# AI ROI for Small Business: Apollo Sees "Zero Payoff" Outside Big Tech. How to Measure Yours - Canonical: https://spiai.ai/blog/ai-roi-small-business - Published: 2026-11-04 - Author: SpiAI - Description: AI ROI for small business: why Apollo found no payoff outside big tech, and a simple way to time, price and prove AI's return in your own team first. --- On October 24, Apollo Global Management's chief economist Torsten Slok published a short note with a blunt title: no signs of AI payoff outside the tech sector. The next morning [Walter Bloomberg's account summarized it](https://x.com/DeItaone/status/2114301898060611618) for more than 300,000 viewers: "AI HAS DELIVERED 'ZERO PAYOFF' OUTSIDE BIG TECH... Revenue per worker and profit margins are rising for the Magnificent Seven but remain flat for the S&P 493." The most useful replies were not the doom takes. [@futureproofmeai added a caveat](https://x.com/futureproofmeai/status/2114706270733336917): "That's an aggregate read, not proof of zero gain at every firm." And [@justfyi turned it into an instruction](https://x.com/justfyi/status/2114941506012504083): "Prove it in your own team first: Pick 1 task. Time 5 runs without AI, 5 with. No clear time saved, no rollout." That is the right response for an owner-led service business. You cannot learn your AI ROI from the S&P 500. You can measure it on one workflow in two weeks. ## TL;DR - Apollo's Slok found revenue per worker and margins rising for the Magnificent Seven and flat for the other 493 S&P 500 companies, with no sign yet of AI-driven productivity gains in the broader group. - That is a macro finding about large public companies. It does not tell you whether AI pays off in a 20-person firm, but it is a good reason to measure instead of assume. - Measure ROI on one workflow: time it before and after, price the time, subtract tool and review costs, and count errors. - Include the review time. AI drafts, your team approves, and the approval is part of the cost. ## What did Apollo actually find? Slok's note compared two groups of S&P 500 companies: the seven largest tech firms and everyone else. For the Magnificent Seven, revenue per worker and profit margins have climbed. For the other 493, both have been roughly flat after inflation, and he sees no acceleration in productivity growth that would signal AI paying off yet. He also argued that AI spending is concentrated in a small, largely venture-backed group, so the broad economic benefits stay limited until mainstream companies turn AI into higher revenue and margins. Two things are worth keeping in mind. First, this is aggregate public-company data, measured in revenue and margins, which move slowly. Early efficiencies may not show up there for a while. Second, large companies adopt AI through long procurement and change programs. A small business can try something on Monday and know by the end of the month whether it worked. ## Why small businesses should measure, not assume The Apollo note is a useful corrective to vendor decks claiming 40% productivity gains. It is also not a reason to ignore AI. The honest position is that nobody, including us, can tell you your ROI without measuring your own work. Most owners skip measurement for understandable reasons: they are busy, the tool is cheap, and it feels faster. "Feels faster" is exactly how a business ends up paying for five AI subscriptions and seeing no change in the numbers that matter. ## How do you measure AI ROI in a small business? Use one workflow, two weeks and a simple worksheet. 1. **Pick one repeatable task.** Choose something that happens at least a few times a week and has a clear finished state: writing quotes, answering a type of customer email, entering invoices, preparing a weekly job report. 2. **Time it without AI.** Have the person who normally does it record start and finish time for five runs. Note any errors or rework. 3. **Time it with AI, including review.** Five more runs, with the clock running through the review and fixes, not just the draft. This is where most claimed savings shrink. 4. **Price the time.** Multiply minutes saved per run by runs per month by the loaded hourly cost of the person doing it. 5. **Subtract the real costs.** Subscriptions, setup time, the reviewer's time and any increase in errors that reach customers. 6. **Look at quality, not just speed.** Count mistakes, customer replies and rework in both rounds. 7. **Decide.