The short version: I tracked every AI-assisted task in my business in a spreadsheet for 30 days, and the honest total was 9 hours and 40 minutes saved, not the 20-plus hours a week you see quoted in reports. Nearly all of that came from three tasks, and on two others AI cost me time rather than saved it. If you want to know whether AI is worth it for your business, stop reading surveys and run your own tracker.
Why a spreadsheet instead of another survey
Every few months another report lands claiming small business owners save 12, 16, sometimes 20 hours a week using AI. I've quoted some of those numbers myself in past posts. But nobody in those reports is showing their working. They ask "roughly how much time do you think AI saves you" and people guess, usually generously, because it feels good to say the tool is brilliant.
So in September 2026 I built a Google Sheet with five columns: task, tool used, estimated time without AI, actual time with AI, and time spent editing or fixing the AI output afterwards. That last column is the one everyone skips, and it's the one that matters.
What I tracked
I logged seven recurring tasks across 30 days:
- Client proposals (11 written)
- LinkedIn posts (22 published)
- Meeting notes and follow-ups (14 calls)
- Email replies to warm leads (around 40)
- Blog first drafts (6)
- Video captions for short clips (9 videos)
- Invoice chasing emails (7)
I used Claude for proposals and long-form drafting, ChatGPT for quick email replies, Otter.ai for meeting transcription and summaries, and Descript for video captions. Nothing exotic, just the tools most consultants my age or younger already have open in a browser tab.
Where AI saved real time
Client proposals were the biggest win. My old process was 45 minutes per proposal, built from a Word template I'd tweak each time. With Claude drafting the first pass from bullet points I gave it, then me editing for 18 minutes, each proposal dropped to roughly 22 minutes total including the prompt writing. Across 11 proposals that's just under 4.2 hours saved.
Meeting notes were the second win. Otter transcribed and summarised 14 calls, and turning those summaries into follow-up emails took about 6 minutes each instead of the 20 minutes I used to spend reconstructing what was said from memory and scrappy notes. That's another 3.2 hours.
Invoice chasing emails, small but real: ChatGPT drafts a polite nudge in seconds, I tweak the name and amount, done in under a minute versus the 8 minutes I used to spend agonising over tone. Across 7 chasers, about 45 minutes saved.
Add those three up and you get roughly 8 hours saved, which already tells you something the headline reports don't: three tasks did almost all the work. The other four barely moved the needle, and two of them went backwards.
Where AI cost me time (the bit nobody wants to admit)
Here's the uncomfortable part. Blog drafts were meant to be my biggest saving. Claude produced a full first draft in about 4 minutes. Then I spent an average of 55 minutes rewriting it to sound like me instead of like every other AI-generated business blog on the internet, the same rhythm, the same "in today's fast-paced world" energy, the same three-point lists. My own drafts from scratch, no AI at all, took about 65 minutes. Net saving: 6 minutes across six posts. Not nothing, but not the transformation people promise you.
Video captions were worse. Descript's auto-captions struggled with client names, British pronunciation, and the odd Hebrew word I drop into conversation living partly in Israel. Fixing captions on 9 short videos took longer than typing them out manually would have on 5 of the 9. That task lost me time, roughly 40 minutes across the month.
The pattern is clear once you see it written down: AI saves you real time on structured, repeatable tasks with low personality requirements, invoices, transcripts, proposal templates. It costs you time on anything that needs to sound distinctly like you, because the editing to strip out the AI voice takes almost as long as writing would have. If your business is your personality, and for a lot of small business owners and consultants it is, that's a cost the case studies never mention.
The real total, and what it costs to get it
Net for the month: 9 hours 40 minutes saved. My subscriptions across Claude, ChatGPT Plus, Otter Business, and Descript came to £71 that month. That works out to roughly £7.30 for every hour saved. Compare that to what an hour of my own billable time is worth and it's a clear yes, but it is not the free lunch the LinkedIn posts about AI make it sound like. If your billable hour is under £15, the maths gets a lot less exciting, and that matters for anyone reading this who runs a smaller margin business than a consultancy.
