloadable data. A subscription needs billing frequency, renewal and cancellation explained before purchase.

Avoid manufacturing urgency or hiding material conditions to force action. Australian businesses must give truthful information and cannot use store policies to override consumer guarantees. The ACCC’s online-shopping guidance and its 2025 return-policy compliance sweep are useful checks for claims, returns language and customer rights.

Leak 4: the cart introduces surprise or conflict

The cart is often where the operational model becomes visible. Minimum quantities, mixed-cart restrictions, account pricing, promotional eligibility, shipping thresholds and stock reservations can all appear here.

Review whether customers can understand:

What they are buying and in which quantity or variant
Whether a rule prevents checkout and how to resolve it
Indicative delivery cost or the basis for calculating it
Promotion conditions and exclusions
Stock changes or backorders
How to edit, save or remove items

The right answer is not always to remove a business rule. It may be to explain it earlier, make progress visible or design a lower-friction way to satisfy it.

Brown Brothers is a useful Emote example. Its multi-brand store required minimum bottle quantities with exceptions for selected products. The interface used a live countdown to help customers complete the order rule, while inventory and product information were managed across the connected brand sites. Conversion design and operational logic were treated together.

Leak 5: checkout creates preventable effort or failure

Checkout friction can include unnecessary account creation, repeated fields, unclear validation, inaccessible controls, unexpected costs, unavailable payment methods, address problems or technical errors.

Work through the checkout on representative devices and conditions:

Guest and returning customer
Common and unusual addresses
Different delivery and payment options
Promotion accepted, rejected and removed
Validation errors and recovery
Out-of-stock or price-change behaviour
Keyboard-only and representative screen-reader journeys against the agreed accessibility requirements
Slow or interrupted connections

Error handling deserves particular attention. A message should identify the field or problem, explain how to correct it and preserve valid information. W3C’s accessible form-notification guidance provides practical patterns for overall feedback, inline errors and successful submission.

Do not optimise checkout by deleting information customers need for an informed purchase. Reduce unnecessary effort, not clarity.

Leak 6: mobile performance or reliability interrupts intent

Customers cannot convert through a page that does not load, respond or remain stable enough to use.

Assess real-user experience, not only a laboratory score. Google’s Web Vitals guidance defines user-centred measures for loading, responsiveness and visual stability. Segment field data by page type and device where possible, because a fast homepage can hide a slow product template or third-party checkout script.

Performance work may involve image delivery, JavaScript, third-party applications, fonts, caching, theme code, hosting requirements or platform limitations. Emote can advise on hosting requirements and coordinate with a suitable hosting partner or the client’s appropriate provider; it does not sell or directly provide hosting infrastructure.

Avoid promising that a specific speed improvement will create a universal conversion increase. Performance removes friction and supports use, but the commercial effect depends on the store, audience and other changes.

Leak 7: the operational promise fails after payment

Conversion optimisation should not end at the purchase event.

An order can create future loss through inaccurate inventory, slow dispatch, unclear tracking, damaged delivery, difficult returns or unhelpful service. These failures can increase cancellation, support cost and negative word of mouth while reducing repeat purchase.

Business.gov.au describes shipping as an essential ecommerce operation and recommends considering reliable, affordable delivery and customer protection. Review the full promise:

Inventory accuracy and source of truth
Order transmission and exception handling
Dispatch and tracking communication
Delivery estimates and service coverage
Returns, refunds and exchanges
Customer-service ownership
Replenishment, subscription or reordering journeys

Some “website” problems are actually data or process problems surfaced by the website. The optimisation backlog should name the owner and dependency rather than disguise every issue as frontend work.

Build the diagnosis from more than analytics

Quantitative data can show where behaviour changes. It rarely explains the reason by itself.

Use a combination of:

Behavioural evidence

GA4 ecommerce events and purchase-journey reports
Funnel and path explorations
Product, category, device and channel segmentation
Error events and performance data
Experiment results

Google says Funnel exploration can visualise the steps users take to complete a task and show where they succeed or fail. The funnel should reflect the store’s actual journey rather than forcing every customer into one ideal sequence.

