What to measure before automating a task

Before automating a task, watch it happen a few times and record what it costs today. Otherwise, a new workflow can feel faster while moving the work into checking, correcting or recovering from mistakes.

You do not need elaborate software for a baseline. A small, dated log is enough if the definitions stay consistent before and after the change.

Record a representative period

Choose a period that includes normal and awkward cases, such as two to four weeks for a task that happens most days. If volume changes sharply by season or campaign, note that context. Count how often the task starts, how long the main work takes and how often someone has to fix or escalate it.

For each instance, record:

Avoid collecting customer details you do not need. The purpose is to understand the workflow, not to create a second store of personal data.

Compare the whole task

After a small pilot, measure the same items over a comparable period. Include setup, supervision, failed runs and corrections. If an automation saves ten minutes of copying but adds fifteen minutes of review, the process has not improved on time alone. It may still help in another way, such as clearer records or fewer missed handovers; define that benefit and test it honestly.

For example, imagine twenty weekly enquiries each take six minutes to copy into a tracker: two hours of hands-on work. If a proposed integration reduces copying to zero but creates two minutes of checking per enquiry plus twenty minutes fixing exceptions, the new total is one hour. The apparent saving is one hour for that illustrative week, before setup and maintenance. Use your own measured counts and timings rather than presenting this example as a result.

Do not report a percentage saving from a handful of convenient examples. Record the number of cases, the range of times and the kinds of exception that still need a person. If the old process changes during the pilot, make that visible too.

Decide what success would mean first

For a hypothetical enquiry-to-tracker workflow, useful acceptance criteria might be: each test enquiry appears once, fields match the original, the owner can spot a failed transfer and the manual route still works. Time saved is useful only if those basics hold.

The baseline also helps you choose whether the task needs a simple rule, better source data or an AI-assisted step. Explore the automation service or describe your workflow when you are ready to discuss a pilot.