From Edge Case to Evidence
ATM Locator Usability, A Researcher-Initiated Investigation
A researcher-initiated investigation prompted by a signal uncovered during an unrelated homepage baseline study, demonstrating the value of proactive inquiry when users struggle in ways that fall outside the original research scope.
I initiated this follow-on investigation to determine whether what we'd seen was an isolated edge case or evidence of a broader usability failure. The central question was simple and falsifiable: was the homepage finding noise, or did it reflect a systematic problem with the ATM locator tool?
A focused, unmoderated study was scoped to answer the question efficiently, without the cost or scheduling overhead of a full moderated engagement. Participants interacted with the locator independently while verbalizing their thoughts. No moderator prompting; responses reflect genuine, unprimed behavior.
Task scenarios
- 1Find ATMs near the participant's current location.
- 2Search for ATMs in a different city or town.
- 3Explore the tool for any confusion, language, error messaging, information density, and filter behavior.
Strong visual foundation, significant functional gaps
After completing a successful search, 90% found the map helpful for identifying nearby ATMs, confirming the core map experience works. The primary failures are upstream: search friction and location permission handling prevent users from getting there.
The search input is the primary failure point
Half of all first-search attempts failed because participants used the browser's address auto-populate function, which the tool could not process. Searches silently failed, with no clear error message explaining what went wrong or how to recover.
Search method, first vs. second attempt
Self-rated ease of search
Scale: 1 = very difficult, 7 = very easy.
The improvement between first and second attempt (5.5 → 6.1) reflects users learning to avoid auto-populate, not the tool improving. Adaptable users succeeded; those who repeated the same behavior continued to fail.
The designed flow worked for only 1 in 10 participants
The permissions problem is two-layered: most users don't grant access by default, and when they do, the tool frequently doesn't function correctly anyway.
These aren't outlier attitudes, privacy-first browser defaults are increasingly common, particularly among users who manage multiple accounts or use shared devices. The tool must be usable without location permission as a baseline expectation, not an edge-case recovery path.
Four targeted improvements with clear rationale
-
High priority
Clear, specific error messaging for failed search inputs
When browser auto-populate causes a search to fail, the tool currently returns no meaningful feedback. Add an error state that explains what happened ("We couldn't recognize that format, try typing a zip code or city name directly") and guides users toward a successful path. This directly addresses the 50% first-search failure rate.
-
High priority
Disable or dim unavailable filters by location
If no ATM in the searched area supports a given filter (e.g., "Deposit Cash"), that option should be visually dimmed with a tooltip, rather than allowing users to apply a filter that returns empty results without explanation. This resolves the confusion surface that first appeared in the homepage study.
-
Medium priority
Explanatory content for "Deposit Cash", process and scope
40% of participants didn't understand what depositing cash at an ATM entails, including which ATMs support it, what the process looks like, and any limitations. Add helper text or a collapsible explainer; list the full set of services available at each ATM rather than relying on filter labels alone.
-
Medium priority
Clarify map logo visual language with a legend
30% of participants were confused by logos displayed on the map, uncertain whether they indicated ATMs, partner banks, or nearby businesses. Standardize ATM markers with a distinct, labeled icon style; if partner logos are displayed, add a legend explaining what each marker type represents.
When findings meet institutional resistance
The research findings were presented to the relevant product and engineering stakeholders. The response highlighted a tension common in UX practice: evidence-based recommendations encountering a strategic narrative that overrides them.
"We're not planning to make changes to the ATM locator at this time. The organization's direction is as a digital-first financial institution, physical ATM and branch access is not a strategic priority, and we don't want to invest in optimizing an experience we're moving away from."
A "digital-first" positioning doesn't eliminate users' need to access physical cash, particularly among the members most likely to use ATMs regularly (the screener for this study). A failed search doesn't disappear because the organization has a digital strategy; it creates mistrust and abandonment in the moment a member needs help. The recommendation to at minimum improve error messaging, a low-cost, low-effort change, was not accepted.
This outcome is documented here not as a criticism, but as a demonstration of a critical skill for senior researchers: producing rigorous evidence, presenting it with strategic framing, and navigating organizational decisions that don't align with user needs, while maintaining credibility for future influence opportunities.
What this study demonstrates beyond the findings themselves:
Due to the confidential nature of the project, supporting documentation is not provided within the case studies, but can be viewed during an in-person interview upon request.