Friction audit
Cataloguing the small hassles in a process to decide where to remove friction — or deliberately add it.
What it means
A friction audit systematically locates the points of effort, delay, and difficulty along a behavioral pathway and evaluates each as either an obstacle to remove or a useful brake to keep or add. Unlike a sludge audit, which is concerned specifically with friction that harms the user, a friction audit is neutral about direction: sometimes the goal is to smooth a desired behavior (one-click checkout), and sometimes to introduce 'good friction' that slows a harmful or impulsive one (a confirmation step before a risky transaction). It draws on a robust but context-dependent finding: small, seemingly trivial friction can move whether people act by far more than its objective cost would suggest, though how much varies widely with the situation. The skill lies in matching the direction of friction to whose interest is served. It matters because friction is a powerful and often underused lever in design — and the same tool can either help or exploit, depending on intent.
How it works
The method begins by mapping a behavioral pathway end to end — every step a person must take to move from an intention to a completed action — and then inspecting each step for the effort, delay, and small hurdles that behavioral scientists call friction. Friction is anything that raises the cost of acting without changing the underlying incentive: an extra form field, a login wall, a phone tree, a waiting period, a document that must be found and uploaded. For each point, the auditor asks two questions. First, how much does this step actually deter people, relative to how trivial it looks? Second, in which direction should it move — should it be smoothed away, left in place, or deliberately strengthened? The answer to the second question is what separates a friction audit from a simple usability review. Smoothing is appropriate when the behavior serves the person taking it and the friction is merely a nuisance. Adding or keeping friction, often called 'good friction,' is appropriate when a moment of effort protects against a choice the person would regret, such as an irreversible transfer or a purchase made in haste. The audit is therefore as much an ethical exercise as a design one: it forces an explicit judgment about whose interest each step serves.
What the evidence shows
The premise that small barriers carry disproportionate weight has deep empirical roots. In a 1965 experiment, Howard Leventhal and colleagues found that university students who received a campus map and a prompt to schedule a specific time were far more likely to get a tetanus inoculation — roughly a tenfold increase over students given the same health information without those logistical aids — while the intensity of the accompanying fear appeal made little difference to behavior. Kurt Lewin's term for such small facilitating or blocking factors, 'channel factors,' captures the core idea. Later field studies sharpened it. Brigitte Madrian and Dennis Shea (2001) showed that switching a firm's retirement plan to automatic enrollment raised new-hire participation to around 86 percent, far above the level under the prior sign-up regime, because the effort of enrolling had itself been a genuine barrier. Bettinger, Long, Oreopoulos, and Sanbonmatsu (2012) found that families randomly offered help completing the federal student-aid form were substantially more likely to file it and to enroll in college, whereas families given only information about aid were not — evidence that the friction of the paperwork, not a lack of knowledge, was the binding constraint. A meta-analysis of default effects (Jachimowicz et al., 2019) reported a medium-to-large average effect of about d = 0.68, but with wide variation that depended heavily on domain and on whether the default matched what people already wanted. The broader claim that choice-architecture 'nudges' reliably move behavior is more contested: a large meta-analysis (Mertens et al., 2022) estimated an average effect near d = 0.43, yet a re-analysis correcting for publication bias (Maier et al., 2022) concluded that the overall effect was no longer distinguishable from zero. The friction-specific findings — auto-enrollment, form simplification, channel factors — remain among the better replicated, but the size of any given effect is an empirical question, not a guarantee.
Related but distinct
The closest relative is the sludge audit. Richard Thaler (2018) coined 'sludge' for friction that works against a person's own interest: the excessive paperwork, hidden cancellation paths, and rebate hoops that firms and bureaucracies use to discourage actions people are entitled to take. Cass Sunstein (2022) proposed the sludge audit as a systematic review of such frictions inside an organization, and several governments have since run them on their own forms and processes. A friction audit shares the mapping technique but not the moral starting point: it treats friction as directionally neutral and asks, step by step, whether effort should be removed or added, rather than assuming all of it is harmful. It also overlaps with choice architecture, the broader practice of designing the context in which choices are made, and with usability analysis, which typically aims only to reduce effort. What distinguishes the friction audit is its willingness to recommend more friction where a pause protects the chooser, a move a pure efficiency review would never make.