Customer evidence

On-site search terms and zero-result searches
Usability observation and customer interviews
Surveys and open-text feedback
Reviews, returns reasons and service contacts
Questions asked before purchase

Operational evidence

Stock and fulfilment exceptions
Payment failures and fraud controls
Dispatch and delivery performance
Refunds, cancellations and returns
Margin and promotion rules

When all three point to the same friction, confidence rises. When they conflict, investigate before changing the experience.

Prioritise the leaks worth fixing

A long audit can be less useful than a short, defensible backlog.

Score each candidate issue against:

Evidence confidence: Is it observed across dependable data or only an opinion?
Commercial consequence: How many valuable customers or orders could it affect?
Customer consequence: Does it prevent, mislead or materially frustrate people?
Effort and dependency: Does the change require data, platform, vendor or operational work?
Implementation risk: Could it affect revenue, accessibility, search, security or fulfilment?

Act quickly on high-confidence, high-consequence faults. Instrument and investigate high-consequence issues with weak evidence. Keep low-consequence ideas in the backlog unless they are genuinely safe and inexpensive.

Turn each priority into a controlled improvement

Write a hypothesis that connects evidence, change and expected customer behaviour.

For example:

Customers selecting compatible parts are abandoning product pages because critical dimensions are hidden in downloads. Bringing verified dimensions and compatibility guidance into the page should increase qualified cart additions without increasing returns.

Then define:

The audience and pages affected
The exact change
Primary outcome and guardrails
Technical and operational dependencies
Quality-assurance requirements
Test method, duration and decision rule where experimentation is viable
Rollback or remediation plan

Not every change can be A/B tested. Low traffic, platform constraints, legal fixes and technical defects may require other evidence. Record the decision and monitor the appropriate before-and-after signals without claiming causal certainty the design cannot support.

Continuous improvement in practice

Bastion Lane Espresso illustrates a prioritised improvement approach. Emote’s public case study describes design, SEO and technical reviews, followed by work on site speed, navigation, shop and product-list pages, campaign support and a rules-based complimentary-product feature. Ongoing maintenance continued after the initial changes.

The relevant lesson is not that every store needs the same audit or feature. It is that established ecommerce performance often improves through a connected programme of UX, technical and operational changes rather than one redesign event.

Frequently asked questions

What is a good ecommerce conversion rate?

There is no universal rate that can determine whether a store is healthy. Product type, price, device, channel, market, customer mix, season and measurement setup all influence the number. Compare like-for-like segments and commercial value before using a benchmark.

Should we fix checkout first?

Only when evidence shows checkout is the material constraint or a serious defect exists. Customers may be lost earlier through traffic mismatch, product discovery, availability or unanswered questions.

Does a faster website always increase conversion?

No outcome is guaranteed. Speed and responsiveness reduce technical friction and support use. The commercial effect depends on the audience, pages affected and other constraints.

Should we copy a competitor’s ecommerce features?

Not without evidence. A feature may serve a different catalogue, customer, operating model or platform. Understand the customer problem and dependency before copying the interface.

How often should an established store run CRO?

Continuous improvement should follow the store’s traffic, release cycle, campaign calendar and internal capacity. The discipline is ongoing measurement and prioritisation, not an arbitrary number of tests.

Can Emote review the store before we rebuild it?

Yes, where an agreed engagement is appropriate. Detailed audits, research, testing and prioritised roadmaps are paid work. An initial meeting can clarify the problem, available evidence and most credible next step.

Diagnose the journey before prescribing the interface

The most valuable ecommerce conversion work begins by locating the real constraint.

The leak may be traffic, findability, product information, order rules, checkout, performance, inventory, delivery or retention. It may cross several of them. Use behavioural, customer and operational evidence together, prioritise by consequence and confidence, and protect the commercial and customer outcomes around every change.

Explore Emote’s Conversion Rate Optimisation and Websites and eCommerce capabilities for a connected approach to the store and its ongoing performance.

If you want to identify and address the ecommerce leaks that matter most, book a meeting with Emote. Emote will first understand the commercial objective, current evidence and operating constraints, then recommend an appropriate paid optimisation pathway.