By Tiago Strammiello & Caleo Tsiapalis, Co-Founders, ClaireAI
In brief
- Define the test window and expected call path before collecting results.
- Keep a call-level record and reconcile every rate with the underlying counts.
- Evaluate the greeting, next step, transfer and record handoff separately.
- The earlier numerical study findings are withheld pending source-record verification.
An after-hours intake audit checks what a caller actually reaches when the office is closed. Use a defined call window, an outcome log and a review of the intended next step. This guide provides a worksheet for a firm to test its own phone route before and after changing coverage.
Research update.Numerical findings previously published at this URL are withheld pending verification of the underlying call records. This edition provides an audit method and does not present a measured industry answer rate or a revenue forecast.
Define the question before placing calls
Write down the number you are testing, the hours being reviewed and the route the firm expects. An answer from a person, a configured voice agent and a voicemail recording are different observations. Decide how to label each before the first call so the results use consistent definitions.
- Scope: the firm numbers and schedules included in the review.
- Window: the dates, local time zone and office-hours boundary.
- Expected result: the greeting and next step for each call path.
- Stop rule: how long the tester will wait and when the call will end.
- Privacy: fictional test details and a plan for handling test records.
Use a call-by-call worksheet
| Field | What to record |
|---|---|
| Call reference | A unique test ID; keep contact details outside a public report. |
| Number and time | The destination, exact timestamp and local time zone. |
| Observed answer | Person, voice agent, voicemail, ringing, disconnect or another observed outcome. |
| Greeting | Whether the greeting identifies the expected firm and coverage. |
| Next step | Message, callback request, consultation path, transfer or no completed next step. |
| Transfer result | Whether the receiving destination answered and whether the fallback worked. |
| Record delivery | Whether the assigned person received the expected intake record. |
| Exception | Unexpected behavior, supporting records and the owner investigating it. |
A proposed audit worksheet, not results from a completed study.
Test the schedule and its exceptions
Include an ordinary evening, a weekend and any configured closure relevant to your firm. If numbers serve different time zones, test their schedules separately. Record which cases were tested; a successful evening call does not establish that every holiday or overflow condition works.
Use the after-hours routing guide to prepare the expected call paths.
Calculate rates from a single set of records
Choose an explicit definition of an answered call. For a live-response rate, state whether it includes voice agents as well as people. Count each eligible test once, then divide the number meeting that definition by all eligible tests. Record exclusions before calculating the rate.
Check that category totals equal the sample size and that all tables use the same denominator. Calculate a combined rate from the underlying counts rather than averaging rounded percentages. Label the result as a description of your test calls; it does not establish a national rate or explain why a call was unanswered.
Separate answering from a completed handoff
An answered call can still stop before a useful next step. Review whether the approved questions were followed, the receiving team got the expected record and any unavailable destination led to the agreed fallback. Track these outcomes separately so an answer-rate change does not hide a routing problem.
Review the warm-transfer test cases for answered and unanswered destinations.
Compare configurations without claiming causation
For a before-and-after review, keep the tested numbers, time windows and scenarios comparable. Note differences in staffing, scripts, routing and phone-provider settings. A small operational test can identify a failure to investigate; it cannot establish that a particular tool caused a revenue increase.
Keep the source worksheet, definitions and calculation version with any results you choose to publish. A reviewer should be able to reproduce every figure from the eligible call records. Withhold any statistic that cannot be reconciled with those records.
Turn an observed problem into a testable change
- No expected greeting: compare forwarding and schedule settings with the intended route.
- Unanswered transfer: review staffed hours, destination ownership and fallback behavior.
- Missing intake record: compare call history with the selected connection and destination.
- Unclear caller response: have the firm approve revised language, then repeat the relevant test.
Investigate a missed call using provider records and the available ClaireAI timeline.
For a ClaireAI evaluation, bring the call paths you want to test to a demo. Standard onboarding is included; the team confirms the supported configuration and launch timing with your firm.
Tiago Strammiello & Caleo Tsiapalis, Co-Founders, ClaireAI. Tiago and Caleo work on ClaireAI’s legal intake workflows. This guide describes a proposed method for reviewing a firm’s after-hours phone route.