# Prioritization and baseline worksheet

From the Dryvn resource library: https://www.dryvn.ca/resources/what-to-automate-first

## Instructions

Use one copy per candidate task. Score the six measures with plain words (high, medium or low), not invented numbers; you are comparing tasks against each other, not producing a precise figure. Score "consequence of a wrong action" first, because it can veto the rest. Decide the lane, then record a two-week baseline before you change anything. Run one change for about four weeks, keep counting the same things, and compare.

## Blank template

### Scoring

| Measure | Your score (high / medium / low) and a note |
| --- | --- |
| Task (one sentence) | |
| Frequency (occurrences in a normal week) | |
| Minutes per occurrence (total, everyone who touches it) | |
| Clarity (could a new hire follow written steps?) | |
| Data availability (is the information already in a system?) | |
| Exception rate (how often it leaves the normal path) | |
| Consequence of a wrong action (how bad, how reversible) | |
| Lane (ordinary software / fixed automation / AI assistance / human decision / not yet) | |
| Reason for the lane | |

### Baseline (two normal weeks, before any change)

| Item | Week 1 | Week 2 |
| --- | --- | --- |
| Occurrences | | |
| Total minutes per occurrence | | |
| Exceptions | | |
| Mistakes found later | | |

### The experiment

| Item | Your answer |
| --- | --- |
| The one change | |
| Owner during the experiment | |
| Who reviews exceptions | |
| Start date | |
| End date | |
| How you will know it worked (which baseline number must move) | |
| Oversight minutes the change adds per week (fill in as you go) | |

## Fictional example (illustrative only)

This describes a made-up residential cleaning company. It is not a customer result or a benchmark.

| Measure | Example |
| --- | --- |
| Task | Re-typing new bookings from the website form into the schedule |
| Frequency | High (about 25 a week) |
| Minutes per occurrence | Low (about 4 minutes, one person) |
| Clarity | High; the steps never change |
| Data availability | High; every field is already on the form |
| Exception rate | Low; occasional duplicate booking |
| Consequence of a wrong action | Low; a wrong entry is caught at the day-before reminder and fixed in a minute |
| Lane | Fixed automation |
| Reason | Rule-based, reversible, information already captured. Checked first whether the booking tool does this on its own. |

Baseline: 24 and 27 bookings; about 4 minutes each; 1 and 2 exceptions (duplicate bookings); 3 and 2 mistakes found later (wrong address or time).

Experiment: connect the form to the schedule so entries are created automatically; owner reviews the day's new entries each morning (about 10 minutes of oversight per week); run for four weeks starting the first Monday of the month; success means minutes per occurrence fall to under one and mistakes found later do not rise.