How much time can automation save a business?

Automation recovers time from four places. Most estimates only count the first.

1. The task itself. Re-typing an enquiry into the CRM, copying invoice lines into a spreadsheet, building the Monday report.
2. Switching around the task. Opening the right system, finding the record, checking the last email. Small per run, large across a week.
3. Checking and fixing. Spotting the typo in the phone number, correcting the duplicate record, resending the invoice with the right reference.
4. Waiting. The quote that sits in an inbox for two days before someone approves it. This does not show up in anyone's timesheet. It shows up as a slower sale.

The fourth one is often the most valuable. Automation removes waiting by triggering the next step the moment the previous one finishes, which you measure as cycle time (enquiry to quote, order to dispatch, month end to report) rather than hours logged.

Real numbers from a live build give a sense of scale. A digital publishing business with 63 web properties used to run inbound sales with a team reading emails, quoting, chasing invoices and upselling across every inbox. It now runs on an AI email sales agent that handles the standard deal end to end. Running cost is around $105 a month. The staff cost it replaced was $9,000 to $9,500 a month.

That is an unusually complete case: one process, high volume, fixed pricing, and a decision to take people out of the standard deal entirely. Most first automations remove part of a role's repetitive load rather than all of it. The method for estimating is the same either way.

• Nobody reassigns the hours. If a task frees eight hours a week and nobody decides what those hours are for, they get absorbed. Decide in advance: outbound calls, client reviews, the backlog project.
• The review step grows. If staff don't trust the output, they check everything, and the saving disappears. Build trust with a parallel run, then reduce checks to exceptions only.
• Exceptions are not routed. If unusual cases land back in someone's inbox with no context, they take longer than before. Send exceptions to a queue with the reason attached.
• Upkeep is ignored. A workflow that breaks silently for two weeks costs the time it saved, plus the clean-up.

Use the same numbers you estimated with: runs per week, minutes per run that are still manual, review time. Add two more:

• Cycle time for the process (enquiry to reply, order to dispatch).
• Error rate (wrong records, missed follow-ups, invoices sent twice).

Record the baseline before you build, check after the first month, and again after the third. The first month includes tuning, so do not judge it on week one.

Say you run a 12-person recruitment agency in Manchester. Three consultants each spend part of every day on the same jobs:

• Logging CVs from email into the applicant system: suppose 120 CVs a week at about 4 minutes each, so around 8 hours a week across the team.
• Scheduling interviews by email back and forth: suppose 30 a week at about 15 minutes each, so around 7.5 hours.
• A Friday client update compiled by hand: about 2 hours.

That is roughly 17.5 hours a week on paper. An automation that reads CVs, extracts the fields and logs them, plus a booking link for interviews, could remove most of the first two. Allow time for checking flagged CVs and the odd scheduling exception, and the realistic net saving is lower than the headline. Still, it is enough to give each consultant several more hours a week for client calls, which is where the agency makes money.

All figures here are illustrative. Your numbers come from your own process.

In an audit, we time the real process with the people who do it, rather than relying on estimates from a meeting. Each workflow gets an hours figure for today, an estimate of what the automation removes, and an honest allowance for review and upkeep. If a workflow would not pay back, we say so before building it.

After go-live, we compare against the baseline and agree with you where the recovered hours go. The number we care about is what your team does with the time.