The Friction Audit

Understanding, Not Disclosure

Operating Model

All Sectors

Across financial services, firms can show that a fee was disclosed, an exclusion stated, a rate change notified, a key risk flagged — and yet customers reach the moment of consequence and report being surprised. The information was technically present in the journey. It was structurally invisible: pre-ticked, defaulted, layered behind a click-through, presented after the decision had effectively been made. The Consumer Understanding outcome reframes this as an outcome failure, not a disclosure success. Most firms have removed friction systematically over a decade of conversion optimisation. Few have a repeatable discipline for finding where positive friction is now absent.
The structural move is to make friction audit a repeatable operating-model discipline — owned, scheduled, evidence-driven, and consequential — rather than a one-off review. The audit identifies where the journey routes the customer past a moment of consequence without surfacing it, and produces a redesign that reinstates positive friction where it is structurally absent:
Owned, scheduled, repeatable
The audit has a named owner accountable to a defined governance forum, a published cadence (typically annual for the full journey set, with high-risk journeys reviewed more frequently), and a documented method that combines journey walkthrough, transaction sampling, behavioural-data analysis, frontline-agent feedback, and customer co-observation. The output is a prioritised set of journey points with evidence of invisibility, a redesign commitment with owner and date, and a post-redesign retest. The design test: can the firm produce, for any major customer journey, the date of the last audit, the findings, the actions taken, and the evidence those actions changed customer experience?
Anchored at high-yield moments
The audit prioritises the journey points where positive friction is most likely to be structurally absent and where its absence carries the highest consequence — fee-incurrence, rate change, renewal, claim, exit, drawdown, and any point where a default operates in the firm's commercial favour. The FCA's digital design review identified bias in layouts, pre-selected defaults, lack of product information, and 'speed over customer interests' as recurring patterns at exactly these moments. The audit walks each of these moments as the customer most likely to be harmed (low financial confidence, time pressure, vulnerability), not as the standard digital native the journey was designed for. The design test: does the audit's coverage map to the moments of consequence in the customer lifecycle, and does it test those moments through the lens of the customer cohort most likely to miss them?
Evidence-driven, not walkthrough-only
The audit triangulates journey walkthrough with data the firm already holds: drop-off rates at high-consequence steps, dwell-time anomalies, call-driver themes, complaint categories, chat transcripts, and frontline-agent feedback on what customers say they did not understand. The Consumer Understanding review (March 2026) named good practice as 'analysing insights from multiple sources, including call listening, complaints, chat transcripts, website analytics, drop-off data, and surveys; reviewing this evidence regularly and prioritising meaningful improvements rather than simply making cosmetic changes.' Drop-off data alone is ambiguous; in combination with complaint themes and call drivers it becomes diagnostic. The design test: for any flagged friction-deficit, can the firm cite at least two independent evidence sources confirming it, and did the redesign measurably move at least one of those signals?
- Each major customer journey has a documented audit owner, a most-recent audit date within the published cadence, and a redesign commitment register showing actions taken, owners, and post-redesign retest evidence.
- Audit findings are triangulated across at least three independent evidence sources (journey analytics, call drivers, complaint themes, chat transcripts, frontline-agent feedback, customer testing) — and a finding supported by only one source is marked as preliminary rather than actionable.
- Where the audit identifies a friction-deficit and the firm reinstates positive friction, post-change MI shows the intended customer-understanding signal moving (call-driver decline, complaint-theme reduction, comprehension-test improvement, journey-abandonment shift at the right point) — and the absence of such movement triggers a further iteration rather than a closed action.
- The audit population deliberately includes customers most likely to be harmed by friction-deficits — low financial confidence, vulnerable circumstances, low digital confidence, time pressure — and the redesign is evaluated against outcomes for those cohorts, not only against the population average.
- A retail bank, after the FCA's digital design review, formalised a quarterly friction audit covering its four highest-consequence digital journeys: arranged overdraft drawdown, foreign-transaction confirmation, savings-rate change acknowledgement, and mortgage product-expiry decision. The audit team — joint between UX research, customer experience analytics, and a Consumer Duty Champion lead — worked from a defined evidence base for each journey: drop-off and dwell-time data at every step, the top ten call drivers in the four weeks following the journey, complaint-theme extraction from the same period, and chat transcript sampling. The overdraft journey audit found that the fee disclosure screen had a median dwell time of 1.8 seconds and a 96% pass-through rate, with subsequent calls to the helpline citing 'I didn't know there would be a charge' as the most common driver. The redesign added a positive-friction step that surfaced the projected fee in pounds for the requested overdraft amount, required an explicit confirmation at that figure, and offered a callback to a money-guidance specialist for customers in arrears. Drop-off at the new step rose, as expected — but the call-driver category fell by more than a third within two quarters, and complaint volume on the same theme fell with it. The journey was no longer optimising for the speed of overdraft drawdown; it was optimising for whether the customer knew what the drawdown would cost.
- A wealth platform, responding to the FCA's research on digital engagement practices and investment outcomes (Occasional Paper 66), ran a friction audit on its drawdown-initiation journey. The audit combined journey walkthrough by trained customer-experience researchers, analysis of session recordings for customers approaching the drawdown decision, the top complaint themes from drawdown customers in the previous twelve months, and structured interviews with the platform's retirement-specialist team about what callers said they had not understood. The audit found three friction-deficits: the projected sustainability of the chosen withdrawal rate was disclosed in a layered tooltip that 78% of customers did not open; the tax implications of crystallisation were summarised in a regulatory-language paragraph that comprehension testing showed only 31% of customers correctly interpreted; and the irreversibility of certain drawdown choices was stated once, in a sentence positioned after the action button. The redesign brought the sustainability projection onto the decision page as a chart with explicit pounds-and-years framing, replaced the tax paragraph with a tested plain-language explainer and a 'show me an example' interactive, and reordered the irreversibility statement to precede the action with an explicit acknowledgement step. Post-redesign comprehension testing of the same cohort moved from 31% to 74% on the tax point; calls to the retirement specialist team rose at the decision moment (the intended consequence) and fell in the four weeks after drawdown commenced. The friction audit had not added friction to the journey indiscriminately. It had reinstated it at the three points the data identified as invisible to the customer.
Common failure modes
The most common failure mode is the audit that produces a report nobody acts on — a finding documented in a quality-assurance log, with no link to the product roadmap, no named owner of redesign, and no commitment to retest. A second is conflating positive friction with friction in general: adding a confirmation modal at every step is not an audit response, it is a usability regression that customers will learn to dismiss without reading. The audit must be selective — friction reinstated at moments of genuine consequence, not sprinkled across the journey. A third is over-relying on a single signal: drop-off rates alone do not tell the firm whether a customer abandoned the journey because they chose to or because the journey defeated them. The FCA explicitly warned against firms being overly reliant on positive online review ratings as an indicator that there are no significant issues, and the principle applies more broadly — multiple indicators are needed. A fourth is auditing the digital flow in isolation from the human channels: a fee disclosure may be invisible online but well-handled in branch, or vice versa, and the audit must follow the customer across channels rather than testing each in a silo. A fifth is treating the audit as a compliance exercise rather than an outcome exercise — producing evidence that the disclosure exists rather than evidence that the customer can see and act on it.

