I track every euro I spend on AI tooling and every hour AI saves me. Not because I’m obsessive about metrics—though I might be—but because vague claims about AI productivity don’t help anyone make investment decisions. “AI makes you 10x more productive” is a marketing slogan, not a business case.
So here are my actual numbers. Two years of AI investment in a real business. Real costs, real savings, real revenue impact. No rounding in my favor.
The Full Cost Picture
Here’s what I’ve spent on AI tooling over 24 months:
Direct AI costs (model subscriptions, API usage):
- Year 1: €4,800 (averaging €400/month, with some months higher during heavy production)
- Year 2: €6,200 (averaging €517/month, increased usage and added tools)
- Total: €11,000
Infrastructure costs (workflow tools, storage, integration platforms):
- Year 1: €1,800
- Year 2: €2,400
- Total: €4,200
Learning investment (courses, experimentation time valued at my consulting rate):
- Year 1: ~80 hours × €120/hour = €9,600
- Year 2: ~30 hours × €120/hour = €3,600
- Total: €13,200
Setup and maintenance time (valued at consulting rate):
- Year 1: ~120 hours × €120/hour = €14,400
- Year 2: ~60 hours × €120/hour = €7,200
- Total: €21,600
Grand total cost over two years: approximately €50,000
That number is higher than most people expect when they think about “AI subscription costs.” The tooling itself—€15,200 over two years—is modest. The real cost is the time investment in learning, building, and maintaining the systems. Anyone who tells you AI is “just a subscription fee” is ignoring the operational investment that makes AI productive.
But that total cost needs context. Because the return side of the equation is what makes the investment worthwhile.
The Time Savings
I track time savings by function. Here’s the weekly time recaptured through AI-assisted operations:
Content production: 12-15 hours/week saved (producing 12-15 pieces versus the 3-4 I could produce manually in the same time) Research and analysis: 5-8 hours/week saved Client deliverables: 8-10 hours/week saved (higher quality with less time investment) Administration: 6-8 hours/week saved Communication and correspondence: 3-4 hours/week saved
Total weekly time savings: 34-45 hours/week
Annualized at approximately 48 working weeks: roughly 1,600-2,160 hours per year.
At my consulting rate of €120/hour, that represents €192,000-€259,000 in equivalent labor value per year. Over two years: €384,000-€518,000.
Even using a conservative internal valuation—not my external consulting rate but a lower internal rate of €60/hour—the two-year value is €192,000-€259,000.
Against a total investment of €50,000, that’s a return of roughly 4-10x depending on the valuation method. The payback period was approximately 4-5 months.
Now, I need to be honest about what these numbers mean and don’t mean. The “saved hours” aren’t all converted into revenue. Some go to more content production (which builds audience and authority). Some go to better client work (which builds retention and referrals). Some go to personal time (which builds sustainability). And some, frankly, go to experimentation that doesn’t produce immediate returns.
The time-to-revenue conversion isn’t 100%. But it doesn’t need to be. Even at a 30% conversion rate—30% of saved hours translating to additional revenue-generating work—the ROI is strongly positive.
The Revenue Impact
This is harder to measure precisely because revenue is influenced by many factors, not just AI adoption. But here’s what I can attribute:
Increased client capacity. Before AI, I could manage 2-3 consulting clients simultaneously. After AI, I comfortably manage 3-4 with better deliverable quality. That additional client represents roughly €30,000-€50,000 in annual revenue.
Content-driven business development. My tripled content output has measurably increased inbound inquiries. Tracking shows that roughly 40% of new consulting clients in Year 2 discovered me through content that wouldn’t have existed without AI-assisted production.
Editorial agency revenue. An entire revenue stream—my AI-powered editorial agency operation—that wouldn’t exist without AI capabilities. This generates consistent monthly recurring revenue from content clients.
Book production. The Subtract to Ship series and Late to the Table volumes were produced at a speed and volume that would have been impossible manually. As I described in my piece on building six books using AI-native methods, the book revenue is directly attributable to AI-enabled production capacity.
Conservative total revenue impact: €80,000-€120,000 additional annual revenue attributable to AI capabilities. Over two years: €160,000-€240,000.
Against a €50,000 total investment, the revenue ROI alone is 3-5x, not counting the time value or quality improvements.
The Non-Financial Returns
Some returns don’t show up in financial metrics but matter enormously:
Reduced stress. Before AI, I was chronically overextended—too many commitments, not enough hours. AI didn’t eliminate the work; it redistributed my effort from production (exhausting) to judgment (energizing). I work the same hours but finish each day less depleted.
Better decision quality. With AI handling data processing and analysis, my decisions are based on more complete information than before. I catch patterns I would have missed, consider scenarios I wouldn’t have had time to model, and enter client conversations better prepared.
