A front desk return you can prove, counted from a baseline taken before go-live

The baseline makes the case

A front desk return is the difference between a period before go-live and a period after it, counted the same way. The decision rules for widening a pilot are in our guide to clinic AI pilot success criteria, and this page covers the counting that feeds them.

Four measures, each with a denominator

The Institute for Healthcare Improvement recommends 4 to 10 measures across outcome, process and balancing types to track whether a change is an improvement.¹ For the front desk, four measures carry the money, plus one balancing measure.

Reply time is the median minutes from a patient's first message to the clinic's first reply, per enquiry, split into opening hours and out of hours. The median keeps one weekend message answered on Monday from distorting the figure.

Conversion is the share of new enquiries that become a booked appointment within 14 days. The denominator is new enquiries, leaving out reschedules and questions about directions.

No-shows are appointments missed without notice, per booked appointment. Staff hours are the hours the team spends on messages, phones and payment chasing, logged for two sample weeks. The balancing measure is complaints or patient-reported problems per hundred bookings, so faster replies stay good replies.

Taking the baseline

Count for at least four full weeks before go-live, and use the same weeks of the previous year where the records exist. A year-on-year comparison removes most seasonal swings.

Hold everything else still while you count. A price change or a marketing push in the same month moves conversion and no-shows on its own.

Watch when the project started. A clinic often buys after an unusually quiet month, and the next month tends to look better whatever is done. A review in the International Journal of Epidemiology notes that this regression to the mean can make natural variation look like real change.² A longer baseline is the simplest defence.

Your numbers, not a benchmark

Set targets from the clinic's own numbers. A systematic review of 105 studies found an average no-show rate of about 23%, with regional averages from 13.2% in Oceania to 43.0% in Africa.³ A private clinic in a wealthy city may sit far below any of those, and the return is calculated on its own rate.

Research on 1.25 million online sales leads found that firms responding within an hour were more than 60 times as likely to qualify a lead as firms waiting 24 hours or longer.⁵ That shows the direction of the effect in business generally, and the clinic's value is the change in its own conversion rate against its own baseline.

Turning the counts into money

Extra bookings are worth the average fee for the services that converted, less the direct cost of delivering them, where the diary had room. Recovered no-shows are valued the same way, per slot filled.

Staff hours are valued at the clinic's own loaded hourly cost from payroll, when the rota is shorter or named hours moved to work that earns, such as recalls and follow-up bookings. Report other hours as capacity, and say where the time went.

Count the full cost, including the subscription published on the pricing page, staff time on set-up and training, and any overlap with tools being retired. For documentation, the documentation cost calculator works out what records cost today and shows its arithmetic.⁷

Discount the forecast

HM Treasury's supplementary guidance on optimism bias describes a demonstrated, systematic tendency for project appraisers to be overly optimistic, and asks them to make explicit adjustments to costs, benefits and timings.⁴ Its benchmark adjustment for outsourcing projects is up to 41% on operating costs.⁴

Write the forecast down, then halve the benefits and add a month to the timeline before presenting it. Approve the investment if it still pays on those numbers, and otherwise wait for the measured result.

Check before go-live that the new system records the same data as the baseline. wai logs every message, draft, approval and signature, and the log can be exported, so reply time and staff activity are countable without a stopwatch.⁸

Where to start

This week, write down the four measures and the balancing measure with their denominators, and pull the last four weeks from the inbox, the diary and the payment records. Ask the front desk to log time on messages and phones for two sample weeks.

Before go-live, freeze prices and campaigns for the measurement window where you can, and note anything you could not freeze. After go-live, count the same way at day 30, 60 and 90. wai runs Admin as a 90-day pilot that ends with what changed in revenue, hours saved and patients cared for.⁶ Put the measured change beside the full cost, and the board decides on a return it can see.

Frequently asked questions