Key takeaways
- AI automation ROI comes down to a simple formula: annual time savings minus annual cost, divided by annual cost, times 100
- The most common mistake is comparing AI cost to employee salary. You're not replacing anyone. You're freeing up hours they currently waste on repetitive tasks
- For a typical build costing EUR 2,000-5,000 with a EUR 500-1,000/month retainer, most businesses break even in 3-6 months
- 74% of executives expect AI ROI within year one, and for simple workflows, that's realistic
- If a task takes less than 30 minutes a week, don't automate it. The setup cost won't justify the return
The formula most people get wrong
Here's how most people evaluate AI automation: they look at what the tool costs, compare it to what an employee costs, and decide it's either cheaper or more expensive. That comparison is wrong. Every time.
You're not replacing a person. You're replacing a process. The question isn't "is this cheaper than hiring someone?" It is "what does this manual process cost us right now, and what would the automated version cost?"
Those are very different questions with very different answers.
The manual process cost includes the hours your existing staff spend on it, the mistakes that happen because humans get tired and bored (and they do get bored of data entry by hour three), the leads that go cold because nobody followed up fast enough, and the opportunity cost of your best people doing work that doesn't need their brain.
The automated process cost includes the build, the monthly running costs, and the time someone spends reviewing the output.
When you frame it that way, the maths usually looks quite different from the "AI vs employee" comparison.
We've run this calculation for over a dozen businesses now. The ones that proceed typically see 2-4x ROI in year one. But the ones who get burned are almost always the ones who skipped the maths entirely and automated something because it sounded impressive, not because the numbers made sense.
The simple AI automation ROI formula
Here's the formula. It's not complicated. The hard part is being honest about the inputs.
Step 1: calculate annual savings
Hours saved per week x hourly cost of that time x 52 weeks = annual savings
Step 2: calculate annual cost
One-off build cost + (monthly running cost x 12) = annual cost in year one
Step 3: calculate ROI
(Annual savings - annual cost) / annual cost x 100 = ROI percentage
Step 4: calculate payback period
Annual cost / (annual savings / 12) = months to break even
That's it. Four lines of arithmetic. The challenge is getting accurate numbers for step 1, because most people dramatically underestimate how much time they spend on repetitive work. Track it for a week before you calculate. Actually track it. Don't guess.
Average SMBs spend 20-30% of their revenue on manual administrative tasks. Automation can reduce those operational costs by 30-40%. So the savings are real, if you pick the right process to automate.
Worked example 1: lead follow-up for an estate agent
Here's a real scenario we see constantly in this vertical.
The situation: an estate agent gets about 30 property enquiries per week through their website, Daft.ie listings, and phone calls. Someone on the team spends roughly 2 hours every day responding to these, qualifying them, and booking viewings. That is 10 hours a week.
The manual cost:
- 10 hours/week at EUR 25/hour = EUR 250/week
- EUR 250 x 52 weeks = EUR 13,000/year
- Plus the leads that go cold because the response took 4 hours instead of 4 minutes (and research shows responding within 5 minutes makes you 21x more likely to qualify a lead)
The automated version: an AI follow-up system responds to enquiries within 60 seconds, asks qualifying questions, and books viewings directly into the agent's calendar. Someone reviews the responses for about an hour a week. That's it.
The automated cost:
- Build: EUR 3,000
- Monthly retainer: EUR 500/month x 12 = EUR 6,000
- Total year one cost: EUR 9,000
The savings:
- 9 hours saved per week (10 minus 1 hour reviewing)
- 9 x EUR 25 x 52 = EUR 11,700 in time saved
- Plus faster response times mean more leads convert (hard to quantify exactly, but even a 10% improvement on 30 leads/week adds up)
The ROI:
- (EUR 11,700 - EUR 9,000) / EUR 9,000 x 100 = 30% ROI in year one
- Payback period: about 9 months
30% isn't going to make anyone faint with excitement. But here's the thing. Year two, there's no build cost. The annual cost drops to EUR 6,000. The ROI jumps to 95%. And that's before counting the extra leads you converted because you responded faster.
