Edward Jones UX Evaluation

Quantifying friction in advisor tools before a line of code was written

Role Lead Product Designer at World Wide Technology
Type Design strategy · Expert evaluation
Evidence Overview and detailed reports plus my evaluator notes, retained
Measurement None retained; recommendations handed off pre-development

Overview

One of three expert evaluators on a quantitative UX evaluation for Edward Jones' Client Care and Commitment Assessment, covering the firm's Account Opening and Financial Statements prototypes and its design system. Each evaluator worked independently against Nielsen's ten usability heuristics and scored every task step with the PURE method; scores were then combined using an inter-rater reliability statistic, so the final ratings reflect measured consensus rather than any one reviewer's taste.

The Challenge

Edward Jones' UX team had produced InVision prototypes for two critical advisor workflows through a series of design ideation workshops. Before committing them to development, the firm needed a defensible answer to a hard question: where will these designs create friction, and how severe will it be? The evaluation had to be more than a critique; it had to produce numbers the organization could rank fixes by, and a baseline it could re-measure against after iterating.

Approach

Three evaluators with over 25 combined years of UX experience worked the prototypes independently, scoring each step on PURE's three-point cognitive-load scale and logging every heuristic violation with a severity rating. We combined the independent scores statistically and summed friction per feature, so 'Complete Owner Information' and 'Create Network Profile' could be compared on one chart. Each finding shipped as an annotated screen with a concrete recommendation. A companion audit compared both prototypes against the EJ Design System across seven domains (accessibility, color, grid, iconography, typography, components, patterns), recording for each guideline whether a specification existed at all.

01 What the prototypes were held to

The rubric, fixed before anyone looked

Nielsen’s ten heuristics, printed in the report in full. Three evaluators worked the prototypes independently against this list, so every finding had to name the principle it violated rather than rest on taste. It is why the findings pages read “Consistency and standards” and “Error prevention” instead of “this feels wrong”.

untested From the retained evaluation report. A published rubric, not a measurement; what it buys is that three reviewers were pointing at the same ten things.

Forty-four, and what the count could not see

The whole engagement in one page: 44 points of friction of varying severity, most of them in consistency, error prevention and efficiency of use. It also states its own limits, which is the more useful half: the evaluators are experts in heuristics but not advisors, the prototypes were InVision rather than software, and the scope stopped at the workflows each prototype happened to demonstrate.

untested The retained Executive Summary. Every limit quoted here is the report’s own wording, not a caveat added afterwards. The overview and approach on this page describe PURE scoring and a combined inter-rater statistic; the severity labels on the findings pages corroborate the three-point scale, but no page in this archive shows the statistic, so treat that part as my account.

02 Where the friction was

A finding is a severity, a heuristic, and a fix

Account Opening, three findings deep. Each one carries a severity, the heuristic it breaks, the crop it was found in, and a concrete recommendation: group the “Select all” control with the checkboxes it governs; give secondary actions a button hierarchy so an action reads differently from a link; hold duplicate-to-other-accounts changes in application state and commit them on “Save and Continue”. Nothing here is a note to think about later.

untested From the retained findings pages. Severity is three expert judgements against a fixed scale, not a measurement of anybody using the prototype.

The same rubric, a second prototype

Financial Statements, scored the same way, and the severities spread: two notable, two easy. The recurring shape of the engagement is visible here; read-only values sitting in input fields that invite typing, and primary actions with no more weight than the action that discards the work. Both are consistency problems, which is why the recommendations point at the design system rather than the screen.

untested From the retained findings pages. The file is named for a friction chart; the page is the second prototype’s findings, and no summed per-feature chart survives in this archive.

03 Design-system audit

Solve the interaction once

The audit’s argument for going after the system instead of the screens: set a standard for a hard interaction, then apply it across the product suite so people only learn it once. Six domains were checked against the two prototypes (accessibility, color, grid, iconography, typography, patterns), and for each one the audit records the same blunt thing: whether a specification exists at all.

untested The retained audit overview lists six domains. The narrative above says seven, counting components; no page in this archive shows a components domain, so six is what the record supports.

Five of six accessibility guidelines unspecified

The audit’s sharpest page, and the one that needed no scoring at all. Of six accessibility guidelines, the design system specified one; references. Structure and hierarchy, keyboard navigation, screenreader text, images and video, and colour all came back NO, which means every product team was deciding accessibility for itself, in private, repeatedly. A yes/no column is a harder finding than a severity score.

untested The retained accessibility audit. Present-or-absent is checkable by anyone with the same design system; nothing here rests on the evaluators’ judgement.

Outcome

The evaluation documented 44 points of friction of varying severity across the two prototypes. The recurring patterns: read-only information dressed in interactive-looking components, primary actions without visual hierarchy, alerts placed far from what they referenced, and workflows stitched across disparate underlying systems: the largest single source of friction. The design-system audit found no specification for structure and hierarchy, keyboard navigation, or screen-reader text, which meant every product team was solving accessibility alone. Recommendations were deliberately scoped as design-system fixes where possible: solve the interaction once, apply it everywhere.

Measurement note

The friction counts and severity scores come from the retained evaluation reports. The work dates to 2019 and assessed pre-development InVision prototypes, so these figures describe designs under review, not software Edward Jones shipped. No post-development measurement was retained.

enesru