Automation paid off if the monthly savings (hours saved times the hourly cost of labour, plus extra profit from faster replies) are larger than the monthly cost of the solution. For that you need three numbers written down before you start: time per task, tasks per month and time to reply. Without a baseline you have nothing to compare against after four weeks.
Most businesses make the same mistake here. They switch automation on, and two months later things feel "somehow easier", but they can't say whether it was worth it. A feeling is not a number. And when it's time to renew the subscription or pick the next task, there is nothing to point to.
What is automation ROI and what goes into the calculation?
ROI (return on investment) tells you what you got relative to what you put in. For automation it has two sides.
Time saved. Hours that you or your staff no longer spend on repetitive work. For a quote, that is retyping line items from last year's Excel. For an enquiry, the first reply. For appointments, going back and forth over messages.
Extra revenue. Enquiries that would have been lost because the reply came too late. A Harvard Business Review study (2011) by Oldroyd, McElheran and Elkington, covering 2,241 US companies, found that the average first reply took 42 hours and that 23% of companies never replied at all. Companies that replied within an hour were nearly seven times more likely to qualify the enquiry than those that replied later. This part is harder to measure, so count it cautiously. More in Speed to lead: what the research says.
Which numbers do you write down before you start?
Take one task and record its baseline. Not five at once.
| Number | How to get it | Example |
|---|---|---|
| Minutes per task | Time ten tasks in a row with a clock | 25 minutes per quote |
| Tasks per month | Count them in email or a spreadsheet | 40 quotes |
| Time to reply | Time between receipt and first reply, from timestamps | 6 hours |
| Share of enquiries with no reply | Count the ones nobody answered | 1 in 10 |
| Hourly cost of labour | Gross pay plus contributions, divided by hours worked | see below |
Don't guess the hourly cost. Take the employer's full cost for the person who does the task and divide it by the hours they actually work. If you do the task yourself, use the realistic value of your hour, not the one you wish it were.
How do you calculate ROI? A simple formula
- Hours saved per month = (minutes before − minutes now) × number of tasks ÷ 60
- Savings in euros = hours saved × hourly cost of labour
- Extra profit = extra deals won × average profit per deal (enter it only if you can justify it)
- Monthly benefit = savings + extra profit
- Net benefit = monthly benefit − monthly cost of the solution
- Payback period = one-off setup cost ÷ net benefit
An invented example, for illustration only: a quote used to take 25 minutes and now takes 10. There are 40 quotes a month. The saving is 15 × 40 ÷ 60 = 10 hours. If an hour of labour costs 20 euros, that is 200 euros a month. If the solution costs more than that and you can't justify any extra deals, automation doesn't pay off for this task. That is also a useful result.
What does research say about measurable effects?
The interesting part is the method. Both studies measured one concrete thing before and after the tool was introduced.
Brynjolfsson, Li and Raymond (NBER Working Paper 31161, 2023) followed more than 5,000 support agents. With an AI assistant they resolved about 14% more issues per hour, and novices 35% more. Zhang and Narayandas (Harvard Business School, Management Science) found chat replies with AI support were 22% faster, and 70% faster for less experienced agents, while customer sentiment improved.
That is customer support in larger organisations, and it doesn't mean you'll get the same result. What it does show is that the researchers picked measurable numbers (issues per hour, reply time) and measured them before the rollout. Do the same.
How do you measure the result after four weeks?
- Week 0: write down the baseline from the table above.
- Weeks 1 to 4: let the automation run and don't change other things at the same time.
- End of week four: measure the same numbers the same way.
- Compare with the formula above and write one sentence: "Task X saves us Y hours a month."
- Review the mistakes. How many messages or quotes did a person have to correct? If it's many, the cost of corrections is part of the calculation.
- Decide: renew, fix the rules or stop.
Don't compare three days before and after. Weekly swings (month end, holidays) can show or hide almost anything.
What to do today
Pick one task and write down its baseline today. If you don't know which, What to automate first helps. How to work out what manual work costs you is covered in The cost of manual work.
Frequently asked questions
How do I calculate automation ROI?
Add up the hours saved multiplied by the hourly cost of labour, then add extra profit from enquiries you kept. Subtract the monthly cost of the solution. The result divided by that cost is the ROI. Write the numbers down before you start, or there is nothing to compare.
When can I first judge whether automation paid off?
After four weeks of measuring, if you wrote down the baseline beforehand. That is long enough for normal weekly swings to show, and short enough to fix mistakes before they spread to everything.
Which numbers do I measure without proper analytics?
Three are enough: how many minutes one task takes, how many tasks there are per month and how long it takes to reply or send a quote. You can read all three from email, a calendar or a spreadsheet, with no special tool.
Sources
Luka Tomič · Founder of Alra
Luka builds AI systems for Slovenian businesses that want to get more done with the same team. He writes about what he sees with clients: where time gets lost and how to get it back.