Worked example 2: content brief generation for e-commerce
The situation: an e-commerce brand creates weekly content briefs for their blog, social media, and email campaigns. The marketing manager spends about 8 hours a week on this, researching competitors, pulling search data, drafting outlines, and writing briefs for freelancers or in-house writers.
The manual cost:
- 8 hours/week at EUR 35/hour (marketing manager rate) = EUR 280/week
- EUR 280 x 52 = EUR 14,560/year
- Plus the inconsistency. Some weeks briefs are thorough, some weeks they're rushed because other priorities took over
The automated version: an AI system pulls competitor content, analyses search trends, generates structured briefs with target keywords and word counts. The marketing manager reviews and tweaks for 30 minutes per week.
The automated cost:
- Build: EUR 2,500
- Monthly retainer: EUR 750/month x 12 = EUR 9,000
- API costs for search and analysis tools: roughly EUR 100/month x 12 = EUR 1,200
- Total year one cost: EUR 12,700
The savings:
- 7.5 hours saved per week
- 7.5 x EUR 35 x 52 = EUR 13,650 in time saved
- Plus consistent quality every single week, which compounds over time as your content library grows
The ROI:
- (EUR 13,650 - EUR 12,700) / EUR 12,700 x 100 = 7.5% ROI in year one
- Payback period: about 11 months
Honestly, that's tight. And this is where the conversation gets interesting, because a 7.5% ROI doesn't scream "do this immediately." But the marketing manager just got 7.5 hours a week back. What do they do with that time? If they spend it on campaigns that generate revenue, the real ROI is much higher. If they spend it on more meetings, well, the maths stays at 7.5%.
This is the exact calculation we run for clients during our AI audit. The numbers tell one story, but the strategy behind how you redeploy saved time tells a bigger one.
Worked example 3: CV screening for recruitment
The situation: a recruitment agency receives about 200 CVs per week across their open roles. A recruiter spends roughly 3 minutes reading each one, comparing it to the job spec, and deciding whether to progress the candidate. That's 600 minutes, or 10 hours a week. On screening alone.
The manual cost:
- 10 hours/week at EUR 30/hour = EUR 300/week
- EUR 300 x 52 = EUR 15,600/year
- Plus the good candidates who get lost in the pile because by CV 150, the recruiter's eyes are glazing over
The automated version: an AI screening system reads each CV, scores it against the job requirements, and produces a shortlist with reasoning. The recruiter spends about an hour reviewing the shortlist and checking edge cases the system flagged.
The automated cost:
- Build: EUR 4,000
- Monthly retainer: EUR 800/month x 12 = EUR 9,600
- Total year one cost: EUR 13,600
The savings:
- 9 hours saved per week
- 9 x EUR 30 x 52 = EUR 14,040 in time saved
- Plus better candidate quality because the system doesn't get fatigued at CV number 147
The ROI:
- (EUR 14,040 - EUR 13,600) / EUR 13,600 x 100 = 3.2% ROI in year one
- Year two (no build cost): (EUR 14,040 - EUR 9,600) / EUR 9,600 x 100 = 46% ROI
- Payback period: about 11.5 months
So year one is basically break-even. Year two is where it pays off properly. This pattern shows up often with higher-cost builds. The upfront investment pushes the break-even point further out, but the year-two numbers look very different once that build cost disappears.
Costs people forget to include
Here's where ROI calculations go wrong even when people use the right formula. They forget costs. Both on the savings side and the expense side.
Costs on the automation side that get missed:
API and infrastructure costs. Most AI systems call external services. Those API calls cost money, sometimes pennies, sometimes pounds. A chatbot handling 500 conversations a month might cost EUR 50-150 in API fees alone. Small, but it adds up and you need it in the calculation.
Maintenance and updates. Systems need adjusting. Your business changes, your processes change, the AI models get updated. Budget EUR 500-1,000/month for ongoing maintenance on any serious automation. That's the retainer in our examples above, and it is not optional if you want the thing to keep working well.
