Many SMEs are becoming digital delayers even when they are experimenting with AI. The issue is rarely enthusiasm – it’s more likely to be unclear ownership, weak data foundations, disconnected tools and a lack of practical measurement. Sean Evers, VP of Sales & Partner at Pipedrive, discusses the strategic investments in software, skills and lessons in AI preparation that smart SME leaders should pay attention to.
Analysts and reporters are increasingly sharing research about the slow nature of productivity gains that can be attributed to newly implemented AI technologies. This mirrors the conversations I have with SME leaders – they are experimenting and implementing, but they can’t say for sure that any productivity gains are consistently turning into revenue.
Don’t just take my word for it. British Chambers of Commerce data shows adoption is accelerating, with 54% of firms now actively using AI, up from 35% in 2025. But according to DSIT research, 75% of AI adopters report workforce productivity gains, but 77% have not yet seen revenue impact.
There’s evidence that many SMEs are showing enthusiasm though they are delaying real adoption by using it without ownership, data discipline or the right commercial measurement. They are accidental digital adoption and value-recognition delayers, without even realising.
Start with the business problem, not the tool
This is not the fault of the average business. The tech industry has overwhelmed many with hype, promises, ambiguity and serious FOMO. That’s not a great environment for making the best business decisions.
AI investment should begin with a specific bottleneck and problem. Let’s take one common business challenge as an example. The Sales team faces slow lead follow-up, poor forecasting, low conversion between sales stages and weak customer retention. What might a business leader do? Well, after exposure to advertising and then doing their own research of vendor claims, here’s what not to do: avoid buying any tools because competitors are doing so or because staff are already experimenting with something informally.
The smartest right first step is to go back to basics and question which workflow, customer issue or revenue leak are you trying to improve solely and right now? Then narrow in on that as the issue to research, gather internal feedback on and scope out the estimated costs and risks to the business around. Really focusing in on the problem should allow you to progress through a set of logical steps to a better-fitted solution out of many possible routes or products.
Fix the foundations before scaling AI
SME leaders have hopefully already discovered that AI will only be as useful as the data, systems and processes beneath it. If they are relying on a situation with disconnected tools, patchy CRM records and unclear handovers, then they know anything layered on top will likely only offer limited value. All those problems will remain and stymie a perfect solution.
It’s not glamorous, but leaders should ‘paint by numbers’ and start off by auditing the existing systems, their pros, cons and trade-offs, before adding any new software.
A basic readiness checklist can consist of: clean the data; ensure clear ownership; integrate the tools; agree the use cases; set employee guidance and define what a better customer experience should look like.
Measure value beyond time saved
Time saved is useful, but it is not the same as business impact. In fact, there are many velocity metrics touted that aren’t the most useful to use to really gain a deeper understanding of what’s really being delivered. What you want is to pair appropriate velocity (how fast the process moves) with value or quality metrics (what you deliver and how it performs).
In our example, put in place the ability to track whether AI improves truly important numbers for your business health, like conversion, forecast accuracy, sales cycle length, retention, customer satisfaction or revenue per customer. There are many best practices or models a leader can employ to help guide this, ranging from the very simple to something much more structured and all encompassing, like McKinsey’s Five-layer framework.
More detailed advice from your tech partners should help show how any proposed solution should connect any productivity to commercial outcomes via a clear and understandable path.
Once that’s clear, investments can be made with business risk more carefully managed. Taking the time to be forensic and logical in approach also helps small businesses avoid taking one-off or failed experiments that never become integrated habits or embedded into well-functioning workflows.
When moving forwards with implementation, it’s likely that, as with all major technology installs, culture becomes the real crux of success. Not only does leadership want to set as few and clear rules for safe and useful employee adoption, they want to show that these will change in time and that’s OK. Advice on using ChatGPT 3.5 is way out of date now – the important thing is that employees are both empowered and accountable for using their tools, and that training or fresh guidance arrives roughly in line with any new functionality or change in the power of the model/solution in question.
Every suggestion given is designed to support confidence in AI decision-making and its use in practice – but none of it is secret or hidden. Many tech providers, strategic partners, major analysts and so on, have freely available frameworks, questionnaires and best practice guides to support progress. Any SME can leverage what they have and who they know to better plan and to stress test their problems before they go to purchase. If helpful, take any lists of terms given in this article and turn them into a quick checklist to get the ball rolling.
Don’t delay – start slow, start sure
SMEs do not need to adopt every new AI tool quickly. What they do need is a practical adoption plan that links AI to business value from the jump.
One thing to bear in mind above all else though is the critical issue of trust. Even well-implemented AI will struggle commercially if it makes the internal or end-user experience feel less human, less transparent or – if a business process – more pressured. Pipedrive’s own recent research into Sales and AI uncovered that 55% of the public do not fully trust AI, while 46% say they value human connection. Faster outreach is not the same as a more trusted buying experience. Wherever it’s applied, AI should offload digital admin so users can focus on areas demanding the human touch, such as customer service and relationship building.
Winners will combine enthusiastic experimentation with very structured discipline. That will look like investigating and understanding clearly stated problems, inputting in pre-cleansed and trusted data, setting accountable owners – and of course the right metrics that show real commercial progress in that particular operation.
