Where is the return on investment from AI?
Why huge AI time savings don’t seem to impact the bottom line
I keep seeing the same thing happen. People tell me they’re saving loads of time with AI, but when leadership looks at the books, nothing seems to have changed.
Your people save 20-30 minutes a day, everyone feels good about AI adoption, but then you’re scratching your head wondering why the P&L looks the same.
Here’s what makes this tricky.
On certain tasks, I’ve personally seen 10 to 20X productivity gains – both for myself, and folks I’ve helped with AI.
Writing proposal documents, summarising long reports, analysing data – the improvements can be staggering.
But that’s not across my entire workweek.
It’s specific tasks that might take up 20-30% of my time.
So the overall impact gets diluted across everything else I do.
I see this in my own work. On certain tasks – writing proposals, creating workshop materials, analysing survey data – I get those 10-20X gains. But I don’t work any fewer hours.
I put those saved hours straight back into the business. I write more content, take on extra client work, or spend the time making what I produce better.
Parkinson’s Law: work expands to fill the time available
Time savings alone aren’t enough – you need a deliberate plan for that freed-up capacity.
Research backs this up – a major study of 5,000+ support agents from McKinsey found AI delivered 14% productivity gains, but only in specific, well-structured work contexts.
I want to share what happened with one of my clients and the practical framework we developed to turn AI time savings into measurable business value.
Case study
Last year, I helped a consulting firm in the UK with approximately 50-60 staff adopt AI across their entire organisation. This was a comprehensive AI adoption where we rolled out tools and trained staff to use them properly.
Six months later, we surveyed the team. The results looked fantastic. 95% of staff were using AI daily, and reported saving an average of 23 minutes per workday. The feedback was overwhelmingly positive. People loved the tools. Productivity felt higher. Morale was up.
But when we looked at the quarterly results, we ask the question “Where’s the money?”
With a very conservative blended average hourly rate of £120, those time savings looked impressive on paper:
55 people × 23 minutes × 21 days = 443 hours. At £120/hour, that's £53k a month.
Looking at this number on the whiteboard, it felt solid and concrete.
“So we’re saving over £50k a month?”
Here’s the thing – it’s not cash savings. It’s extra capacity.
Extra capacity distributed in tiny amounts across the entire team.
And that’s where many AI business cases start to fall apart.
Saved time doesn’t automatically become value
Saved time is like potential energy. It needs a clear path to become actual value. Economists describe this as the productivity J-curve – the idea that new technologies often need complementary changes before benefits show up in financial results.
It’s capacity. And capacity only becomes value when you deliberately convert it.
The savings come in tiny fragments that are hard to use.
Research on attention residue shows why this matters – scattered minutes are genuinely hard to convert into productive deep work.
Or the work speeds up but it’s the wrong step in the process.
This is basic operations theory: speeding up non‑bottleneck steps doesn’t increase overall system throughput.
Other blockers include:
- Managers can’t figure out how to redeploy people
- Quality checks eat into the headline savings
- People take the time as a well‑being dividend – less stress, fewer late nights, more breathing room
A note on the human side of things: even hard-nosed managers should pay attention to that last point.
Stressed employees make more mistakes and leave for other jobs.
Less stress means better work and people who stick around.
None of this is bad, but it means we need to think a lot more about the conversion from time to value.
The value goes beyond just productivity
There’s a whole other layer here too: most people talking about AI gains tend to focus on productivity – doing the same work faster.
But as well as Productivity, AI delivers at least three other types of value that are harder to measure but equally real:
First, new Capabilities. Your team can now do things they simply couldn’t before – analyse patterns in large datasets, generate multiple creative options quickly, or handle complex research that would have required specialists.
Second, Smarter Decision-Making. AI helps us consider more variables, crunch more data, easily use decision-making frameworks, catch errors we’d miss, and spot patterns in information. The decisions get better, even if they don’t take less time.
Third, Accelerated Learning. People pick up new skills faster when AI can guide them through complex processes step by step.
I write about all of these in more detail in my free whitepaper “Opportunity and AI Adoption” (no email opt-in required)
These broader benefits are real, but they’re even harder to capture in traditional ROI calculations.
This means your AI investment might be delivering value that doesn’t show up in time‑saved metrics at all.
Note when AI enabled new capabilities, improved decision quality, or accelerated learning, these often don’t hit this month’s P&L, but they ARE building competitive advantage.
Why AI fluency matters more than immediate ROI
There’s another critical factor we haven’t discussed yet – AI fluency. Your team is building skills that will define the next decade of business.
Even if you’re not seeing cash returns today, you’re building organisational muscle memory. Every prompt your team writes, every workflow they test, every tool they experiment with – they’re learning how AI actually works in practice. This is important because AI capability is accelerating. The tools available in 12 months will be far more powerful than what we have today.