Staff time learning the system. There's always a ramp-up period. Your team needs to understand what the automation does, how to review its output, and when to override it. Budget 2-4 weeks where productivity might actually dip before it improves. Nobody mentions this in the pitch, which is why we're mentioning it here.
The first month. It is almost always slower, not faster. You're configuring, testing, feeding the system examples of good and bad output, fixing edge cases. (This is normal. It doesn't mean the automation isn't working.)
Savings that get missed:
Error reduction. Humans make mistakes on repetitive tasks. Those mistakes cost money to fix. Automation doesn't eliminate errors, but it makes them consistent and fixable.
And then there's opportunity cost. What could your team be doing instead? This is the hardest number to quantify but often the biggest one. If freeing up 10 hours a week means your sales person makes 20 more calls, or your recruiter places one more candidate per month, the downstream revenue dwarfs the direct time savings.
Speed. Faster response times, faster turnaround, faster delivery. In some businesses, being first to respond literally wins the deal.
When AI automation is NOT worth it
We'd rather lose a sale than sell something that doesn't make sense. So here's when you should not automate.
The task takes less than 30 minutes a week. If someone spends 25 minutes a week on something, automating it will cost thousands to build and save maybe EUR 650/year in time. The maths doesn't work. Just keep doing it manually.
The task requires different judgment every time. If every instance is genuinely unique, if there's no pattern, automation won't help. AI is brilliant at pattern recognition. It is bad at novel situations. Negotiating a contract? Making a hiring decision based on culture fit? Leave those to humans.
You don't have a documented process yet. This one catches people out. You can't automate something that doesn't have clear steps. If different people on your team do the same task differently, and nobody agrees on the "right" way, you need to document the process first. Then automate the documented version.
The volume is too low. Handling 5 customer enquiries a week doesn't justify a EUR 3,000 build. Handling 50 does. There is a threshold where manual effort is simply cheaper, and there's nothing wrong with that.
You're automating to avoid fixing a broken process. If your lead follow-up is slow because your CRM is a mess and nobody updates it, automating the follow-up will just send faster responses from a broken system. Fix the process, then automate the fixed version. (This is genuinely the most common mistake we see.)
The payback period question
Most business owners don't actually care about ROI percentages. What they want to know is: when do I get my money back?
The payback period formula is simple:
Total year one cost / monthly savings = months to break even
For a typical engagement:
- Build cost: EUR 3,000-5,000
- Monthly retainer: EUR 500-1,000
- Monthly time savings: EUR 800-1,500
That puts break-even at 3-6 months for straightforward workflows. More complex systems with higher build costs (EUR 5,000+) might take 6-12 months.
74% of executives want to see AI ROI within year one. For simple automations like lead follow-up, chatbots, and basic workflow automation, that's very achievable. For complex multi-system integrations, set expectations at 12-18 months.
Here's what the payback timeline typically looks like:
- Month 1-2: setup, configuration, staff training. You're spending money and not saving much yet
- Month 3-4: system is running, savings start accumulating, some teething issues still being resolved
- Month 5-6: break-even point for most simple automations. The system is stable and your team knows how to work with it
- Month 7-12: pure return. The build is paid for, you're only covering the retainer, and savings compound
- Year 2+: this is where ROI gets serious. No build cost, just the retainer versus the full annual savings
The businesses that see the fastest payback share two things: they picked a high-volume, time-heavy process to automate first, and they had a clear, documented workflow before they started.
Your next step
The boring answer to "should I invest in AI automation?" is: it depends on your specific numbers. But now you have the formula to find out.
Pick your most time-consuming repetitive task. Track how long it takes for one week. Run the formula. If the ROI is above 20% in year one and the payback period is under 8 months, it is probably worth doing.
If you're not sure about the numbers for your business, or you want someone to run the calculation with you, that's what the AI audit is for. It is EUR 500, takes about two hours, and you'll walk away knowing exactly what's worth automating and what's not. If you decide to go ahead with a project, the audit fee gets credited against the build cost.
No pressure. Just maths.