Nine numbers.
One definition each.
These are the numbers on your dashboard, in your portal and in your monthly review. Each one has a single definition, written down here, so nobody argues about what it means.
What it looks like
Tap a number to see how it is measured. Every value comes from the sample inputs below; change them and the dashboard follows.
Hours returned Time
Person minutes for every task agents completed, added up, ÷ 60
Each agent run records the task it replaced. The minutes per task are set with you in the deep audit and checked during the pilot.
Change the sample inputs
Sample values to show how each number reads. Not a client, not a result.
How each one is measured
The same words and formulas appear in your portal and your monthly report.
Hours returned
Person minutes for every task agents completed, added up, ÷ 60
Each agent run records the task it replaced. The minutes per task are set with you in the deep audit and checked during the pilot.
The capacity agents hand back. It turns into money only when that time goes to selling, serving or building.
Cost per hour returned
(AI usage + Humonix plan) ÷ hours returned
AI usage is metered on every run: tokens and voice minutes, at cost. The plan is your monthly fee.
Compare it with what an hour of a person costs you. Same definition as the token burn page.
Approvals turnaround
Median time from a draft entering your queue to your decision
The approval queue stamps each draft and each decision. Next to it: the share of drafts approved without edits.
Slow approvals hold work back. A low share approved without edits means the Brain is missing knowledge; fix that before adding more agents.
Site score
Score out of 100 from the site check
The same check as the free audit: titles, descriptions, structured data, speed basics and more, run on a schedule.
How well search and AI engines can read your site. Fix what fails first, then what needs work.
AI-answer visibility
Questions where you are cited ÷ questions tracked
A fixed list of the questions your customers ask, built from your Brain and checked in ChatGPT, Perplexity, Gemini and Google AI.
Follow the trend, and which competitors are cited where you are not. The list stays fixed so months compare.
Leads from site
Form fills, chats and booked calls from your website, counted once per person
Lead intake on your site. Duplicates are merged by email or phone; leads from an ad click are counted under ads.
The demand your site creates on its own.
Cost per lead
Ad spend ÷ leads from ads, per channel. All in: (ad spend + Humonix plan) ÷ all leads
Spend from the ad platforms; leads from your intake, matched to the click that brought them.
The first line judges the ads. The second shows what a lead costs you in full.
Ad spend vs results
Spend this month against the cap you set, next to the leads it brought
Spend read from the ad platforms each day. Spend stops at your cap; budget changes wait for your approval.
Spend should follow results. When leads stall while spend climbs, we propose a change and you decide.
Response time
Median time from a new lead or customer message to the first reply
Timestamps on each message and each reply, by an agent or a person, counted around the clock.
New leads cool quickly. This shows whether someone answers in time.
How we review them with you
Same numbers, same order
Time first, then visibility, then demand. One page, the same each month.
Your inputs, written down
Minutes per task and the hourly cost are set with you and kept on the page, so you can change them.
One decision per number
Each review ends with what to change: a rule, a budget, a skill or nothing.
Set up your own numbers
We agree on the inputs with you before anything runs, so the first month already reads true.
Book a call