How to use Friction Analysis
Friction analysis is the diagnostic methodology behind most high-impact UX fixes: list every step, estimate its cost, and rework the most expensive ones first.
Friction Analysis
A diagnostic method that lists every step and decision in a user flow, scores each one's cost, and prioritises the highest-cost steps for redesign.
Friction analysis is a structured way to find out where a flow is leaking users. You walk the flow as the user does, write down every distinct action and every decision they're forced to make, then assign each one a cost — in time, cognitive load, emotional effort, or perceived risk. The steps with the highest scores become your redesign queue.
The method applies to any flow with a goal at the end: onboarding, checkout, account recovery, support ticketing, settings changes. It sits inside the broader practice of UX optimization but is method-agnostic — you can run it from a wireframe, a live site, or a recorded session.
Most flow problems aren't caused by one broken thing. They're caused by a stack of small frictions — a redundant confirmation, an unclear label, an unexpected field — each of which costs a few percent. Compounded across six or seven steps, that's how a 40% drop-off becomes normal.
Friction analysis is what you do before you start testing. It produces a ranked list of suspects, each tied to a specific step, so your A/B tests have hypotheses worth running instead of opinions worth arguing about.
Step 1 — Map every step and every decision
Start by walking the flow yourself, slowly, writing down each discrete action. A "step" is anything that requires the user to do something: click, type, choose, wait, read, decide. Steps that feel automatic to you — a cookie banner, a region selector — still count.
Then mark every decision separately. A step that offers three shipping options is one step but three decisions. Decisions are where users stall, abandon, or pick wrong, so they deserve their own row in the audit.
A typical Shopify checkout, mapped honestly, runs 18–25 steps and 6–10 decisions between cart and confirmation. If your map shows fewer than that, you've under-counted — go back and split the steps your eye glossed over.
Map from the user's machine, not yours
Auto-fill, saved sessions, and a fast dev laptop hide most of the friction. Walk the flow in a private window, on mobile data, with no autofill. The map you produce there is the one that matches your actual users.
Step 2 — Score the cost of each step
For each step, score four cost dimensions on a 0–3 scale: time (how long it takes), cognitive load (how hard it is to figure out), emotional cost (does it feel pushy, risky, or annoying), and risk (could the user lose progress or make a mistake they can't undo).
Sum the four scores into a single friction score per step. A score of 4 is a normal step that's working. A score of 8+ is a candidate for redesign. A score of 10+ is almost certainly hurting conversion right now.
Friction scores across a typical Shopify checkout flow
In the example above, the 3DS challenge and the shipping-method decision dominate the friction profile. That's where you focus — not on the email field, even if it's the first place users drop. Drop-off and friction are different things; friction analysis finds the cause, drop-off only flags the symptom.
Step 3 — Benchmark against realistic ranges
Scoring is subjective unless you calibrate. Compare your scores against typical ranges for the same step type, so a 7 on shipping-method choice means the same thing on Tuesday as it did on Monday.
The table below is the calibration sheet we use across audits. Treat it as a starting point — your flow's context will shift the bands, but the relative ordering tends to hold.
Typical friction-score ranges by step type
| Step type | Healthy | Watch | Redesign |
|---|---|---|---|
| Form field (single input) | 2–4 | 5–6 | 7+ |
| Choice between 2–3 options | 3–5 | 6–7 | 8+ |
| Choice between 4+ options | 5–6 | 7–8 | 9+ |
| External verification (OTP, 3DS, captcha) | 6–7 | 8–9 | 10+ |
| Wait / loading state | 2–3 | 4–5 | 6+ |
| Account creation requirement | 5–6 | 7–8 | 9+ |
| Upsell or cross-sell prompt | 4–5 | 6–7 | 8+ |
If a step scores in the Redesign band, the question is no longer whether to change it — it's what to test first. The score doesn't tell you the fix, but it tells you a fix exists and is worth the engineering hours.
Step 4 — Attack the highest-cost steps first
For each high-cost step, generate three hypotheses: remove it, defer it, or simplify it. Removal is the cleanest win and the rarest — most steps exist for a reason. Deferral (moving it later in the flow, when commitment is higher) is often the quietest win. Simplification — fewer choices, better defaults, clearer copy — is the most common.
Turn each hypothesis into a single A/B test with a measurable downstream metric. If you change the shipping-method step, the metric isn't "clicks on shipping" — it's checkout completion rate from that step forward. Friction analysis only pays off if the redesign survives a test.
Don't redesign the whole flow at once
It's tempting to rebuild every Redesign-band step in one release. Don't. You'll ship four changes, see a 6% lift, and have no idea which change earned it. Ship the top-scoring step first, measure, then move down the list.
Friction analysis FAQ
A heuristic review judges a screen against a checklist (Nielsen's ten, WCAG, mobile-first principles). Friction analysis judges a flow against the user's own goal, step by step. They overlap, but friction analysis is sequence-aware — the same field is fine in one position and painful in another.
No. The map and the scoring can be done from the live site alone. Session recordings are useful as confirmation — they show you where real users hesitate or rage-click — but they're not a prerequisite. Start with the map.
For a single flow (checkout, signup, account recovery), a complete analysis is 4–8 hours of focused work: 1–2 hours mapping, 2–3 hours scoring with a second reviewer, and 1–2 hours writing up hypotheses. Bigger flows scale linearly with step count.
Two reviewers, independently, then reconciled. Solo scoring drifts — you score the same step differently on day three than on day one. Two reviewers force you to defend each score, which is where the real insight comes from.
Friction analysis is the diagnostic phase of UX optimization. It tells you what to fix and in what order. UX optimization is the broader practice that includes the testing, the design work, and the iteration loop after the diagnosis is done.
Yes — it's flow-agnostic. The method works for onboarding, account recovery, returns initiation, support ticketing, settings changes, anything with a goal state. The cost dimensions (time, cognitive, emotional, risk) apply everywhere.
Look for asks the user didn't expect or didn't consent to: forced account creation, marketing opt-ins pre-ticked, surprise fees, language that implies commitment before they're ready. Each of those is +1 emotional cost. Real user research validates the scores, but the heuristic catches 80% of cases.
Always. Mobile typically scores 1.5–2x higher per step than desktop on the same flow, because small screens compress decisions and slow data networks change the wait costs. Run two scoring passes and prioritise mobile fixes first if mobile is your majority traffic.
That's a sign the flow needs structural rework, not field-level polish. Step back and ask whether the flow itself is the right one — can steps be removed entirely, can the order change, can the goal be reached in fewer screens. Then re-score the proposed redesign before building it.
Quarterly for high-traffic flows, plus after any release that touches the flow. Friction creeps in — a new tax line, a new compliance banner, a new payment option — and a quarterly cadence catches it before it costs a meaningful share of conversions.
Get an AI expert review of your site
Paste your URL — Metricuno's AI runs the same heuristic checks a senior CRO consultant would, scoring your page and prioritising the fixes that'll move conversion fastest.