Only 19% of businesses are actually getting value from their AI investment. Here's how to make sure you're one of them 👇
That 19% figure comes from Microsoft's 2026 Work Trend Index, based on trillions of anonymised Microsoft 365 signals and a survey of 20,000 knowledge workers. It's not an outlier either. McKinsey's parallel research found 88% of companies report regular AI use in at least one business function, yet only 39% can point to any actual bottom-line impact from it. BCG puts a sharper number on the same problem: 60% of companies globally aren't generating material value from AI despite serious investment.
In other words, adoption has outpaced impact almost everywhere. The real challenge now isn't whether to use AI. It's deciding where it creates measurable business value and how to scale it responsibly, so your organisation ends up on the right side of that 19% line rather than the wrong one.
That shift is one of the clearest signals coming out of Microsoft's FY27 strategy too, a strategy our own CTO, James Pearse, saw unveiled first-hand at this year's Microsoft MCAPS event. The focus is moving from experimentation to execution. Organisations are being asked to demonstrate outcomes, put governance in place, and prepare for AI becoming a standard, operational part of the business, not a side project.
This is where most businesses should start, not with another pilot, but with a clear-eyed look at where AI can move the needle on productivity, efficiency, or customer experience.
Focus on the areas where teams are already spending significant time on repetitive, manual, or administrative work. Useful questions to ask internally:
The organisations seeing the most success aren't the ones using the most AI tools. They're the ones solving real business problems with the tools they have.
Example: A finance team spending three days a month manually reconciling supplier invoices isn't a candidate for a flashy AI pilot. It's a candidate for straightforward automation with a clear, measurable time saving. That's a better starting point than a broader "innovation" project with no defined output.
Action step: Pick one process, run a 90-day pilot with a defined before/after metric (hours saved, error rate, turnaround time), and use that as the template for the next one. Value compounds faster when it's proven, not assumed.
This is often where organisations get caught out. As AI becomes embedded in everyday operations, it gains access to business data, applications, and processes, which makes governance just as important as innovation.
Worth asking:
Strong governance doesn't slow AI adoption. It's what allows organisations to scale it with confidence. The businesses making the fastest progress are typically the ones that invested in security, identity, and governance before expanding deployment, not after.
This matters more than most leaders assume. ISACA's 2026 research found that a quarter of organisations have no active AI policy at all, and more than half of professionals say they don't know how long it would take to shut down an AI system in the event of a security incident.
Example: An employee pastes a client contract into a public AI tool to summarise it quickly, with no malicious intent, but with sensitive data now sitting outside the company's control. This is the everyday reality governance needs to catch, not a hypothetical edge case.
Action step: Before expanding AI access, confirm three things: who can access what data, whether that activity can be logged and audited, and whether someone owns the response plan if something goes wrong.
As adoption grows, visibility becomes critical. Many organisations now have multiple AI tools in use across departments, often adopted by individual teams without central oversight.
That creates risk around:
Leaders need confidence that they know where AI is being used, what information it can reach, and whether that usage aligns with company policy. Without that visibility, it's nearly impossible to manage risk or prove value, and this is often where organisations realise their AI adoption has outpaced their governance strategy.
Example: A marketing team adopts a free AI writing tool for social copy. A sales team separately signs up for an AI meeting-notes app. Neither is reported to IT, and six months later, nobody in the business can say with confidence how many AI tools are actually in use, or what data has passed through them.
Action step: Run a simple internal audit, even a short survey asking teams which AI tools they use day-to-day, before investing further. You can't govern what you can't see.
Much of the AI conversation so far has centred on tools like Copilot. The next conversation is about agents, and it's already starting.
As organisations automate more workflows and processes, AI agents will play a growing role in day-to-day operations. The challenge isn't deploying them. It's managing them.
Key questions for leadership:
This isn't just a technology question. It's an operational one. Business leaders, IT, and security teams all need a shared understanding of how agents will be governed and controlled as usage scales.
The scale of the shift is significant: Microsoft's 2026 Work Trend Index recorded a 15x year-over-year increase in active agents across the Microsoft 365 ecosystem, rising to 18x in large enterprises. Agent adoption is no longer experimental. For many organisations, it's already operational, whether governance has caught up or not.
Example: An AI agent is given permission to draft and send routine customer emails to speed up response times. Without a clear escalation rule, it also starts responding to complaints and refund requests, situations it was never scoped to handle.
Action step: Before deploying any agent, define its scope in writing: what it's allowed to do, what it must escalate to a human, and who reviews its output. Treat this the same way you'd treat onboarding a new employee, with clear responsibilities from day one.
This might be the most important question of all.
For years, many organisations measured AI success by pilots launched, licences purchased, and technologies deployed. That's changing. The organisations getting real value from AI are now measuring:
Deployment is only the starting point. The real measure of success is the value AI creates for the business, and if initiatives aren't tied to measurable objectives, it becomes very difficult to justify further investment. McKinsey's research puts a number on this gap directly: while 88% of organisations report regular AI use, only 39% can attribute any measurable EBIT impact to it, and just 23% say they're actually scaling an agentic AI system anywhere in the business.
Example: A business rolls out an AI tool to every department because leadership wants "AI adoption" to look strong in a board update. Twelve months later, usage is high but nobody can say what it actually changed: no time saved, no cost reduced, no customer metric improved.
Action step: For every AI initiative, define the outcome metric before rollout, not after. If you can't name what "success" looks like in numbers, that's a signal the initiative needs sharper scope before it gets resourced further.
Start with an honest look at where your organisation actually stands:
Answering these honestly is the clearest path from AI experimentation to measurable business success.
Not sure where your organisation stands against these five questions? Book a free AI readiness assessment with our team, and we'll help you identify your highest-value opportunity, check your governance foundations, and build a scaling plan with metrics attached from day one.