Answer

How do I know whether the automation is actually working?

Four numbers, reviewed weekly: items handled without a person, items sent to a person, how often an automatic output had to be corrected, and hours saved against the baseline you measured before go-live. If the correction rate is falling and the automatic share is rising, it is working.

Last updated: 22 September 2026

Without a baseline there is no answer

The most common reason a company cannot tell whether a project succeeded is that nobody wrote down what the situation was beforehand. Volume per week and minutes per item, measured for one ordinary week before anything is built, is enough, and it takes an afternoon.

The correction rate is the honest metric

Volume figures flatter the project; corrections tell you the truth. A rising correction rate means something changed: a new product line, a new customer type, a supplier with a different document format. That signal is more useful than any satisfaction survey, and it arrives weeks earlier.

Watch one outcome number as well

Pick one business outcome the automation should move and track it separately: quotes sent within a day, calls answered, invoices booked before the deadline, no-shows per month. One number, measured the same way every week, is worth more than a dashboard nobody opens.

Asked next

How soon should we see movement?

Within the first month on volume, and by the second on hours. If neither has moved by then, the process or the scope was wrong, and that is worth saying out loud.

Do we need a dashboard for this?

No. A weekly summary arriving by email or chat is read; a dashboard usually is not. The numbers matter, the interface does not.