This is also the honest answer I give clients who ask whether they should hire someone to set all this up for them rather than fumble through it alone. If you're going to spend real money working out where AI pays off in your business, it's worth understanding what an AI consultant costs before you commit, because the tools themselves are cheap and the real expense is the hours spent figuring out which tasks are worth automating in the first place.
How to run your own 30-day tracker
You don't need anything fancier than a spreadsheet. Here's the exact method:
Want AI doing the heavy lifting in your marketing?
I build the systems that handle the boring 80 percent, so you get your week back. Done properly, with the human kept in.
- Step 1: Pick five recurring tasks you already do most weeks. Don't guess at new ones, track what's real.
- Step 2: For two weeks, log time without AI as your honest average from memory or a quick check of your calendar.
- Step 3: For the following two weeks, log time with AI, including every minute of editing, correcting, or rewriting afterwards. Not just the generation time.
- Step 4: Total both columns. Subtract. Anything with a positive net after four weeks is worth keeping. Anything close to zero or negative, drop it.
If you want to turn this into something clients or your team can see rather than just a private spreadsheet, this is exactly the kind of thing that works well as an interactive calculator on your own site, people love plugging in their own numbers and seeing a personalised result far more than they trust a generic stat.
What this means if you're deciding whether to bother with AI at all
The discipline of tracking daily, small habits rather than expecting one big transformation is the same principle Duolingo built its entire growth strategy around, small consistent reps beat one dramatic overhaul. AI in a small business works the same way. It is not going to hand you back a whole day. It hands you back 20 minutes here, 6 minutes there, and only adds up if you track it and only keep the tasks that earn their place.
It also rewards businesses that already think in terms of automated chains rather than one-off tools, the same logic IFTTT built its whole marketing message around years before generative AI existed, connect the small repeatable things and let them run without you. My proposal workflow now runs largely on autopilot from a template through to a drafted first pass, and that's the version of AI that's paying off, not the version where I ask an AI to be creative for me.
If your business leans on video content, it's worth knowing where AI helps versus where it doesn't before you invest hours fixing captions. GoPro built a brand entirely on user-generated footage and even they lean on human editors for the final cut, not automated tools alone, because raw footage still needs a human ear for what lands.
The uncomfortable bit I'd rather not admit
I went into this expecting AI to have transformed my working week. It hasn't. It has shaved off just under two hours a week, on the specific tasks I chose carefully and edited, and it quietly cost me time on the tasks tied to my personal voice. If you're a service business, a consultant, or anyone whose value is partly your personality on the page, expect the same split. The admin gets faster. The voice work does not, and pretending otherwise is how businesses end up with AI-flattened blogs, LinkedIn posts, and emails that all read like nobody in particular wrote them.
Frequently asked questions
How much time does AI save a small business per week?
Based on my own 30-day tracked total, roughly 2 to 3 hours a week net, once you subtract editing and correction time, not the 12 to 20 hours some surveys claim. The real number depends entirely on which tasks you pick, structured repeatable tasks save the most, personality-driven content often saves the least.
Which small business tasks benefit most from AI?
Proposal and quote drafting, meeting transcription and follow-up emails, and templated admin like invoice chasing showed the clearest time savings in my tracking. Tasks that require a distinct personal voice, like blog posts and social content, showed little or no net saving once editing time was included.
Is it worth paying for multiple AI tools as a small business owner?
Only if you track where the time goes. My four subscriptions cost £71 for the month and saved just under 10 hours, roughly £7.30 per hour saved, which was worthwhile against my billable rate but would not be for every business. Track your own numbers for at least two weeks before deciding.
Why does AI sometimes cost more time than it saves?
Because editing AI output to sound like you, correct names, or fix tone often takes nearly as long as writing from scratch. This is most common with content tied to a personal or brand voice, video captions with names or regional phrasing, and any writing meant to sound distinctly human rather than generic.
Related reading: The AI Notetaker Problem: What Happens When You Record Every Client Call and AI Phone Answering for Small Business: The Real Cost, the Real Setup, and What It Does to Your Calls.
If you want the full breakdown, here is everything I know about AI marketing.
