Report the sequence separately: offered or delivered → verified live answer → buyer billability → application → issued or placed business → collected and retained revenue. A later stage must have its own evidence.
This is a published measurement framework, not a report of RingFunnel conversion rates or an industry benchmark. No private call records are included. Download the measurement dictionary for your own internal reporting.
Define the cohort before calculating
Record the observation period, time zone, product, source, delivery model, eligible destinations and exclusion rules. Use a private call reference to reconcile records. A cohort of calls delivered this week should keep those same calls when later outcomes are reviewed; do not divide a later week's sales by unrelated calls.
If retries or transfers create multiple log rows, decide whether the question concerns routing attempts or unique call opportunities. Preserve both counts where useful. Disclose the matching and duplicate rules so another analyst can reproduce the denominator.
Minimum definitions for a call report
| Measure | Evidence required | Limit |
|---|---|---|
| Delivered | Authoritative routing record for the defined destination | Not proof of a live answer |
| Live answered | Verified human-answer evidence under a documented review method | Duration alone cannot distinguish voicemail |
| Buyer billable | Applicable buyer agreement and confirmed billing classification | Not publisher payability or a sale |
| Publisher payable | Applicable publisher agreement and payout classification | Separate obligation from buyer charge |
| Application / issued / placed | Authorized agency or carrier records with stage and date | Stage definitions and later reversals matter |
| Collected commission | Confirmed financial record linked to the cohort | May change after reversals or chargebacks |
Keep unknown separate from no
Use explicit true, false and unknown states for outcomes. An unavailable recording is not a confirmed unanswered call. An integration that returns no sales field is not evidence of zero sales. Show the number of classified and unclassified calls alongside the reported rate.
For example, if 24 of 30 delivered calls have verified live answers and six are unclassified, the verified-answer share is 24/30, or 80%. Do not call the other six missed calls without evidence. This example is hypothetical and does not describe RingFunnel results.
Use named denominators
- Verified-answer share: verified live answers / all delivered calls in the cohort, with unclassified count reported separately.
- Application rate per live answer: verified applications / verified live answers for the same cohort. State whether outcome coverage is complete.
- Net call spend: confirmed call charges less confirmed credits attributed to that cohort.
- Cost per placed policy: attributable cohort cost / verified placed policies. Identify whether labor and other variable costs are included.
- Observed contribution: collected commission less reversals, net call spend and documented attributable variable costs. Fixed overhead remains separate unless included explicitly.
A zero denominator makes a ratio unavailable, not zero. Missing costs prevent a complete contribution calculation. Preserve both the as-of date and the original cohort dates when values change.
What a publishable benchmark needs
Before publishing a performance benchmark, document the sample size, sampling frame, date range, source mix, vertical, definitions, missingness, exclusions and time allowed for outcomes to mature. Explain whether the data represents one buyer, a selected set or a broader population. Avoid extrapolating a small or selected sample to the industry.
Use aggregated, permission-cleared data. Suppress small groups that could expose a person or business, and remove consumer details, recordings, private commercial terms and identifying references. A methodology page is not evidence that such a dataset has been collected or approved for release.
Version and correct the result
Keep the original extraction, calculation version, reviewer and revision history privately. A public result should explain material corrections and link to the method used. This framework is version 1, published September 23, 2026. Questions or corrections can be sent through the RingFunnel contact page.
For a working scorecard, see how to measure a call test. For consistent buying definitions, use the glossary and buying specification.