A clearer picture of
your phone operation.

Every home-service operation between roughly $2M and $15M in revenue runs its front office on the same assumption: that the phones are basically fine, because the ones that ring get answered and the ones that book get booked. The systems these businesses already own reinforce it. They count events accurately and price none of them.

What DialWorth measures

Three components, computed from a client's own call records rather than from a benchmark or an industry constant. Capacity, the calls that were never handled because staffing sat in the wrong place on the utilization curve, modelled at half-hour resolution with queueing mathematics that accounts for callers hanging up. Conversion: calls answered but not booked, decomposed by pipeline stage and normalized so a rep fielding emergency calls does not look skilled for reasons unrelated to skill. Turnover: the fully loaded cost of CSR churn, including the ramp-period booking shortfall nobody invoices and therefore nobody counts.

The three overlap, so they are reconciled rather than added. A call lost to abandonment cannot also be counted as a conversion loss, and the engine enforces that structurally rather than by review.

The commitments the method holds to

These are the constraints worth knowing about, because each one costs us a number we could otherwise have printed.

• No point estimate. Every headline figure ships as a range with a most-likely value. A single number implies a precision the underlying data does not support, and it is the number that gets quoted.
• No causal claim. This is observational data. We measure association and model counterfactuals under stated assumptions; we do not say one thing caused another, and the report builder refuses to render language that does.
• No unsourced parameter. Every assumption carries whether it was measured from your data, derived from it, stated by you, or defaulted, and the defaults are counted on a limitations page rather than buried.
• No unearned proof. There are no testimonials on this site and no cross-client percentiles, because there is not yet a consented dataset large enough to publish one honestly. When there is, it will be stated with its sample size.
• No result the data cannot support. Seven data-quality gates run before anything is computed. If one fails you receive a readiness report instead of a figure, at no charge, and no partial result is published.

Why the method is published

The calculation method is public, down to the thresholds. Every threshold quoted there is asserted against the engine's own configuration by an automated test, so the published method cannot quietly drift from the code that runs. A methodology that has gone stale is worse than none: it is a specific, confident, wrong claim, and the first person to check it is the one deciding whether to trust anything else on the page.

That is the standard this practice is willing to be judged against. It is a deliberately narrow business, one measurement, done properly, for one kind of operator, and the narrowness is the point.