Using it in practice
A workable audit tends to proceed in stages. The pathway is drawn out and each step is instrumented so that drop-off can be measured rather than guessed — where people abandon a form, how long a step takes, how many return later. Each friction point is then classified by whose interest it serves: the person acting, the organization, or a third party. Points that impose cost on the person for the organization's benefit are candidates for removal; points that impose a small cost on the person for the person's own protection are candidates to keep or strengthen. Changes are then tested rather than assumed, ideally against a control, because the direction and magnitude of a friction effect are hard to predict from inspection alone. Removing a confirmation step may speed desired orders and also multiply mistaken ones; adding a cooling-off period may curb regret and also deter legitimate use. The discipline of the method is that it refuses to treat less effort as automatically good or more effort as automatically protective, and instead makes the trade-off visible and measurable.
Limits and caveats
Several cautions apply. The judgment about whose interest a step serves is rarely as clean as the method implies; the same confirmation screen can be described as consumer protection or as a conversion-killing obstacle depending on who is asked, and organizations have an obvious incentive to classify self-serving friction as protective. Good friction can shade into paternalism, or into sludge wearing a virtuous label, when the added effort mainly serves the designer's preferences rather than the chooser's welfare. Effect sizes are heterogeneous: a friction that halves take-up in one setting may barely register in another, so an audit that assumes a large effect can misallocate attention. Because the same technique smooths desired behavior and obstructs unwanted behavior with equal ease, it can be turned to manipulation as readily as to help — the identical map of a checkout flow can be used to strip out dark patterns or to install them. Finally, friction interacts with fairness: barriers that look trivial to a designer can fall much harder on people with less time, lower literacy, or less reliable technology, so a step judged a minor brake in the aggregate may function as a wall for some. A responsible audit records not only how large each friction is but on whom it lands.
Examples
A bank's friction audit removes steps from opening a savings account (helpful) while adding a deliberate pause and warning before a large irreversible transfer (protective good friction).
A retailer audits its checkout and finds a mandatory account-creation step before payment; removing it in favor of guest checkout smooths a desired behavior, while a brief 'confirm your delivery address' pause added before an expensive express order is a deliberate brake in the other direction.
A plan administrator audits the retirement sign-up path and discovers that new hires must choose a contribution rate and fund allocation on their first day; switching to automatic enrollment with an easy opt-out removes the friction that had kept many from ever starting.
A trading app inserts a mandatory cool-down screen and a typed confirmation before a highly leveraged trade, treating impulsive risk-taking as a behavior worth slowing rather than smoothing.
A public agency audits a benefits application and finds claimants must upload the same identity document three times across separate portals; the audit flags this as sludge to remove, not a useful brake, because the effort serves no one but the process.
First described in Behavioral design practice.
Key references
- Bettinger, E. P., Long, B. T., Oreopoulos, P., & Sanbonmatsu, L. (2012). The role of application assistance and information in college decisions: Results from the H&R Block FAFSA experiment. The Quarterly Journal of Economics, 127(3), 1205-1242. doi.org/10.1093/qje/qjs017
- Jachimowicz, J. M., Duncan, S., Weber, E. U., & Johnson, E. J. (2019). When and why defaults influence decisions: A meta-analysis of default effects. Behavioural Public Policy, 3(2), 159-186. doi.org/10.1017/bpp.2018.43
- Leventhal, H., Singer, R., & Jones, S. (1965). Effects of fear and specificity of recommendation upon attitudes and behavior. Journal of Personality and Social Psychology, 2(1), 20-29. doi.org/10.1037/h0022089
- Madrian, B. C., & Shea, D. F. (2001). The power of suggestion: Inertia in 401(k) participation and savings behavior. The Quarterly Journal of Economics, 116(4), 1149-1187. doi.org/10.1162/003355301753265543
- Maier, M., Bartos, F., Stanley, T. D., Shanks, D. R., Harris, A. J. L., & Wagenmakers, E.-J. (2022). No evidence for nudging after adjusting for publication bias. Proceedings of the National Academy of Sciences, 119(31), e2200300119. doi.org/10.1073/pnas.2200300119
- Mertens, S., Herberz, M., Hahnel, U. J. J., & Brosch, T. (2022). The effectiveness of nudging: A meta-analysis of choice architecture interventions across behavioral domains. Proceedings of the National Academy of Sciences, 119(1), e2107346118. doi.org/10.1073/pnas.2107346118
- Sunstein, C. R. (2022). Sludge audits. Behavioural Public Policy, 6(4), 654-673. doi.org/10.1017/bpp.2019.32
- Thaler, R. H. (2018). Nudge, not sludge. Science, 361(6401), 431. doi.org/10.1126/science.aau9241