Competitive positioning. Being early to AI-native operations has positioned me as a practitioner, not just a commentator. Clients hire me partly because I’ve done what I advise—my operational experience is the credential.
Sustainability. Before AI, my output level was only sustainable with significant personal sacrifice. Now the same output level is sustainable indefinitely because I’ve shifted to a less depleting mode of work. This matters for a solo founder thinking about the next 10 years, not just the next quarter.
How to Calculate Your Own AI ROI
If you’re considering AI investment, here’s the framework I recommend:
Step 1: Baseline your current operations. Before changing anything, track how you spend your time for two weeks. Categorize by function (content, research, client work, admin, etc.) and by type (production, judgment, coordination, waste).
Step 2: Identify the AI-addressable hours. From your baseline, identify hours spent on production-heavy tasks that AI could handle. Typically 40-60% of total hours for knowledge workers. Be conservative—overestimating AI savings leads to disappointment.
Step 3: Estimate realistic time recovery. Assume AI saves 50-70% of the time on addressable tasks (not 100%—you still need to direct and review). Multiply addressable hours by this recovery rate. This is your estimated weekly time savings.
Step 4: Value the saved time. Use a rate that reflects what you’d realistically do with the time. If you’d use it for billable work, use your billing rate. If you’d use it for business development, use a fraction of your expected return per hour spent on development. If you’d use it for personal time, value it at whatever prevents you from burning out.
Step 5: Estimate total costs. Include tooling, infrastructure, learning time, setup time, and ongoing maintenance. Most people underestimate the time investment by 50-70%. My experience: plan for 100-150 hours of setup and learning in the first year, plus 50-80 hours of maintenance.
Step 6: Calculate. If (annual value of saved time + incremental revenue) > (annual costs × 1.5), the investment is likely worthwhile. The 1.5 multiplier accounts for the things you’ll underestimate—setup time, maintenance, failed experiments.
For most knowledge workers I’ve run this calculation with, the ROI is positive within 4-8 months. The variance depends on how production-heavy their current work is (more production = higher ROI), how expensive their time is (higher rates = faster payback), and how systematically they implement (workflows > ad hoc use).
The Investment Curve
Something that caught me off guard: the ROI isn’t linear. It follows a curve:
Months 1-3: Negative ROI. You’re investing time and money in learning and setup. Output may actually decrease temporarily as you’re building systems rather than producing work. This is the valley that kills most AI adoption efforts—people try it, see no immediate return, and quit.
Months 4-6: Break-even. Your first workflows start producing returns. Time savings begin to appear. But they’re modest because your systems are still being refined.
Months 7-12: Accelerating returns. Workflows are stable, agents are refined, and you’ve learned where AI helps most. Returns compound as each workflow improvement benefits all future work.
Year 2+: Compound growth. Your knowledge base is rich, your workflows are mature, and new capabilities layer on top of existing ones. The marginal cost of each new AI application drops because the infrastructure is already in place.
I’ve seen this curve play out in my own operations and in every client I’ve guided through the transition. The patience required during the first three months is the single biggest predictor of long-term success. Founders who push through the negative-ROI valley almost universally end up with strongly positive returns. Those who quit in month 2 never get there.
What the Numbers Don’t Capture
One caveat: ROI calculations assume your time has consistent value, which isn’t true. An hour spent on a breakthrough strategic insight for a client is worth far more than an hour spent formatting a report. AI’s real value isn’t just saving time—it’s redirecting your time from low-value to high-value activities.
If I save 40 hours per week on production work and spend 10 of those hours on strategic thinking that generates a single insight worth €50,000 to a client, the ROI calculation based on hours saved dramatically understates the actual return.
This is the dimension that AI skeptics miss when they argue that AI “just saves time.” Time is not fungible. An hour of judgment work is categorically different from an hour of production work. AI doesn’t just give you more hours. It gives you better hours—hours spent on the work where your unique capabilities create the most value.
And that, more than any number I’ve shared today, is the real return on AI investment. Not just making things faster, but making things possible that were previously out of reach for a solo operator.
Takeaways
- My total AI investment over two years was approximately €50,000 (including tooling, infrastructure, learning, and setup time), returning 3-10x depending on valuation method.
- Weekly time savings of 34-45 hours come from content production, research, client deliverables, administration, and communications combined.
- The ROI curve is non-linear: negative for months 1-3, break-even at months 4-6, accelerating from month 7+, and compounding in year 2—patience through the early valley is the biggest success predictor.
- Include all costs in your ROI calculation—tooling is only 30% of the true cost; learning and setup time account for the majority.
- AI’s deepest value isn’t just saving hours—it’s redirecting your time from production (where your value is limited) to judgment (where your unique capabilities create the most impact).
