Back to BlogTLDR: Log every callback for a month with the office, the field, and the stage it was caught at. The list will concentrate: a small number of offices and a small number of fields produce most of the calls. Fix the form for those fields and have the conversation with those offices.
Why treat callbacks as a metric rather than a nuisance?
Because a callback is the visible end of a process failure, and it is cheap to count. A lab that logs callbacks for four weeks knows more about its intake quality than one that has argued about it for a year, and the log points at specific fixes rather than general complaints.
Callbacks also cost more than the minutes they consume. Each one puts a lab coordinator into a slightly adversarial conversation with a customer, and relationships wear down under that. Offices that get called often start to feel the lab is difficult, whatever the merits.
There is a second, less obvious cost. A case that generates a callback has stopped, and stopped cases are the ones that miss dates. The callback is not only a symptom of poor intake data. It is a direct cause of late delivery.
What exactly should you log?
Four fields per callback: which office, which piece of information was missing or unclear, what stage of the workflow it was caught at, and how long the case sat before the answer came back. Four fields is small enough that people actually record it.
Keep it to a shared sheet with one row per call. Resist the urge to build anything, and resist the urge to add fields. A log with four columns gets filled in; a log with twelve does not, and an empty log tells you nothing at all.
After four weeks, three questions answer themselves:
- Which offices appear most often, adjusted for how much work they send?
- Which fields appear most often, across all offices?
- How many were caught at the bench that could have been caught on arrival?
The three fixes, in the order they pay
1. Fix the form for the fields that repeat. If the same two fields generate half your calls across many offices, that is a form problem, not an office problem. Change the prompt, explain the consequence next to the field, and make blocking fields genuinely required.
2. Move detection to arrival. Most of the cost of a callback comes from how late it is. A case checked on the day it lands generates a call the office can answer from memory. The same case checked two days later generates a chart pull. This is the single change with the largest effect, and it does not require the offices to do anything differently.
On the platform side this is what prescription intake is for. SmartRX reads the prescription on upload, cross-references it against the scan files and surfaces gaps before design starts. SmartScan flags scan-quality issues, including distortions, missing mesh, contamination artifacts and unclear margins, for a technician to judge. Both raise the flag; people decide what it means.
3. Have the conversation with the outliers. After the first two fixes, whatever remains is concentrated in a handful of offices. Go and see them, bring the log, and frame it as turnaround rather than blame. "Your cases average two days longer than our other accounts and here is why" is a conversation about their patients. "Your prescriptions are incomplete" is a conversation about their competence, and it goes badly.
How much improvement is realistic?
Enough that it is worth doing, and not enough to reach zero. Some calls are legitimate clinical conversations that should happen. The target is removing the avoidable ones: the repeated blank fields and the gaps found late that could have been found on day one.
Be sceptical of anyone promising to remove callbacks entirely, including software vendors. A lab that never calls a referring office is either getting perfect prescriptions, which does not happen, or guessing, which is worse than calling.
The honest goal is a shift in the mix. Fewer calls overall, and the ones that remain happening on day one about something that genuinely needed a clinician's judgement, rather than on day three about a shade guide.
A four-week plan
- Week 1: start the log. Four columns. Tell the team why.
- Week 2: keep logging. Do not change anything yet, or you will not know what worked.
- Week 3: keep logging. Start drafting form changes for the top two fields.
- Week 4: keep logging, then sort. Fields by frequency, offices by frequency per case sent.
- Week 5: ship the form change and move the check to arrival.
- Week 9: re-run the log for two weeks and compare. Same columns, same definitions, or the comparison is meaningless.
Related reading
- the denture prescription checklist every lab should send out
- the remake starts at intake, not at the bench
- why faster design alone will not fix your turnaround time
Frequently asked questions
What should we record for each callback?
The office, the specific information that was missing or unclear, the workflow stage where it was caught, and how long the case waited for an answer. Four fields, one row per call.
How long should we log before acting?
Four weeks. Long enough for the pattern to be stable and short enough that the team stays engaged. Change nothing during the logging period so the before and after comparison means something.
Should we charge offices for incomplete prescriptions?
It is rarely worth it. The calls concentrate in a small number of fields and offices, and a form change plus one conversation usually fixes more than a fee does while keeping the account.
Can callbacks be reduced to zero?
No, and aiming for zero pushes teams to guess rather than ask. The goal is to remove avoidable calls and shift the remaining ones to day one, when the office can still answer easily.