** If the saving is clear and quality held, roll it out to that workflow. If not, stop or change the approach. Then move to the next task. ## A worked example of the math Here is the structure, with round, hypothetical numbers to show how the calculation works. Use your own. | Line | Without AI | With AI | |---|---|---| | Minutes per quote, including review | 40 | 22 | | Quotes per month | 60 | 60 | | Hours per month | 40 | 22 | | Loaded cost at $45 an hour | $1,800 | $990 | | Tool cost per month | $0 | $50 | | Monthly total | $1,800 | $1,040 | | Monthly difference | | $760 saved | The same worksheet often comes out the other way. If review takes nearly as long as writing from scratch, or errors increase, the saving disappears. That is a useful result too. It tells you to stop paying for that use. ## Where AI usually pays off first in service businesses The clearest returns tend to come from work that is repetitive, text-heavy and easy to check: - First drafts of routine customer replies and follow-ups. - Pulling data out of emails, PDFs and forms into your systems. - Summarizing long threads, calls and job notes for handoffs. - Preparing recurring reports from data you already have. Returns are weakest where the work is rare, high stakes or depends on context the AI does not have, such as pricing unusual jobs or handling a sensitive complaint. Those are better treated as AI-assisted, with a person firmly in charge. ## Why ROI often disappears after the pilot Even when a test looks good, savings leak away in predictable ways: - **Review creep.** People start rechecking everything twice because they do not trust the output, and the saving evaporates. - **Tool sprawl.** Each person picks their own AI tool, nobody shares what works, and costs multiply. - **No change in the work.** The task gets faster but nothing else changes, so the freed time is not used for anything that earns money. - **Errors reaching customers.** A few bad drafts that slip through can cost more than months of saved time. The fix for most of these is the same: a defined workflow, a named reviewer, and a monthly look at the numbers. AI drafts, your team approves, and someone owns whether it is still worth it. ## What else should you measure besides time? Time saved is the easiest number to capture, but it is rarely the whole return. Three other measures are worth a line on the worksheet: - **Response speed.** How long a lead or customer waits for a first reply. Faster replies can win work that slower ones lose. - **Capacity.** Whether the same team can handle more jobs, quotes or clients without overtime or a new hire. - **Consistency.** Whether every customer gets the same quality of follow-up, not just the ones who catch your best person on a good day. Track these alongside the hours. Revenue and margin are what Apollo measured, and they are what you should eventually see move too. ## Verdict Verdict: Hold on broad AI spending, Buy measured pilots. Apollo's finding is a reason to stop assuming AI pays off, not a reason to ignore it. Pick one workflow, time five runs without AI and five with, include review time and errors, and price the difference. Roll out only what clearly saves time without hurting quality, then measure again in a month. That is how a small business gets the payoff the S&P 493 has not shown yet. **Want to know where AI would actually pay off in your business?** In a free 30-minute assessment we pick the workflow with the clearest return, set up a simple before-and-after test and show you where the review step goes. [Book a free assessment](https://spiai.ai/) ## FAQ ### What is AI ROI for a small business? It is the value AI creates in your business, usually time saved or revenue gained, minus what it costs, including subscriptions, setup and the time people spend reviewing its output. ### What did Apollo say about AI payoff? Apollo chief economist Torsten Slok wrote on October 24, 2026 that revenue per worker and profit margins are rising for the Magnificent Seven but flat for the other 493 S&P 500 companies, with no clear sign yet of AI-driven productivity gains outside tech. ### Does Apollo's finding mean AI does not work for small businesses? No. It is an aggregate view of large public companies using slow-moving financial measures. It is a good reason to measure your own results rather than assume them, not proof that AI cannot help a small business. ### How long does it take to measure AI ROI? About two weeks for one workflow. Time five runs without AI and five with AI, including review, then compare cost and quality. ### What costs should I include when calculating AI ROI? Tool subscriptions, setup time, the time people spend reviewing and fixing AI output, and the cost of any errors that reach customers. Leaving out review time is the most common mistake. ### Which tasks give the fastest AI return? Repetitive, text-heavy tasks with a clear finished state, such as routine customer replies, data entry from documents, summaries and recurring reports. ### What if the test shows no savings? Stop paying for that use or change the approach. A negative result is useful because it keeps you from spreading a tool across the business that does not pay off.