Koda: two agencies built the wrong fix. The data showed the real one.
- Industry
- plastic surgery clinic
- What we did
- analyzed the full patient journey from first inquiry to surgery before building anything
- Result
- the real bottleneck found, $1.5M in recoverable revenue identified, and a third wasted automation spend avoided
The problem
Koda had already paid two separate agencies a combined $14,000 to build AI voice agents. Conversion was still stuck at 4.6%: 138 surgeries from 3,808 leads.
Everyone assumed the problem was after-hours response time. So both agencies built after-hours voice agents. Twice. Nothing moved.
What we found
Before writing a line of code, we went through the data on every lead.
- 64% of the leads they won or still had open (1,460 of 2,284) arrived after hours
- 63% of the leads they lost (958 of 1,524) also arrived after hours
Almost identical. After-hours was never the problem.
The real constraint was two coordinators who physically couldn't handle the inbound call volume during the day. The data also showed:
- a 42% no-show rate on booked consultations
- a 75% drop-off between consultation and booking surgery
- lead qualification that was fundamentally broken
What it was worth
Fixing those leaks added up to about $1.5M in recoverable revenue, with the no-show problem alone fixable through a proper reminder and confirmation flow.
We then built a prototype operations platform around the real bottlenecks: lead pipeline management, pre-op coordination and four AI agents to take the load off the coordinators.
Why it matters
Two agencies automated a problem that didn't exist, because nobody looked at the data first. That's why every engagement we run starts with the numbers, not the tools.
Could this be happening in your business?
If you've already spent money on automation that didn't move the numbers, the leak is probably somewhere you haven't looked yet. See how Agent OS finds it, or book a call.
