Skip to main content

Calculator

What does your marking actually cost?

Four numbers you already know, and arithmetic you can check. We do not assume review time goes to zero, because it does not.

103,000
2 min120 min
1 min30 min

This is not zero and we have not set it to zero. The default of 3 minutes assumes a reviewer reads the drafted evaluation, checks the cited evidence and releases it. Raise it if your work is long-form or high-stakes.

$20$300
Assessor hours today

40 hrs

per month

Review hours with Scorafy

6 hrs

per month

Net saving after the subscription

USD 1,941

per month

Assessor hours freed

34 hrs

per month

The arithmetic

120 x 20 min ÷ 60 x USD 60
USD 2,400
less 120 x 3 min ÷ 60 x USD 60 review
USD 360
less Growth plan subscription
USD 99
Net per month
USD 1,941
Net per year (x12)
USD 23,292

Growth covers 500 AI evaluations a month. Annual billing reduces the subscription line further and is not applied here.

What this model assumes

So you can argue with it properly.

  • Review time is real and it is yours to set

    The workflow is AI evaluation plus human review, not AI instead of humans. The default of 3 minutes per submission is a starting point, not a measurement of your team - move it until it matches what a reviewer would actually spend reading a drafted evaluation and checking the evidence behind the scores.

  • The subscription line is the list price for your volume

    The calculator picks the smallest self-serve plan whose monthly AI-evaluation allowance covers your submissions, and subtracts it at the monthly price. Annual billing is cheaper and is deliberately not applied, so the figure is conservative.

  • Nothing here is a customer average

    There are no benchmark figures, industry averages or typical-saving percentages on this page. Every number on screen is derived from the four sliders you moved.

  • Time is only part of the case

    The calculator values hours. It does not price consistency across assessors, evidence attached to every score, a release gate, or an audit trail you can produce when a result is challenged - and at low volume those are the whole argument.

  • Setup effort is not modelled

    Building the rubric the first time takes real work. It is once per assessment rather than once per submission, so it does not scale with volume, but it is not free and this page does not pretend it is.