A woman working from home with back pain
getty
quickbooks 2026 The AI Impact Report includes a survey of 34,000 small businesses across the United States in partnership with the University of Chicago. On paper, this data is impressive. The percentage of respondents who say they use AI at least regularly increased to 77%, up from 48% in mid-2024. Furthermore, 41% of those surveyed reported increased revenue and 74% productivity improvement through the use of AI. Figures like these reveal one side of the story regarding the use of AI for small business purposes. However, the “fine details” of this study reveal another aspect. When measuring the use of AI in small businesses, the most important question is not whether companies are using AI, but whether they have the ability to measure whether AI is successful.
When researchers asked how companies (companies) measure the improvement from these benefits, the answers became very vague. More than 50% said their overall feeling was that their business had improved. Less than half of respondents reported tracking specific metrics. Productivity numbers are based on self-reporting, not time surveys. “AI” returns were based on correlations rather than controlled measurements. This area is also where most small and medium-sized businesses currently reside. Such, Gartner predicts By 2026, global AI spending is expected to reach $2.52 trillion. Much of this money is being spent without any real way to measure the return.
Why most companies fail to measure AI results
The AI should start at a point after the measurement problem. Many small businesses lack baseline data (how long it takes a human to complete a task) to pass to AI. Without this data, it is impossible to measure what percentage of time AI saves in performing tasks. Similarly, without tracking the conversion rates of previous human-generated campaigns, there is no basis for comparison with AI-generated campaigns. There is no baseline. Therefore, no measurable comparison of improvement is possible.
The second question is how to assign credit (attribution) for changes if a company uses an artificial intelligence system to email customers and hires a new salesperson during the same period. Employers cannot determine whether it has been used or not. Increase revenue with AIhiring new salespeople, seasonal changes, or market changes unrelated to technology. We previously discussed the AI investment gap. The fundamental problem remains the same. Companies buy systems but don’t have a definition of a successful outcome before they start.
55% were predicted by. There are so many Foresters. (55%) AI projects fail to achieve their intended goals. This does not necessarily mean that the initiative failed because of the technology, but simply that there was no way to measure success at the time the project was launched. Even a successful tool can look like a failure if you don’t clearly establish how “success” is defined before using the tool. Similarly, a project may appear successful with respect to its metrics, but as long as the metrics are wide-ranging and include chance events, they add little or nothing to the organization.
5 numbers every small business should track
5 ways to measure AI ROI It was published last year and is still going strong. But QuickBooks data shows that many companies don’t track them. Here’s a simplified AI measurement framework that any small business can get started with this week.
5 numbers every small business should track
Source: Business AI Research Institute
First, you need to determine how much time each task requires. What are the three most commonly used AI-based tasks? Before using AI, record the time (in seconds) required to complete the same task. Record your time using a real stopwatch or timer. Then, calculate the time saved by comparing the two times and multiplying the time difference by the hourly rate. For example, if a task that previously took 1.5 hours now takes only 0.33 hours using AI, this is a savings of 1.17 hours. If a business owner earns $100 per hour, that’s $116 in time savings per case. Record these numbers weekly.
Next is the output quality. Track how much, if any, edits are required for each piece of content generated by artificial intelligence. If 80% of your AI-generated content requires only minor adjustments, you can get value from your AI tools. If 80% of your AI-created content requires a complete rewrite, and it takes time to do so (with no net time savings), you’ve simply changed your workflow without increasing productivity.
Third, revenue per AI-supported activity. Compare revenue from campaigns written using AI to campaigns written by humans for the same number of customers and time period. The two sets must contain comparable customer lists, offer types, etc. The amount of revenue you get from an AI-drafted campaign compared to a human-drafted campaign is your AI marketing ROI. If there is no difference in revenue, then using AI has saved time in creating email marketing campaigns, which is a completely different value than increased revenue.
Fourth is the error rate. When using AI for tasks such as processing data entry, bookkeeping, and creating reports, monitor how often errors need to be corrected. Compare your current error rate to the error rate when you run these tasks manually. If you see fewer errors than before, the AI may be improving its level of accuracy. Conversely, if the number of errors stays the same or increases, and the overall workload also increases, the AI will generate additional errors and the review will take longer.
Fifth, the cost and value of the tool. Calculate the cost of all AI subscriptions your company has. Then divide that number by the documented value (from previous measurements) produced by those subscriptions. For example, if your company spends $500 per month on AI, and you measure an additional $2000 per month as a result of using these tools (time saved or increased revenue), you can clearly see your ROI. However, if you can’t identify that your company spent $500 a month and generated at least a certain amount of dollars in return, you should do one of the following: Correct the measurement method Or the tools you use, maybe both.
The dangers of recruitment without measurement
American Chamber of Commerce According to a report, 58% of all small businesses are currently using generative artificial intelligence (AI), and 93% expect their business to grow in 2026. This is great news for business optimism. However, if you can’t measure results with some degree of certainty or evidence, you’ll continue to spend money on unproven tools. What’s worse, each new tool claims to solve a problem that the previous tool may not have addressed, making you want to buy even more tools.
“trust” McKinsey’s AI in 2026 The Trust Survey shows that companies can invest in building trust infrastructure (including measurement and governance) for their AI projects, allowing them to build projects faster and having fewer compliance issues. This idea also applies to small and medium-sized enterprises. Small business teams can build “trust” in AI by showing it how much time it has saved or how much profit it has made.
In addition to restaurants, other businesses are reporting similar results. For example, one restaurateur said he saw a 20% increase in reservations after using an AI-based reservation service. He attributed this increase in bookings by comparing bookings from May this year to May last year. However, if booking comparisons are made without taking into account differences such as changes in customer review activity on websites such as Google, the number of dining areas (outdoor vs. indoor), or even the menus offered at the time of comparison, the data collected will not separate variables and will result in anecdotal narratives rather than evidence-based data.
Small businesses don’t need to hire a data science team to properly measure artificial intelligence. What they need is discipline. See what metrics are being measured. When do those metrics start? Check to see if any changes were made to coincide with the start of your AI email campaign. You probably hired a new salesperson in the same month. Measure each separately. Track revenue from deals generated by sales reps in one column, and revenue generated from AI campaigns in another column. Separation provides clarity. Don’t just throw all your metrics into one group and hope it gives you a better picture.
QuickBooks’ 77% adoption rate is true. Many of these companies will realize productivity gains from AI. However, “a lot” is not a business plan. The companies that can take advantage of AI over the next 12 months will be those that stop making educated guesses (i.e., “this will help”) and start doing the math. Identify three things you need to automate. Record how much time you spent. Rate the quality compared to before. Calculate what percentage of your total revenue comes from work done by AI. Calculate the costs associated with using AI. These numbers indicate whether the AI is working or not. The same number also identifies where to fix/fix the issue.
Run measurements for 30 days. Positive results will provide the evidence needed to decide on further investment in this area. If you don’t see any improvement, that’s a reason to evaluate ways to improve your strategy before wasting another dollar. You’ll understand in either case. Most businesses today operate on guesswork. Do something different than most people.
