The UX Audit Checklist: 47 Points We Check Before Any Redesign
Published:
Reading time:
Category:
Content
Most redesign briefs start in the wrong place. A founder sees a low activation rate or a high drop-off on step three of onboarding and immediately reaches for a new layout. The problem is that a layout change without a diagnosis is a guess, and guesses are expensive. This UX audit checklist is the diagnostic that runs before the redesign brief gets written.
It is the same 47 checks Morphic runs on client products before recommending anything. It sits behind engagements like the SaaS onboarding fix that lifted activation from 34% to 61% in 90 days, and the PDP quality-selector redesign at MissPompadour that produced a 10.88% lift in revenue per user for new visitors over a 28-day A/B test at 94% significance. The checklist did not produce those results; the precise diagnosis did. The checklist is what made the diagnosis systematic rather than a series of educated opinions.
Morphic is publishing it because the checklist is not the moat. The judgement about what the findings mean is.
One Rule Before You Start
Pick the metric this audit serves before checking a single point: activation rate, trial conversion, checkout completion, or whichever number matters most right now. An audit without a target metric produces a long list of things someone does not like. An audit anchored to a metric produces a sorted queue of changes, ordered by their expected impact on that number.
Every check below is easier to answer once you know what you are optimising for. Some findings will matter enormously for your metric; others will not, even if they are technically wrong. That distinction is the judgement layer the checklist cannot provide. It comes from knowing your product, your users, and the number you are trying to move.
Category 1: Funnel and Analytics Forensics (Points 1-7)
Ground the audit in behaviour before opinion. Funnel drop-off is not random: Baymard Institute's decade of checkout usability testing has documented over 140 distinct causes of cart abandonment, finding the average large e-commerce checkout contains 23.48 form elements when the practical minimum is closer to 12. The same principle applies to SaaS onboarding, signup sequences, and any funnel with a conversion target at the end.
Funnel mapped end to end: every step from arrival to the target metric, with conversion rate per step documented. The biggest single-step drop-off is your audit's centre of gravity.
Device split per step: a mobile rate below roughly 70 percent of desktop at any step is a finding in itself; Google's research found 53 percent of mobile visits are abandoned when load time alone exceeds three seconds.
Traffic-source behaviour differences: paid, organic, and referral users failing at different funnel points usually signals an expectation mismatch set before the product loads.
Drop-off recency: diff your current funnel numbers against your changelog, because a new drop-off with a deployment timestamp is a bug, not a design problem.
Rage signals: repeated clicks, rapid back-navigation, and form restarts in session data are the closest thing to watching a user shout.
Time-on-step outliers: long time-on-step with high drop-off is almost always a copy or information architecture problem.
Segment survival: if your target customers convert at a lower rate than general traffic, the product is optimised for the wrong visitor.
Category 2: First-Contact and Comprehension (Points 8-14)
The five-second layer: does a stranger understand what this is and why it matters? The most common first-contact failure is jargon, meaning internal vocabulary leaking into the interface, a category of problem teams are structurally blind to because the team knows what every label means.
Value proposition passes the stranger test: someone outside your industry can say what the product does and why it matters after five seconds.
Primary action is unmistakable: one visually dominant next step per screen, not three equally-weighted CTAs competing.
Jargon inventory: list every label a new user must understand before acting; each one is a friction point.
Price and commitment clarity: what this costs and what happens after the click are answerable before the click.
Trust surface at first contact: social proof and security signals positioned where doubt actually occurs, near forms and payment steps, not just in the footer; Baymard found 18 percent of users actively look for security indicators before entering payment details.
Speed as design: first meaningful render on a mid-range phone connection, because slow reads as broken before any design decision is evaluated.
Expectation continuity: the ad, search result, or post that brought users here promises what the first screen delivers.
Category 3: Flow Friction (Points 15-24)
Walk every critical flow as a first-timer, on both desktop and mobile. This is where the why behind Category 1's numbers lives. Baymard's checkout research found nearly one in five shoppers abandons a purchase due to a too long or complicated process, and that a 20%-60% reduction in form elements is achievable for most checkouts without removing any necessary information.
Premature asks: every demand for card details, permissions, configuration, or data placed before the user has experienced value is borrowing activation points the product has not yet earned.
Decision load per step: count decisions, not fields; screens demanding choices users lack context to make are the classic activation killer.
Reversibility signalled: unsignalled permanence causes abandonment even when nothing is actually permanent.
Progress visibility: in any multi-step flow, users know where they are and how much remains.
Dead-end inventory: screens where the task is technically complete but the next step is invisible; analytics logs these as drop-offs, users experienced them as endings.
Form field necessity: every field justified against the target metric; the ideal checkout contains roughly 12 form elements and most audited sites have nearly double.
Input forgiveness: fields accept paste, autocomplete, and format variation.
Error recovery in place: mistakes correctable inline without restarting sections.
Interruption survival: lost progress on return is a silent churn machine, particularly on mobile.
Mobile path walked, not assumed: the complete critical flow completed on an actual phone, because rendering correctly is not the same as designing correctly for mobile context.
Category 4: Conversion and Persuasion (Points 25-32)
At every moment you ask the user for something: have you earned it yet? The MissPompadour PDP quality-selector work is a useful frame. The original interface had the information users needed, but the layout buried it under scrolling and dense text. The redesign surfaced comparison points and reduced cognitive load without changing what was being asked. The result: +10.88% revenue per user for new visitors, +3.16% conversion for returning visitors, and +EUR 8.88 average order value for new desktop visitors, all from a 28-day A/B test at 94 percent significance. The mechanism was removing friction from the decision, not adding persuasion to it.
