You have enough patients to learn from and not enough to run an experiment
A single practice in a mid-size metro will never generate the volume that a statistically clean page test requires.
Split testing a button color needs thousands of sessions per variant. One practice with a normal new-patient flow will not produce that in a year, and a test read before it is ready will point you confidently in the wrong direction.
So change the unit of evidence. Grade twenty recorded calls against a simple rubric: was a specific time offered, was the coverage question answered, was a follow-up agreed. Patterns show up in twenty calls that would take twenty thousand sessions to see in a page test.
Then make changes big enough to be visible without statistics. Adding an early-morning block, answering the phone through the lunch hour, or replacing a callback promise with a booked time are changes you can read in a month of booked-appointment counts.
Keep a written log of what changed and when, with one change at a time. Sequential before-and-after is imperfect, but it is honest, and it beats a test that never reaches significance.
A family in New Albany will not cross a bridge for a rough estimate
Coverage and cost uncertainty is a heavier obstacle when the visit also requires crossing a state line.
Households in Floyd County and Clark County generally carry plans organized under Indiana, and a Kentucky practice may or may not be in network with them. A vague answer at intake is where those inquiries die.
Give the front desk a one-page reference: what you are in network for, what you file out of network, what you do not accept, and the state each distinction applies to. Details for any individual patient are worth confirming with the plan directly, and the script should say so plainly rather than guessing.
Where you are out of network, do not hide it and do not improvise. A written estimate with the out-of-network portion stated, sent while the caller is still on the phone, converts better than a confident verbal number that turns out to be wrong.
Track this as its own metric. Booking rate on Indiana-side inquiries versus Jefferson County inquiries will tell you within a month whether the coverage conversation is working or whether you should stop advertising across the river.
After-hours coverage should book the visit, not take a message
In a metro built around shift work, a large share of inquiries arrive when nobody is at the desk.
Overnight and early-morning schedules are ordinary here. The inquiries that come at six in the morning or nine at night are not lower-intent, they are simply arriving from people whose day runs differently.
A message service that only takes a name loses most of them. What works is a path that ends in a specific time: self-scheduling for defined appointment types, or a text reply within minutes that offers two real slots.
Missed calls need the same treatment. An automatic text within seconds, from the number they dialed, offering a time rather than an apology, catches a meaningful share of callers before they open the next tab.
Read the leak before you fix it. Pull the call log for a month and count unanswered calls by hour. The lunch hour is usually worse than the night, and it is easier to fix.
Read the no-show rate by hour, and the seven o'clock block defends itself
Failed appointments are rarely spread evenly, and the average hides where the schedule is actually breaking.
Break kept-appointment rate down by appointment hour, by day of week, and by whether the patient was new. Practices that do this usually find one or two slots doing most of the damage, and they are not always the ones people assume.
Early blocks often perform well in a shift-work town because they suit the patient rather than the practice. Late afternoon blocks in the middle of the commute frequently do not, especially for patients driving out toward Oldham County or back across the river.
Once you can see it, respond to the pattern rather than to the average. Confirm high-risk slots differently, ask for a card on file for the ones that fail most, and put your reliable slots in front of new patients.
Confirmations should carry arrival detail. Where to park, which door, how long the visit runs. For a practice on a narrow street in an older neighborhood, that alone reduces late arrivals and the cancellations that follow them.
The specialist referral you sent out is a booking you never got back
Every outbound referral is an existing patient leaving your schedule with no guaranteed return date.
General practices refer constantly and track it almost never. The patient goes for the extraction, the endodontic treatment or the surgical consult, and whether they come back for the restorative work is left to chance.
Build a referral ledger with a return date on every line. Who was referred, for what, when, and what appointment is scheduled with you afterward. Anything without a return appointment gets a call.
The same logic applies to diagnosed treatment that was never scheduled. It is usually the largest unworked list in the practice and it costs nothing to contact, because those patients already know you.
Work it deliberately rather than in bursts. A set hour each week, with the list sorted by value and by how long it has been sitting, produces more production than most advertising at a fraction of the cost.
For patients with no dental plan, a membership option is an intake tool
A steady flow of relocating and shift-based workers arrives without employer dental coverage, and price is their first question.
When a caller has no plan, the conversation usually ends at the cost of an exam. Nothing else about the practice gets discussed, and the caller keeps dialing down the list.
An in-house membership plan changes the shape of that call. Instead of a number, the front desk has an offer: what it covers, what it costs monthly, and what today's visit would be under it. Terms are worth reviewing with your own counsel before you launch one.
Measure it as a conversion path, not as a discount. Track how many uninsured callers book, how many enroll, and what those patients produce over a year compared with the plan-covered average.
The same script should exist for financing on larger cases. A treatment plan presented without a payment path is a treatment plan that gets declined in the operatory, and case acceptance is a conversion step like any other.
Questions we actually get
- What should we measure first?
- Three numbers, separately: how many inquiries you receive, how many become booked appointments, and how many of those are kept. Most practices report only the first and assume the rest. The gap between booked and kept, broken out by hour and by whether the patient was new, is usually where the largest recoverable loss is sitting.
- How many calls do we need to review to find problems?
- Twenty is generally enough to see the pattern, and doing it monthly matters more than doing a large batch once. Grade against a fixed rubric so you are comparing like with like, and share the results with the team as coaching rather than as a scorecard. The point is to fix the script, not to blame the person reading it.
- Is online booking worth adding for conversion reasons?
- It helps when it writes into the real schedule and offers genuine openings. It hurts when it collects a request that a human then has to call about, because the patient believes they are booked and finds out otherwise. Start with one or two appointment types you can safely release and expand from there.
- Should we run A/B tests on our landing pages?
- Rarely at single-practice volume. You will not accumulate the sessions to read small differences, and half-read tests produce confident wrong answers. Sequential changes, one at a time, judged on booked and kept appointments over a month or a quarter, is the honest method at this scale.
- How long before intake changes show up in production?
- No one can promise a timeline. What is predictable is the order: answered-call rate moves first, booking rate follows, kept appointments after that, and production last because larger treatment is scheduled out. Give each change a full month before judging it, and keep a written log so you know what caused what.
