Browser tool · Service business operations
Which office task should you automate first?
Compare two tasks with transparent ratings, editable weights and clear reasons to investigate, keep a person involved or defer.
Written by Fireflax · 7 minute exercise
Reviewed by Codex — source and implementation review ·
Compare two recurring office tasks using visible criteria. The score is a conversation starter; high consequences, poor data access and unclear rules can stop a task regardless of its score.
Use fictional or non-sensitive details. Inputs and edited output stay in this page's memory and clear on reload. Nothing is sent, booked or connected. JavaScript is needed for the tool; the explanation and worked example below remain readable without it.
Use the same 0–4 scale for both tasks.
0 means none or very low; 4 means very high. For frequency, 0 is rare and 4 is daily or more. For manual effort, compare low to high effort per occurrence. Rate repeatability and usable data access from absent to strong; exception rate and error consequence from low to high. These anchors are provisional judgments, not research-validated weights.
The fictional defaults compare enquiry assignment (A) with complaint resolution (B). Change all ratings to describe your own tasks, without entering customer records.
The score and the gates are both visible.
Score = 100 × weighted sum of frequency, effort, repeatability, access, (4 − exceptions) and (4 − consequence), divided by (4 × total weight), rounded to the nearest whole number. At least one weight must be nonzero. Equal inputs and weights produce tied scores.
- Defer: consequence is 3 or 4, or access/repeatability is 0 or 1. Changing weights cannot bypass these gates.
- Investigate native feature: a feature exists or you have not checked yet.
- Human-assisted: exceptions are 3 or 4, or the remaining score is below 60.
- Rules-first: the remaining case. This suggests a bounded investigation, not permission to connect live systems.
Sources and review notes
Drafted with Codex and checked against the documented formulas and Fireflax sample on the review date. Examples and ratings are fictional planning assumptions, not customer results. Inputs stay in browser memory.
- Native feature or custom connection
Check existing tools before adding another system.
- Enquiry sample
A bounded example, not evidence that every task fits.