Demo, Runs in this browser only, Fictional data
Test one headcount request against the work
Enter the workload behind one request and what a short trial showed. See the capacity arithmetic, which numbers are evidence and which are estimates, and whether a bounded trial is justified yet.
Arithmetic on the numbers you enter, in this browser. It never recommends hiring, firing or changing anyone’s role, and the decision stays with the person you name.
01 The request
A support request with a counted workload and a ten-draft trial. Invented figures and roles. Never enter employee names, pay or performance details.
Turn on JavaScript to edit this sample and run the check yourself. The result below was computed from this sample by the same rules when the page was built.
- Workflow
- Support: how-to tickets
- Weekly volume (items a week)
- 180
- Volume is
- Counted
- Time per item today (minutes)
- 9
- Time is
- Timed
- Team time for this work (hours a week)
- 24
- Needs judgment or a relationship
- Rarely
- Drafts sampled (drafts)
- 10
- Usable drafts (drafts)
- 7
- Time to check a draft (minutes)
- 1.5
- Check time is
- Timed
- Who checks every draft
- Support lead
- Who approves a trial
- CEO
- Fewest drafts that count (drafts)
- 10
- Usable drafts needed (%)
- 70
- Trial length (weeks)
- 6
02 The result
TRIAL JUSTIFIED The evidence supports proposing a 6-week trial
The evidence supports proposing a 6-week trial. The work takes 27 h a week today, 3 h more than the team’s time. The decision stays with CEO.
- Work today
- 27 h a week
- Team time
- 24 h a week
- Trial rate
- 70% (7 of 10)
- With drafts
- 12.6 h a week
Capacity arithmetic: Support: how-to tickets
| Line | Working | Result | Basis |
|---|---|---|---|
| Work today | 180 items × 9 min ÷ 60 | 27 h a week | Evidence |
| Against team time | 27 h − 24 h available | 3 h a week over | Evidence |
| Trial rate | 7 usable ÷ 10 sampled | 70% | Evidence |
| With drafts | 180 × (1.5 min check + 30% redone × 9 min) ÷ 60 | 12.6 h a week | Projection, not a measurement |
Proposed trial
- Drafts for every item, each checked by Support lead before anything goes out.
- A weekly quality check: stop the trial in any week where usable drafts fall below 70%.
- CEO approves before it starts and decides what happens after it.
- If the trial rate holds, the work needs about 12.6 h a week against 24 h available.
Keep in mind
- This check does not recommend hiring, firing or changing anyone’s role, and it is not a forecast of savings.
Each download is labelled as a demo export of fictional data checked in a browser, with the rules version and an input fingerprint. Entries that start like a spreadsheet formula are saved as plain text.
03 The rules
This is the whole check, with your current settings filled in. Nothing else is weighed, and the same entries always give the same result.
- Work today = weekly volume × minutes per item ÷ 60. With drafts = volume × (check minutes + share of drafts redone × minutes per item) ÷ 60.
- A figure marked Estimated keeps that label in every result built from it.
- A trial result counts only when at least 10 drafts were sampled and checked.
- A trial is justified only when at least 70% of sampled drafts were usable, a named person checks every draft, and a named person approves.
- Work that often needs judgment or a relationship stays with a person: no trial is proposed.
- Missing or estimated inputs produce a bounded evidence step, not a decision.
- A justified trial runs 6 weeks, with a weekly quality check, and stops in any week below 70%.
- The check never recommends hiring, firing or changing anyone’s role.
04 What this shows
What it proves
- Every figure shows its working, and each is labelled as evidence, an estimate or a projection.
- Where the inputs cannot support a decision, the check says so and names a bounded way to gather what is missing.
- A named person approves. The check cannot.
What it does not
- It is not a forecast, a measured saving or proof that a trial will work.
- It is not HR or legal advice, and it does not judge anyone’s performance.
- A real review uses your own tasks, data and tools.
A demonstration of the method on fictional data, run entirely in this browser tab. It is not an AI model, not a customer implementation, not proof of delivered savings, and not a substitute for your own procurement or security review.
Want a check like this built on your own work?
Is this a fit? → Ask about a Consult block See the worked example
Consult: $2,000 for 5 hours. Build: $45,000 minimum over 3 months, from $15,000 a month. The fit guide explains which, if either, suits your work.