You can’t identify operational bottlenecks until you start using AI. We discovered our consulting client’s real constraint wasn’t writing speed – it was client review cycles. We only found that out after six months of AI use showed us exactly where time was actually stuck in the system.
Think of it this way: companies that wait for perfect ROI proof before adopting AI will be competing against teams who’ve been building AI fluency for years. It’s like waiting to adopt email until you could prove it saved money. By the time the ROI was obvious, everyone else had already figured out how to use it effectively.
What this means for different organisations
The size of your organisation completely changes how you should think about AI adoption and ROI.
If you’re running a small business or working solo, you have a massive advantage right now – and it won’t last forever.
You can experiment with a new AI tool this afternoon. If it doesn’t work, you’ll know by tomorrow and try something else. No procurement process, no IT security review, no change management committee. This nimbleness is gold.
I see this with my clients all the time. A solo consultant can completely transform their service offering in a month. They spot an AI capability, test it with a client, refine the approach, and suddenly they’re delivering something their bigger competitors can’t touch. Meanwhile, larger consultancies are still writing the business case for the same tool.
When I save an hour on proposal writing, I immediately know what to do with it. Take on another client, improve my workshop materials, or tackle that marketing project I’ve been avoiding. There’s no committee meeting about capacity allocation. The path from time saved to value created is short and clear.
But you need wins fast. Pick AI uses that directly connect to revenue. If AI helps you respond to leads faster, measure conversion rates. If it helps you create content, track engagement and leads. You can’t afford to build AI fluency for its own sake – you need it to pay the bills.
The good news? You can move fast enough to find those wins. Test an AI tool on Monday, measure results by Friday, scale up or move on next week. This speed is your competitive edge – use it before the window closes.
If you’re leading a larger organisation, you face different challenges. If you’ve got 23 minutes saved daily across hundreds of people – that’s massive potential. But those minutes are scattered across departments, teams, and time zones. You can’t just tell everyone to “use the time wisely” and expect magic.
Every AI implementation needs approval, compliance reviews, security assessments. By the time you’ve rolled out a tool, smaller competitors have already tested five others and moved on to what actually works. This is frustrating, but it’s your reality.
What works in larger organisations is systematic capacity conversion. Pick specific teams or processes where saved time can accumulate into something meaningful. If your marketing team saves 2 hours per campaign, commit to running more campaigns. If your analysts save time on reports, have them tackle the backlog of strategic questions nobody had time for before.
You do have one advantage – you can afford to invest in AI fluency without immediate returns. Your scale means you can treat the first year as education. Build those skills across your organisation while your competitors wait for proof. When AI capabilities take another leap forward – and they will – you’ll have hundreds of people ready to use them.
Both scenarios share something important: the organisations winning with AI aren’t waiting for perfect ROI calculations. Small businesses are exploiting their speed to find immediate applications that drive revenue. Larger companies are building capabilities at scale. Both understand that the capacity AI creates only becomes valuable when you actively direct it somewhere specific.
Key insights
- Time savings aren’t cash – they’re capacity. 23 minutes saved per person sounds impressive, but unless you have a plan to convert that capacity into value, it stays theoretical.
- The work expands to fill the time. People reinvest saved time into doing more or doing better. This isn’t failure – it’s human nature. Plan for it.
- AI delivers four types of value, not one. Beyond productivity gains, you get new capabilities, better decisions, and accelerated learning. Often only one of these shows up in time-tracking and user surveys.
- You’re building AI fluency, not just saving time. The skills your team develops today will matter more than this quarter’s time savings. AI capability is accelerating – teams need to keep pace.
- Bottlenecks only reveal themselves through use. You can’t optimise a system you haven’t tested. Start using AI to discover where the real constraints are.
- Your organisation size determines your AI strategy. Small businesses and solos have a speed advantage they should exploit right now. Larger companies need to invest in widespread fluency even without immediate ROI.
- This isn’t an argument against AI adoption – it’s an argument for better planning. The companies winning with AI aren’t the ones with perfect ROI models. They’re the ones who started early, learned fast, and designed systems to convert capacity into value.
If you’re seeing time savings but no financial impact, you’re not failing. You’re in the middle of the process. The question isn’t whether to continue with AI – it’s how to design the bridge from saved time to business value, and take advantage of the OTHER opportunities AI brings us, like new capabilities, smarter decisions, and accelerated learning.
If you’ve seen similar time savings but mixed ROI signals, what’s blocking the conversion in your situation? I’d be interested to compare notes. Please drop me a note on LinkedIn and let me know your thoughts.
