Measurement

Measurement

The calculation model, in full

An ROI claim without a visible calculation model is worthless. Amoris has no results yet — which makes publishing the method the honest thing to show instead.

What claim are we trying to earn?

Not this:

“500 people replied.”

This:

“This work produced X incremental qualified opportunities compared with an untouched control group, from leads you had already paid to acquire.”

The second requires two things that cannot be created after the fact: a baseline capturedbefore anything runs, and a control group held back.

Which metrics actually count?

Three tiers, reported separately and never blended — because blending is how activity gets sold as outcome. An agency reporting only the first two is reporting effort.

TierExamplesWhat it proves
ActivityLeads analysed, contacted, delivered, bounced, cleanedThe work happened
EngagementClicks, replies, positive replies, opt-outsThe cohort is alive
CommercialQualified conversations, appointments, opportunities, conversions, revenueThe only tier that matters

How is incrementality calculated?

A randomly selected slice of the eligible cohort receives nothing — same characteristics, same period, no contact. It is withheld at the suppression gate before the first send, so it costs nothing to run.

incremental conversions
  = treated conversions − (control conversion rate × treated volume)

Worked example

Eligible cohort              10,000 leads
Held back (control)           1,000 leads
Treated                       9,000 leads

Control conversions              12   → control rate 1.2%
Treated conversions             198   → treated rate 2.2%

Expected without intervention  9,000 × 1.2%  = 108
Incremental conversions        198 − 108     =  90

The headline is 90, not 198. The 108 would have happened anyway.

These figures are illustrative arithmetic to show the method. They are not Amoris results, and no Amoris results exist yet.

Why hold a group back at all?

Some dormant leads convert on their own. Without a control, every one of those gets credited to the intervention. That is the first objection a sceptical finance person makes, and they are right to make it.

If a holdout is refused

Some clients will refuse, because deliberately not contacting leads feels wrong. The fallback is the same cohort's conversion over an equivalent prior period.

This is weaker, and the weakness goes in the report rather than being glossed. It cannot separate the intervention from seasonality or from other marketing running in parallel.

When is attribution agreed?

Before a pilot starts, in writing, while nobody has money at stake. Under any revenue-linked arrangement this is the invoice.

Agreed up frontWhy it matters
Attribution window — how long after contact does a conversion count?Long windows over-credit; short ones under-credit long cycles
What counts as recovered — first purchase, or lifetime value?An order-of-magnitude difference
Multi-touch — your own marketing keeps runningBoth sides can claim the same conversion
Who measures and reports it?Whoever holds the CRM holds the number
How disputes resolveCheap to agree now, expensive later

What can be proven before any conversion lands?

Conversions take weeks. These are available in week one, and they are how the system proves itself before revenue arrives.

  • Agreement rate on rejections

    Week 1

    Of the leads the system declined to contact, how many do you agree with?

  • Classification agreement

    Week 1

    Do you agree with why the system thinks each lead died?

  • Trigger precision

    Week 1

    Every strong trigger must cite a real, verifiable change

  • Evidence traceability

    Week 1

    Every factual claim traces to a source in your data

  • Rep overturn rate

    Week 2

    How often the person who takes the call disagrees

The first is the most useful. If a review of 50 declined leads produces 45 agreements, the judgement layer is calibrated — proven without a single conversion.

What is never claimed

  • — No ROI, revenue or conversion figure before it is measured against a baseline
  • — No results from one engagement presented as a forecast for another
  • — No client name without written permission
  • — No blending of activity metrics into commercial ones
  • — Where the weaker pre-period baseline was used instead of a holdout, the report says so

No engagement has completed yet

Every number on this page is either illustrative arithmetic, clearly marked, or a market statistic with a named source. This page exists so that when there are results, you already know exactly how they were calculated — and so you can decide right now whether that method is one you would trust.