Some of retail’s most overlooked customer experience failures are not failures at all. They are features working exactly as designed: The page loads, the button responds, the notification sends, and the system changes the order status.

Technically, everything works. And yet, for the customer, something feels wrong.

Inside the retail organisation, the ‘feature’ may be considered a success. It launched correctly, passed testing and performs exactly as specified. The project is complete. The box is ticked. There may even be a small round of applause.

But our 2026 Online CX Report shows that customers do not experience websites through technical specifications. They arrive with years of experience in digital behaviour. They instinctively look in certain places, swipe in certain directions and attach familiar meanings to familiar words.

They do not know how the retailer’s systems are configured. They do not know what an internal order status means. They were not in the project meetings. And they should not need to be.

This is e-commerce’s functionality gap: The distance between something technically working and it working in the way customers reasonably expect.

The feature works. The shopper hesitates

Take Zara’s product imagery, for example. The images load perfectly; the functionality behaves exactly as intended. But shoppers move through the images vertically, not horizontally. 

That sounds like a small distinction until you consider how deeply horizontal swiping has become embedded in everyday digital behaviour. Our phones reinforced it. Social platforms reinforced it. Tinder practically turned it into a reflex.

Consumers swipe from right to left repeatedly, often without consciously thinking about it. Then they arrive at Zara, and that familiar action no longer produces the expected result.

Nothing is broken. There is no technical error. But the shopper has to stop, reconsider and learn a different interaction. Zara has not created a functional problem, but a moment of cognitive friction.

That does not necessarily make Zara’s design wrong. Breaking conventions can be intentional, distinctive and occasionally brilliant. But being different does not automatically translate into innovation. Sometimes, it simply makes the customer think harder.

Sometimes functional is almost invisible

Live chat provides another surprisingly simple example. In Humii’s research, 91 per cent of shoppers expect to find live chat in the bottom-right corner of a website. Nobody formally taught them to look there; they developed the behaviour through repetition. Over time, shoppers have learnt that the bottom-right corner is where immediate support usually lives.

Yet some retailers position their live-chat button in the bottom-left corner. The button works. The chat opens. The technology is fully operational.

Coles and Woolworths provide a useful comparison. Coles positions its chat icon on the bottom left. Woolworths places its support entry point where shoppers are more accustomed to looking. In Humii’s comparison, shoppers found the Woolworths option considerably easier to locate.

The issue was not whether Coles offered chat; it was whether shoppers saw it without having to look for it.

The feature exists, but it sits outside the customer’s learned pattern. This is the kind of friction traditional testing can miss. Testing confirms that the button appears, opens and connects correctly, but shoppers’ experience asks a different question: Would I instinctively know where to find it?

Those are two very different measures of success.

‘Dispatched’ does not always feel dispatched

The language used after purchase exposes an even more interesting gap: A customer receives an email announcing: “Your order has been dispatched.”

They click the tracking link and find that the carrier is waiting for the parcel, the shipping information has only just been received, or no meaningful movement is visible. Internally, from the retailer’s perspective, ‘dispatched’ may mean that fulfilment is complete, a tracking number has been generated, or a shipping label has been created. To the retailer’s system, the order has progressed successfully.

But to the customer, dispatched means something much simpler: The parcel has left the building; it is on its way.

Take Ralph Lauren, for example: Our mystery shopper received a dispatch communication while the carrier tracking showed little or no meaningful movement. The result was confusion: If the order has been dispatched, why does the delivery company appear not to have it?

There may be perfectly valid operational reasons. The first carrier’s scan may be delayed. The parcel may already be moving. Data may take time to synchronise between systems.

But the customer does not see the operational explanation. They see a retailer saying one thing and a carrier appearing to say another. 

Every component may have worked correctly:

• The warehouse updated the order.
• The email was triggered.
• The tracking link was generated.
• The carrier page loaded.

Four functional systems. One confused customer. The problem is not necessarily the technology, but an internal milestone has been presented as a customer outcome. Retailers measure dispatch from the moment their system changes status. Shoppers measure it from the moment their parcel appears to move.

That gap matters.

A refund is not complete when the system says it is

The same disconnect can appear in refund communication. A customer receives an automated message:

“Your refund has been processed.” But when they check their bank account, nothing shows.

To the retailer, the refund workflow may be complete: The instruction has been issued, the internal status has changed and the automated email has been sent. But to the customer, a refund is complete when the money is back in their account.

Again, ‘processed’ may be technically accurate. But it lacks the context the customer actually needs:

• When were the funds released?
• How long should the bank take?
• When should the customer become concerned?
• Who should they contact if the money does not arrive?

Of course, retailers do not control every bank or financial institution, and refund settlement times can vary.

But they do control the expectation they create.

Humii’s data shows that almost 30 per cent of refunds take more than three days to reach shoppers after the refund notification, while approximately 90 per cent of retailers tell customers to expect the money within one to three business days.

That creates a predictable moment of concern: Day three arrives, the money does not.

Has the refund failed? Should the customer contact the retailer? Is the bank responsible? Does the stated timeframe begin when the return is lodged, approved or processed?

The system knows. The customer often does not. ‘Processed’ is an internal milestone presented as a completed customer outcome. Those are not always the same thing.

The customer experiences the gaps between the systems

None of these examples represents a traditional technical failure: Zara’s imagery works. Coles’ live chat works. The dispatch email sends. The tracking link opens. The refund notification arrives. If you test each component independently, each one may pass.

But customers do not experience components independently; they experience the gaps between them. They experience the moment when an interaction behaves differently from every similar interaction they have learnt before.

They experience the space between ‘dispatched’ and the parcel actually moving.

They experience the days between ‘refund processed’ and the money returning to their account.

They experience the difference between information existing and information being useful.

This is why internal testing alone will never reveal the entire customer journey. People inside a retail organisation have context customers do not. They know where information lives, what status labels mean and why a feature behaves in a particular way. Real shoppers arrive without that knowledge.

Retailers need to ask a better question

‘Does it work?’ remains an important question. However, it is simply no longer enough.

Retailers should also be asking:

• Does it behave the way shoppers reasonably expect?
• Is the information available when it is needed?
• Does our language mean the same thing to customers as it does internally?
• Are we measuring the completion of a system process or the completion of the customer’s experience?
• Would someone outside our organisation understand this without explanation?

A website can be fully operational and still create unnecessary effort. A notification can be accurate and still be misleading. A feature can be available and still be difficult to find. A process can be complete internally and remain unresolved for the customer.

The next time someone says, “But it works,” ask the more important question: For whom?

Working as designed is not the same as working for the customer.

• Learn more about the Online CX Index powered by Humii and Inside Retail here.

• About the author: Mareile Osthus is the founder and CEO of Humii, and a former CCO of The Iconic.