A simplified mobile screen does not complete a purchase on its own. Somebody still needs to understand the total, enter information, correct a mistake and recognise whether payment occurred. An interface creates commercial cost when one of those tasks forces departure or repetition. Useful improvement preserves the decision while removing avoidable work from the entire journey.
A device difference locates a question rather than supplies its cause
Salesforce Research data published through Statista for the third quarter of 2025 records differences between baskets not resulting in orders on computers and mobile devices. In footwear the percentages were 70% and 82%; in food and beverages, 65% and 77%. These are activities at the retailers observed by the provider internationally, rather than outcomes at a particular Spanish store.
Three categories: baskets without orders by device
Share of baskets not resulting in a completed order. Salesforce Research, third quarter of 2025, international data.
Tap or hover for details. Select a series above to show or hide it.
Collapse notes
Exact Statista table, source Salesforce Shopping Index. Operational data from observed commerce sites; category-level basket/site counts and aggregation details are not supplied. This is not a retail census or a Spanish rate. Observed differences can include intent, offer and buyer composition; they do not identify a causal device effect.
Explore the data table
| Category | Computer | Mobile |
|---|---|---|
| Footwear | 70.0% | 82.0% |
| Food and beverage | 65.0% | 77.0% |
| Electronics and accessories | 71.0% | 75.0% |
Mobile can support research before an order is completed elsewhere. The offer, availability and arriving customers can also differ. Treating the entire gap as a screen defect turns incomplete diagnosis into redesign spending. Review needs to identify which action could not be completed by somebody whose circumstances permitted continuation.
The problem can be a commercial condition as well as the interface
The US Baymard survey reported by Statista for 2025 helps separate interruption reasons. Among checkout reasons after excluding mere browsing, extra costs appear in 39% and process length or complexity in 18%. A website error and inability to calculate the total in advance are different problems. These percentages do not describe mobile purchases alone.
United States: checkout reasons combine price, effort and failures
Five declared reasons in the Baymard 2025 series, excluding mere browsing or lack of purchase intent.
Tap or hover for details. All values are also available in the data table.
Collapse notes
Baymard Institute / Statista. The statistic lists 1,026 adults without separately identifying the exact filtered reason-question base. Multiple responses must not be added. Exact fieldwork dates are unpublished. These are US data, not mobile-only activity or recoverable sales. The current primary page covers another edition; historical values come from the Statista 2025 table.
Explore the data table
| Category | Reports the reason |
|---|---|
| High extra costs | 39.0% |
| Long or complicated checkout | 18.0% |
| Website errors or crash | 15.0% |
| Total not visible in advance | 14.0% |
| Card declined | 8.0% |
An unexpected total calls for price, delivery and conditions before commitment. A field requiring somebody to consult a document calls for advance notice and a route back. Enlarging a button does not necessarily solve either difficulty. The mechanism determines the work: clarifying a condition, preserving an entry and communicating action state are different interventions.
The task needs to survive the keyboard, enlarged text and delay
The Web Vitals documentation separates loading, interaction and visual stability. A page can appear quickly yet respond late to an action; it can also move a control during selection. This helps locate technical difficulty rather than promises increased conversion for reaching a metric.
Priority information depends on the decision. A real photograph can explain the product; a heavy resource delaying price and conditions can interrupt that explanation. When the keyboard appears, fixed headers and buttons should not cover the field or instruction. After selection, visible state needs to distinguish waiting from error to avoid unnecessary repetition.
The W3C reflow criterion addresses preserving information and function when content reorganises, with exceptions for material requiring two dimensions. The commercial objective is that enlarging text retains conditions required to buy and the possibility of correcting a choice.
The minimum pointer-target size criterion addresses dimensions and spacing of controls, also with exceptions. Its practical reading includes links, quantities and neighbouring options as well as the main action. A purchase requires correctly selecting several elements rather than simply hitting the last button.
Resuming does not keep an offer available indefinitely
Switching applications to find an address or taking a call can separate the beginning from completion. Returning with valid information retained avoids repetition. If availability or price changed, the journey should disclose that before confirmation. Continuity preserves the return point and current conditions rather than promises that a selection remains available without limit.
A localised error needs to identify the information requiring correction without clearing everything else. Interrupted communication after submission presents a different question: whether the operation occurred. Inviting another payment without establishing that outcome can duplicate its consequence. Confirmation and assistance should follow the result the retailer can establish rather than place that uncertainty on the buyer.
Useful measurement retains starts, completions and pending work
The commercial unit is the operation rather than each visit. Comparing valid completed purchases with eligible operations started within a defined cohort and window follows the work. When somebody returns on another device, reporting retains reliable links and leaves other connections unknown. Counting only the final device can erase part of the journey; automatically assigning everything to the first exaggerates its role.
Lower average duration does not establish ease either: it can fall because more people left early. Valid completions, errors and assistance needs make time interpretable. The next investment should respond to an observed difficulty and retain the conditions that changed instead of assigning every subsequent variation to the interface.
The mobile journey ends when the purchase can be recognised and the buyer retains what is needed to receive it. Hiding limits to increase a click moves costs to service, returns or later explanation. Complete improvement removes effort while retaining information: the offer is understood, the action responds and a mistake permits continuation from what remains valid.
In this article
Marian Vasile
Marian Vasile is a MarTech and growth professional. His work spans product marketing, SEO, paid advertising and deep online tracking, including behavioral analytics and attribution.