Most practices I sit down with already have the data. What they don’t have is the habit of looking at it on a cycle short enough to act on — and that, not the reporting software, is what costs them.
Here is the fork I care about most. If our schedule fill rate — booked slots divided by available slots — drops below 75% and we are not in a planned slow season, we have one of two problems: a demand problem or a scheduling friction problem. The distinction matters, because the fixes are not the same.
Demand problem or friction problem
A demand problem means fewer people are asking for us. A friction problem means the same number are asking and fewer of them end up in a slot. If inquiry volume held steady while bookings fell, that is friction — the calls came in and something in our own process ate them. If inquiries and bookings fell together, that is demand. The two look identical on a fill-rate chart, which is why the chart alone isn’t enough.
They also want opposite money. Demand is answered with marketing, referral relationships, or reactivating patients we already have. Friction is answered with phone coverage, callback speed, how far out the first genuinely available appointment sits, and whether someone can book without speaking to a human at all. Spend marketing budget on a friction problem and we buy more inquiries that leak out of the same hole — we pay twice and the fill rate doesn’t move. That mistake usually traces back to a monthly review that surfaced the symptom without the diagnostic.
Why weekly, not monthly
The case for a weekly cadence is straightforward. Staffing, marketing spend, and cash flow can all be adjusted while a trend is still small. Reviewed monthly, the same signals arrive after the month they describe has already been paid for.
That is not a figure of speech. Payroll has run, the ad budget is spent, and the empty slots from three weeks ago cannot be resold — unused clinical capacity is the most perishable inventory in the business. A four-week soft patch is a single data point in a monthly review. Weekly, it is four chances to intervene, and the first one is cheap.
The honest limit is that weekly numbers are noisier. A holiday, a provider out sick, one bad-weather day — any of those will swing a week. That is why we plot week by week rather than judge a single week: we want direction, not a reaction to one bar on a chart.
The five dashboards we’d actually run
None of these are proprietary. They map onto the four things a practice can actually adjust — the schedule, staffing, marketing spend, cash — plus the one that explains most of what the other four are saying.
The demand and schedule dashboard
Three numbers: new patient bookings broken out by source, schedule fill rate, and no-show and cancellation rate. Fill rate below 75% outside a planned slow season is the fork above. Spikes above 8–10% on no-shows and cancellations are the second flag — and the spike matters more than the level, because a sudden jump usually points at something we changed, like a new reminder flow or a longer wait to be seen.
The by-source breakout is not decoration. A flat-looking aggregate is often one referral source quietly collapsing while another covers for it, and by the time the total moves, the collapse is months old. The question: are we selling the capacity we already pay for?
The friction dashboard
Everything between a person deciding to contact us and a name landing in a slot. Calls that rang out or went to voicemail, how long before someone called back, inquiries that never became appointments, and how far out the first genuinely available slot sits. That last number is the one most practices don’t track and most patients decide on.
The question: of the people who wanted us this week, how many did we lose to our own process? This is the dashboard that settles the fork, which is why I would build it second, not fifth.
The staffing and capacity dashboard
Hours rostered against hours of booked demand, by provider and by day of week. Fill rate is an average, and averages hide shape — a comfortable 78% week can be a Monday at capacity and a Thursday half empty — a staffing problem wearing a healthy number as a disguise. The question: are we paying for coverage on the days without patients and short on the days with them?
The marketing dashboard
Spend by channel, placed beside booked new patients by source from the first dashboard. Divide one by the other and we have cost per booked new patient — computed from our own two numbers, not borrowed from anyone. Booked, not inquired; the gap between those is the whole point.
One caution worth stating: attribution is messy, and plenty of new patients won’t trace cleanly to a channel. Record what we can, mark the rest unattributed, and read the trend rather than the decimal.
The cash dashboard
Cash collected this week against what was billed, what is still sitting unbilled, and outstanding balances by age. In turnaround work the first question is always where the cash is leaking; this is that question asked weekly. Revenue earned is not cash received, and that gap is where busy practices get blindsided.
The threshold question, answered honestly
Two numbers here carry a real threshold: 75% on fill rate, and a spike above 8–10% on no-shows. For the other dashboards I am not going to hand over an industry benchmark, and I would be suspicious of anyone who does without knowing your specialty, payer mix, appointment lengths, and market.
The right threshold for those is our own recent baseline. Run each one for a couple of months, note the normal range, and let the alert be a departure from that range rather than from someone else’s average. A borrowed number is worse than no number, because it manages to be reassuring and wrong at once.
The last piece has nothing to do with numbers. Each of the five needs a name attached — the person expected to have looked, and to say what they are doing about it. A dashboard nobody owns is a screensaver.
Where this work actually lives
Building this and keeping it running — the data plumbing, the weekly rhythm, and the agents that handle the intake and scheduling work these dashboards point at — is delivered through Interactive Intel, Alton Worldwide’s agentic-AI practice: small, Miami-based, every engagement led by me personally and scoped around an outcome you can measure. How we think about the sector sits on our healthcare page.
If you want a smaller starting point than a reporting programme, the fixed-price AI Opportunity Scan is it: one workflow, two weeks, the payback math in writing before you commit to anything larger. And if you only ever build one of the five, build the first — fill rate, plotted week by week.
Related reading: AI for medical practices · AI consulting for small business