Does using more kinds of AI mean a business is getting more value?
A business using text AI throughout its customer correspondence could count as using one AI technology. Add an occasional image generator and the count rises, even if nothing useful changes for the staff. That's an illustrative comparison, not a finding about particular businesses. The ONS July 2026 analysis of UK business AI use tracks the number of technology categories used by adopting businesses with 10 or more employees. The average rose from around 1.4 to 1.6 since late 2023. These are categories, not subscriptions or individual apps. ONS calls the count an imperfect proxy for intensity: not every technology is relevant to every business, and the count doesn't show how extensively each is used. I wouldn't turn that measure into a shopping target. A repair shop doesn't need image generation or robotics to prove that its text assistant is earning its keep. Nor does using just one category establish that it's doing well. Before adding another tool, I'd ask what the current one has let someone stop doing. If the receptionist no longer rewrites routine collection messages, that is a benefit worth checking against the time spent correcting drafts. A second category should solve another actual problem, not improve an adoption score. This is a buying judgment drawn from the report's measurement limits, not evidence that fewer tools always work better. What would justify adding a second AI tool in a small business that already has one useful application?
Comments
A second tool could earn its place by replacing the first. I'd give the challenger the same ordinary job and ask whether it's easier to get an acceptable result—not find it a new chore just to justify keeping both. Before the trial, agree which subscription would go if it wins. Otherwise the incumbent gets to stay because it's familiar and the newcomer gets to stay because it's promising. A small business can end up paying indefinitely for a comparison nobody finishes. That's a buying rule I'd use, not a finding from the ONS report.
I'd check whether there's a separate subscription to cancel. If the old AI feature comes with software the business still needs for bookings or accounts, replacing its answers may not remove any spend. The challenger then has to justify an extra bill, even if it wins the same-task comparison. I'd have the person paying the invoices confirm what would actually disappear before describing the switch as a saving. A better tool can be worth paying more for; it just needs to be bought on those terms.
I'd try that comparison on the device where the job happens. For a shop owner replying from a phone, start with the customer message in the usual inbox and finish with a usable reply back there. Include finding the attachment and moving the answer across, not just generating it. A tool might write a better reply on a laptop yet be awkward enough on the phone that the owner leaves it until after closing. That's a trial I'd suggest before switching: can this person finish the job at the counter, or have we given them homework?
The report's newer question about extensive use gives a different view from counting technology categories. Among AI-using businesses with 0 to 9 employees, 17% reported extensive use, compared with 9% of those with 250 or more. ONS cautions that these are early indications and that larger businesses are more likely to be unsure about the extent of use. So I wouldn't turn that comparison into a ranking of who's ahead, either. But it is a reason not to read a low category count as evidence that AI barely touches the working day. For a very small firm, I'd ask which ordinary jobs already depend on it before recommending another tool. Extensive use still doesn't tell us whether those jobs got easier.