In the evening, back at her hotel, our auditor would open a document and try to remember how long she had sat in the waiting room. She would write “about five minutes”, because that was honestly the best answer she had. We were selling a patient journey audit, and its durations came out of a whole day’s memory. This article tells how we fixed that: not by changing our audit method, but by building an application that follows it step by step and raises an alert as soon as a measurement is missing.
Accuracy is what makes the audit valuable
Our audit outline existed before the application and has not changed since: eleven sections, fifty questions, put to every clinic we visit in exactly the same way. The document held up; the answers written up after the fact, much less. Some stayed vague, others contradicted each other from one page to the next: the summary table and the detailed sections did not give the same duration for the same consultation.
Yet the value of an audit rests on measured gaps: the delay between booking an appointment and its confirmation, how much notice a reminder gives, the time between the end of the consultation and the moment the report is sent. Those are the figures that, once set against a clinic’s volumes, let us tell it what its current organization costs. An audit built on estimated figures produces an estimated conclusion, and a recommendation resting on it is hard to defend.
We built the application around the audit
So we built a measurement application, with one simple rule: it had to fit into the way we audit, not replace it. We kept the same outline, question for question. The report handed to the client stayed the same document. The only thing that changed is when the information is entered: during the visit, instead of in the evening.
The capture screen therefore mirrors how a visit actually unfolds. Six phases can be timed, from booking the appointment to leaving the clinic, and their timers are free to run at the same time, because the durations of a real journey overlap. Seven events are timestamped with a single tap, from the appointment booked to the report received. A notes field stays on screen at all times, to dictate things as they come and file them under the right section later. And because an auditor works standing up, phone in one hand, everything can be operated with a thumb, and nothing is lost to a network drop: every entry waits on the phone until the connection returns. Figure 1 walks through a demonstration visit and shows what the application records at each moment.
Eleven answers fill themselves in, the rest is written as before
Eleven of the report’s fifty questions are never typed in again: their answer follows from what was timed and timestamped. The confirmation delay is computed between the timestamp of the booking and the timestamp of the confirmation, the waiting-room time comes straight from its timer, and the channels used summarize themselves as “email and SMS” when a clinic sends its confirmation over both. These are precisely the answers the version written in the evening left empty or approximate.
Those computed answers appear read-only, together with an explanation of where they come from, and a button lets the auditor correct them by hand: she was in the room and the server was not, so her value always wins over a computed one. The other thirty-nine questions are filled in as before, with the same words as on the original document. Nobody had to relearn their job to use the tool.
A checklist before the export
During the visit, the application follows the outline like a checklist: every phase to time and every event to timestamp has its own box, and what is still missing can be seen at a glance.
Before the export, it walks through that list and raises alerts: it lists the fields that will print “Not recorded” in the report and flags the timers that are still running, because the duration of a phase that has not finished does not exist yet. Strict mode turns those alerts into errors: while one remains, the export waits. Figure 2 shows that check on the visit of figure 1, whose check-out timer is still running. It served from the very first campaign: two answers there printed “Not recorded” because no timer had run on the phases concerned, and the report said so instead of inventing a duration.
The result is a report that is both complete and accurate: complete because the whole outline is walked through and nothing goes missing in silence, accurate because every figure carries its provenance, a timer, a timestamp or the auditor’s own hand. That is also what makes it a reference point. Once a system is deployed, the same measurement simply has to be run again: the outline has not moved, and the two reports compare question by question, before against after. It is the tool for the stage called “We measure before we quote a price” on our method page.
Your process can get the same measurement
Auditing clinics is only the first process measured this way. The skeleton of the application does not change from one process to the next: the capture that survives a network drop, the computed answers, the checklist before export and the report produced by the server. What belongs to the process, its phases, its questions and its calculation rules, fits in a configuration file whose every version is kept. So we deliver a dedicated measurement application for every process we need to evaluate, one that speaks its vocabulary and produces its report, rather than a generic tool in which one would have to recognize one’s own process.
This is the first stage of our automation method: measure what actually happens, refuse to invent what the measurement does not say, then automate what deserves it. Those measured durations are the raw material we need to evaluate a process before putting a price on it.
If you audit sites, branches or points of sale on a regular basis, or if you want to put a number on an internal activity nobody has ever timed, the next step commits you to nothing. A 30-minute conversation, free of charge, is usually enough to describe your process and the application that would measure it. The question that matters is then what the measurement will show in your case. Write to us at the bottom of the page, or book a slot in our calendar directly.