Friction is found by walking, not by reviewing
Nobody who built a booking journey can see it. This is not a failure of attention; it is a property of knowledge. The person who configured the system knows which button leads where, knows what the confirmation will say, and knows that the deposit is redeemable. A patient knows none of these things and finds out by trying.
The audit is therefore conducted by walking, not by reviewing. Somebody unfamiliar with the clinic attempts to book, on a phone, on a normal connection, without help, and records every point at which they hesitate, guess or stop. Hesitation is the signal. A patient who pauses to work out whether the deposit is refundable has encountered friction whether or not they continue.
The criteria are ordered by when they occur in the journey, which makes the resulting list actionable in sequence. Early criteria affect everybody; later ones affect the subset who reached them. A clinic with a blocking fault at criterion two gains nothing from perfecting criterion eight.
Two criteria are included that are not obviously about friction. Accessibility appears because a journey that excludes people is the most complete form of obstruction. The failure path appears because a journey that breaks silently is one the clinic will never fix, since nobody who experienced it will report it.
How to run the walk
Recruit somebody outside the clinic. A friend, a relative, anybody who has not booked with you. Give them one instruction: book a consultation for a named procedure, using their phone, and speak aloud everything they are thinking.
Record the session, or take notes yourself, and do not help. The urge to help is strong and it destroys the data. Every intervention is a friction point you have just concealed.
Note the timestamp of every hesitation, question and error. Afterwards, ask three questions: what did you think this was going to cost, when did you think the appointment was, and what did you think would happen next. Wrong answers indicate ambiguity even where the journey completed.
Then repeat the walk yourself for the routes the first walker did not take: telephone, social message, and the third-party listing if you have one. Score each criterion by the worst route a patient would realistically use.
Score the accessibility criterion separately using a keyboard alone, as described in the accessibility and readability audit.
Common scoring errors
Walking it yourself first. You will complete it and conclude it is fine. Start with somebody else.
Scoring on a desktop. Score the phone journey and treat desktop as secondary. The failures are on the phone.
Counting a long form as thoroughness. Clinical history collected before an appointment exists is a burden placed on the wrong side of the commitment, and it is also personal data collected before there is a clear basis for it.
Treating an emailed confirmation as immediate when it arrives in twenty minutes. The patient has left the page. Score by what happens on the screen as well.
Ignoring the deposit terms because staff explain them by phone. The criterion asks what is stated before payment.
Fixing friction without creating pressure
There is a version of friction reduction that shades into pressure: countdown timers, scarcity messages, prefilled commitments and default opt-ins. These raise completion and lower the quality of the decision, and in a clinical context they interact badly with consent. This instrument scores none of them as improvements, and the consent and cooling-off checklist scores some of them as failures.
The distinction worth holding is between removing an obstacle and removing a decision. Showing availability removes an obstacle. Preselecting the soonest appointment removes a decision. Stating the deposit terms removes an obstacle. Making the deposit non-refundable in small print removes recourse.
Fix in the order the journey happens, and re-walk after each change with a new person. Walkers become expert quickly and stop finding things after the second attempt.
Finally, connect the failure path to somebody. Errors that are logged and unread are not logged. Once a fortnight, somebody should look at what broke and for whom. That habit does more for booking performance over a year than any single change to the form.