New benchmark data out this year puts a hard number on something a lot of business owners have quietly suspected: adopting AI and benefiting from AI are two very different things. Around 70% of UK small and medium businesses now report using AI in some form, yet only 31% see a positive return on that investment, and roughly three in four adopters report no meaningful change in revenue at all. Meanwhile 85–91% of organisations say they’re increasing their AI budgets regardless. Spend is going up. Payoff, for most, isn’t following.

The single biggest reason cited is a skills gap — over 60% of SMEs point to it as their primary barrier, and separate research suggests the vast majority of UK organisations have some form of AI skills shortfall. Only around 14% of UK workers have had any formal AI training. Put simply: businesses are buying tools faster than they’re building the ability to use them well.

Why “having AI” isn’t the same as using it

The pattern behind the weak ROI numbers is a familiar one. A business signs up for a Copilot licence, or asks staff to “have a go” with ChatGPT, and treats that as the AI strategy. There’s no clear list of what tasks it should actually be used for, no time set aside to learn it properly, and no way of checking afterwards whether it made anything faster or better. The tool arrives; the habits and workflow around it don’t. Research backs this up directly — staff who get more than 81 hours of AI training a year report gaining nearly two extra working days a week in productivity, while untrained staff mostly just add another app to their list.

The gap isn’t a technology problem. It’s an adoption problem, and it’s entirely fixable with the kind of focus a lot of SMEs simply haven’t had time to give it yet.

Three questions worth asking before spending more

Before renewing or expanding any AI subscription, it’s worth answering three plain questions. First: can you name the two or three specific tasks in your business where AI use is actually expected, rather than just “available”? Second: has anyone had proper time — not five minutes between meetings — to learn what the tool is good at and bad at? Third: do you have any way of knowing, even roughly, whether it’s saving time or just adding a step? Most businesses currently seeing no ROI would answer no to at least two of these.

If you’re not sure where to start turning general-purpose AI access into something that actually moves the needle for your specific business, that’s precisely the gap ApplyAI is built to close — helping SMEs work out which tasks are worth automating, train staff properly around them, and measure whether it’s actually working.

Don’t mistake budget growth for progress

The fact that 85–91% of organisations are increasing AI spend regardless of whether it’s working is worth sitting with for a moment. In most areas of business, spending more without checking the return would be treated as a warning sign, not a strategy. AI seems to have earned a temporary exemption from that scrutiny, partly because the technology is moving fast enough that “wait and see” feels riskier than “spend and hope.” For a large enterprise with money to experiment, that’s a defensible bet. For a small business watching every pound, it’s a much harder one to justify without some way of checking whether last quarter’s spend actually did anything.

None of this is an argument against using AI — the businesses genuinely seeing productivity gains are proof the tools work when they’re used properly. It’s an argument for treating adoption as a project with a defined scope and a check-in point, rather than a subscription you renew out of habit because everyone else seems to be doing it too.

The takeaway

Buying AI licences isn’t the hard part anymore — most businesses have already done that. The return comes from deciding exactly where it’s used, training people properly around it, and checking honestly whether it’s paying off. Spend less time worrying about which tool to buy next and more time on whether the one you already have is actually being used well.