Related Patterns

Evidence & Supporting Facts

FCA Digital Design in Customers' Online Journeys (July 2025) identified 'lack of friction' as a named area for improvement: 'Positive friction in journeys can drive good outcomes... Firms can consider if adding friction to their journey could help customers achieve good outcomes.' The review also flagged bias in layouts, lack of product information, and 'speed over customers' interests' as recurring poor practice — the three friction-deficit patterns the audit is designed to surface.

FCA Consumer Understanding Review (March 2026) identified good practice as analysing 'insights from multiple sources, including call listening, complaints, chat transcripts, website analytics, drop-off data and surveys' and reviewing the evidence 'regularly' to prioritise 'meaningful improvements rather than simply making cosmetic changes' — the operational core of the audit method. The same review named reliance on sales data or absence of complaints as evidence of understanding as explicitly inadequate.

FCA Consumer Understanding Review (March 2026) identified governance and oversight failure modes that the audit is designed to remediate: unclear accountability ('who is responsible for decisions, or how those decisions are made'), one-off initiatives without governance, and weak feedback loops where 'monitoring was described but there was no clear evidence of a process to feed this information back into governance or to use it to improve communications.'

FG22/5 Chapter 8 (Consumer Understanding outcome) requires firms to support customer understanding — not merely to provide information. The friction audit is the operating-model discipline through which a firm can evidence that the journeys it has designed do support understanding at the moments of consequence, rather than asserting it from the existence of disclosures.

AI Automation Lens

Journey analytics platforms can construct, at scale, the evidence base the friction audit requires: dwell-time distributions at every step, drop-off heatmaps, repeat-attempt patterns, page-view sequences, and the divergence between expected and actual paths. Machine learning can identify segments of customers whose journey behaviour diverges from the population average — accelerating through decision pages, abandoning at specific friction points, returning to the same screen multiple times — and surface them as candidate audit subjects. Natural language processing applied to call transcripts, chat logs, and complaint records can identify the linguistic markers of post-journey surprise: 'I didn't realise', 'nobody told me', 'where was that?'. These markers correlate to specific journey points and let the audit work backwards from the moment of harm to the moment the journey routed past. AI can also test redesigned flows synthetically before customer exposure — simulating user attention, skim-reading patterns, and decision-fatigue effects to flag where positive friction has been added in form but not in effect. The objective is to make the friction audit continuous rather than periodic, evidence-rich rather than walkthrough-dependent, and grounded in what customers actually do rather than what the firm intended them to do.