"Let Me Talk to My Husband": Recall Automation vs. an AI Sales Agent for Ortho Cases That Never Start

A mother and her twelve-year-old sit through the entire consult. Scan, photos, the 22-month plan, the retainer talk, the number: $5,400 with $1,200 down. Then she says the thing almost every parent says. "Let me talk to my husband tonight."

Your treatment coordinator marks the case pending and moves to the next chair. Eleven days later your software sends the family a text: "Hi Jennifer! Just checking in, are you ready to get started?" No reply. The case sits in the pending column until someone archives it at the end of the quarter. Nobody in your practice ever learns what her husband actually said.

The leak: pending cases get treated like appointments nobody booked yet

Every orthodontic practice has a pending list. Almost none of them work it like a pipeline.

The consult already happened, which means the expensive part is done. Chair time, the scan, the coordinator's hour, the marketing dollars that got the family through the door. What is left is a decision being made in someone's kitchen, without you in the room, usually against one or two other practices the family also visited.

That is a race, and races are won on responsiveness. Roughly 78% of sales go to whoever responds first, and average response time in B2B sales sits somewhere around 42 hours. An eleven-day "just checking in" text is not a response. It is a reminder that you exist, sent to someone who already knows you exist and is waiting for a reason.

Why it happens: your software models chairs, not decisions

Practice-management systems are built around the appointment. Every object in them has a date, a chair, and a provider. A pending case has none of those. It is a conversation in an unfinished state, so the system stores it as a flag and fires time-based recall messages at it.

Time-based is the problem. The system knows eleven days have passed. It does not know that Jennifer's real objection was the $1,200 down payment, not the treatment plan. It sends the same message to the family that needs a payment structure, the family waiting on a second opinion, and the family that is moving in June.

Meanwhile your treatment coordinator, the one person who could tell those three families apart, is chairside most of the day and does follow-up in the gaps. Not because she is disorganized. Because there are no gaps.

The framework: work pending cases like a pipeline

Five steps. None of them require buying anything.

1. Record the objection, not the status

"Pending" is not information. Before the family leaves the room, your coordinator writes one sentence in the chart: the actual reason this case did not start today, in the parent's words. Money. Spouse. Second opinion. Timing. Insurance verification. Kid is not ready.

If you do nothing else on this list, do this one. You cannot follow up on a status. You can follow up on a reason.

2. Sort the list by objection, not by date

Most practices work pending cases oldest-first. That is backwards. It puts the coldest cases at the top and buries the family who consulted Thursday and is deciding this weekend.

Group the list into the four or five objection buckets you actually see. You will find the buckets are wildly unequal, and one of them, usually money, is carrying half your unstarted revenue.

That imbalance is the useful part. If forty percent of your unstarted cases stalled on the down payment, you do not have a follow-up problem, you have a payment-options problem that follow-up cannot solve. Sorting by objection tells you which of those two you are actually looking at before you spend a quarter texting people.

3. Match the follow-up to the objection

Each bucket gets a different next action, written once and reused.

Money: send the two or three payment structures in writing, with the monthly number in the first line. Not "we have financing options available."

Spouse or second decision-maker: offer a ten-minute call with both parents on the phone, and give two specific times.

Second opinion: ask directly what the other practice quoted and what the family liked about it. Most parents will just tell you.

Timing: book a dated re-consult on the calendar right now, six or eight weeks out, rather than letting them call you when they are ready.

Insurance: finish the verification yourself and send the result. Do not hand a parent homework.

4. Put a clock on it, not a drip

Set response windows instead of a send schedule. An inbound question from a pending family gets an answer inside the hour during business time. The first post-consult touch happens the same day, not on day eleven. After that, three touches over ten days, each one carrying new information, the payment structure, the appointment slot, the insurance answer, and then you stop.

Three useful touches in ten days beats nine cheerful ones over two months, and it protects the relationship when the answer turns out to be no.

5. Every case ends in yes, no, or a date

A pending case gets three allowed endings: started, declined, or scheduled for a specific future re-consult. "Still pending" is not an ending. Set a rule that nothing stays in the column past 45 days without one of the three, and review the count in your Monday huddle.

The number you are managing here is not case acceptance. It is how many cases resolved at all.

Where recall automation is still the right tool

None of this makes your practice-management recall useless. It is very good at the jobs it was built for: retainer checks, observation recalls, appointment reminders, aligners are ready. Those are time-based events with one correct message, and automating them beats a person doing it by hand every time.

Recall automation fails on pending cases specifically because a pending case is not a time-based event. It is an unresolved objection. And if your pending list is genuinely short, say a single-doctor practice running six consults a week where the coordinator has real time to work every one by hand, a spreadsheet and a calendar reminder will beat any software you could buy. The tool only starts to matter when volume exceeds the attention available to it.

The proof

Tykon's framework benchmark, measured across our testing: above a 32% win rate, achieved with roughly 27% of the work compared to a human sales team.

The second number is the interesting one. Near-parity on outcome is worth noting, but reaching it on about a quarter of the labor is what changes staffing math for a practice. It means pending-case follow-up stops competing with chairside time for your coordinator's attention. The work happens whether or not anyone has a free twenty minutes on a Thursday afternoon.

What this looks like when it runs itself

Tykon's system, James, runs the pipeline above without a coordinator holding it together. It answers new consult inquiries in under 60 to 90 seconds around the clock, captures the objection out of the conversation itself, and works each pending family on the track matching their actual reason for waiting. Payment structures to the money bucket, a two-option call invite to the spouse bucket, a dated re-consult to the timing bucket. Once the case starts, the same system asks for the review and prompts the referral.

Before you evaluate any of that, do step one by hand. Pull every pending case from the last 60 days and count how many have a recorded objection in the chart. If it is under half, you do not have a follow-up problem yet. You have a capture problem, and that one is free to fix.