Most companies that claim fast response times don't explain how they measured them. We're going to.
This post covers the measurement rig we built, the measurement run we completed, what the clocks actually measure, and — importantly — what we refused to publish and why. If you're skeptical of speed claims in general, this is for you.
We didn't test callback speed in a lab. We ran the measurement through our live production system — the same path a real missed call takes.
A missed call enters our system via webhook, gets logged to our database, triggers a callback dispatch, and the Vapi voice platform dials the lead. We instrumented every handoff.
The measurement clock starts at missedat: the moment our system ingests the missed-call event. The measurement clock ends at ringingat: the moment Vapi reports that the outbound call has been initiated.
We ran this across every logged call in the measurement run — verified against our D1 database — all tagged as measurement runs so they don't contaminate production data. Every call went through the live production code path. The results were HMAC-signed at ingestion.
Across the measurement run, the average dispatch time — missed_at to call origination — was approximately 1.6 seconds. The 95th percentile was approximately 2.3 seconds.
Those are fast numbers. Fast enough that we almost published them.
We didn't publish "1.6 seconds average." Here's why.
When we dug into the Vapi ringing_at timestamp, we found something important: it fires at call origination — the moment Vapi instructs Twilio to dial. It does not fire when the destination handset physically starts ringing.
We verified this directly. We compared ringingat against Vapi's createdAt and against the Twilio leg's datecreated. They matched to the millisecond across six separate calls. Vapi emits ringing_at at creation, not at the moment audible ringing begins.
We then measured how long it actually takes for a handset to physically start ringing after Twilio initiates the call. Using direct calls to an external handset with a separate timer: the median was 2.45 seconds. Minimum 2.0 seconds. Maximum 2.96 seconds across our test calls.
The math is simple: if dispatch to origination averages 1.6 seconds, and origination to audible ring adds another 2.4 seconds, publishing "1.6 seconds" as a "phone ringing" speed would be wrong by a factor of more than two.
We refused to publish it.
Missed call to your phone ringing: under 8 seconds. Machine-verified. July 2026.
This is an upper bound, not an average. Here's how we derived it:
missed_at to call origination span at p95 across the full measurement run.Summing two independent p95 figures overstates the joint 95th percentile — the real number is almost certainly lower. We chose to be conservative rather than aggressive. The bound holds even if your carrier is slower than our test handset.
The speed-to-lead research is clear: companies that respond to missed calls within minutes convert at dramatically higher rates than those that respond within hours. The industry standard is measured in hours, not seconds.
We care about that advantage being real, not just claimed. A number you can't defend in a conversation with a skeptical customer is a number that erodes trust. We built the measurement rig so the claim is defensible — every timestamp in our system, every call in the run, every derivation documented.
If you want to see the methodology in detail, the full runbook and raw call log are maintained internally. We don't publish them publicly, but we don't hide them either.
Measurement methodology: STAT-REGISTRY v2026-07-07. Claim source: STAT-001-BOUND. Call count verified: D1 database, 2026-07-07T03:34 UTC. All measurement calls tagged source=measurement and excluded from production reporting.
You'd never install a furnace without testing it — but most HVAC owners have never test-called their own after-hours line. Here are four scripted scenarios, what a pass looks like for each, and a simple scorecard you can run this week.
We're a founding cohort. There's no public price because a price before a conversation locks you into a number before we understand your business. Here's why that's actually better for you.
When you miss a call, we don't call back in three seconds. There's a short delay built in — and it's there for a reason. Here's what the holdout is, why it exists, and why it makes callbacks work better.
Call (866) 401-5020 — any hour — and Jacqueline answers: the same engine your customers will hear. Ask anything about the cohort; she'll book your founder call on the spot.
Apply for a founding spot →