Ask-to-value sequencing: value demonstrations precede escalating commitments.
CTA language states outcomes: "Start my audit" beats "Submit."
Pricing page answers the four buyer questions: what it costs, what I get, what happens at the limits, and how I leave, all without a sales call.
Objection handling positioned at the objection: guarantees and cancellation clarity sit at the commitment point, not three screens away.
Comparison readiness: buyers will compare you to alternatives; the question is whether that happens on your page or off it.
Friction asymmetry check: if the paid action requires more steps than the free action, you are training users to stay free.
Urgency honesty: fake countdown timers convert short-term and poison long-term trust.
Post-conversion confirmation: the moment after paying confirms the decision was right and routes to value immediately.
Category 5: Retention and Return (Points 33-40)
The audit layer almost everyone skips, and where churn quietly lives. Most audits stop at conversion; the retention layer is where the value of the product is realised or lost. Research on SaaS churn consistently identifies the first two weeks after activation as the period with the highest drop-off risk, and most post-activation problems are not product quality problems, they are product discoverability problems.
First-win moment designed: the first meaningful output is unmistakable and acknowledged, not a silent state change.
Empty states teach: "Create your first report to see performance by channel" is a retention mechanism; "You have no reports yet" is an apology.
Habit hook exists: a designed reason to return, built into the product rather than left to email marketing.
Feature discovery pathways: undiscovered features are unrenewed contracts.
Re-onboarding for returners: a user back after three weeks is neither new nor current, and the product should reorient rather than assume.
Notification economy: users who mute everything because signal-to-noise collapsed are lost before they officially churn.
Help where the trouble is: each recurring how-do-I question in your support log is an interface failure with a support salary attached.
Cancellation flow audited honestly: dark-pattern exit flows trade churn data and reputation for a one-month retention number.
Category 6: Consistency, Accessibility and Resilience (Points 41-47)
Infrastructure-layer checks: invisible when right, corrosive when wrong. Accessibility failures cluster exactly where conversion failures do, in forms and flows, and that is not a coincidence. The WCAG AA standard is a useful proxy for baseline quality regardless of whether your user base includes users with disabilities, and statistically, it does.
Same action, same name, everywhere: "Save" vs "Update" vs "Apply" across sections erodes the user's confidence in their own understanding.
Contrast meets WCAG AA: a minimum ratio of 4.5:1 for body text and 3:1 for large text and UI components, automated scan plus human verification.
Touch targets at recommended minimums: WCAG recommends 44x44 pixels with breathing room on dense screens, and Apple's and Google's platform guidelines specify similar minimums.
Keyboard and screen-reader survival of critical flows: the complete activation and payment flows navigable without a mouse and announced correctly.
State coverage: loading, error, offline, and slow-connection states exist and inform rather than alarm.
Internationalisation tolerance: long names, foreign formats, and translated string lengths do not break layouts, and fixing layout assumptions costs ten times more after translation than before.
Degradation behaviour: a well-designed failure state preserves the user's context and gives a path forward; a blank screen or a technical error code does neither.
Scoring: From 47 Checks to a Next Sprint
Raw findings are noise until they are prioritised. Score every failed check on two axes: estimated impact on your chosen metric (1 to 5) and estimated effort to fix (1 to 5). Plot the results in a 2x2 matrix. The high-impact, low-effort quadrant is your next fortnight. In most audited products it contains three to five fixes, often shippable without major design work.
The rest of the matrix matters too: high-impact, high-effort findings become sprint planning inputs; low-impact, low-effort fixes are polish items; low-impact, high-effort problems go to the backlog indefinitely. The matrix forces the prioritisation conversation that most audit findings skip, which is why most audit findings produce no action.
When to Run It Yourself vs Bring In Senior Eyes
Run it yourself when the budget says so. Honestly executed, this checklist finds real problems. The limits of self-auditing are structural, not motivational: your team knows where everything is and what every label means, which is exactly the knowledge your users lack and the friction you therefore cannot feel. Categories 2 and 3 in particular reward stranger eyes, because the comprehension and flow checks are hardest to answer honestly when you built the product.
If you would rather have senior eyes run all 47 points against your actual data, with findings delivered as an effort-scored, ticket-ready fix list inside two weeks, Morphic runs this diagnostic through its user research service and inside the design kick-off workshops included in both plans. Either way: diagnose before you redesign. It is the cheapest decision in this entire process.
The Bottom Line
Most redesigns are solving the wrong problem. A nav revamp when the drop-off is at the form. A visual refresh when the issue is a premature permission ask on screen two. The audit exists to make that diagnosis before the brief goes to a designer. Run the 47 checks honestly, score by impact and effort, and you will find a shorter list of higher-value fixes than any redesign was going to produce.
Want senior eyes on it instead, delivered as a ticket-ready fix list? Start with the free 3-day trial and see the process before paying anything: see pricing.
Key Takeaways
A layout change without a diagnosis is a guess, so this 47-point checklist runs before any redesign brief gets written.
Pick one target metric before checking a single point, because an audit for a metric produces a sorted fix list while an audit of everything produces opinions.
Baymard Institute documented over 140 causes of cart abandonment and found the average checkout has 23.48 form elements when roughly 12 is the practical minimum.
The retention layer (points 33 to 40) is the audit layer most teams skip, and most post-activation problems are discoverability problems, not quality problems.
Score every failed check on impact and effort: the high-impact low-effort quadrant usually holds 3 to 5 fixes shippable within a fortnight.
Emon Datta
Founder, Morphic Agency
Morphic Agency is a Melbourne-based UX design agency. We've shipped 50+ products across 15+ countries.
