Over the last 12 months or so, it’s been almost impossible to escape the hype around AI. Every fire industry or marketing tech conference I attend, or my LinkedIn feed, is full of claims that AI will reshape the way we all work (or already has!). Yet in my own experience — working in a larger-side SME with international operations — I’ve found it difficult to assess whether companies are really changing their ways of working with proven ROI, or whether we’re mostly still at the stage of experiments and pilots. What is hype, what is reality?
In the absence of clarity, one alternative is to look at the most credible external research, some good old-fashioned (quite ironic) desktop research. Over the last 12 months, a series of major reports from McKinsey, Deloitte, IDC, PwC, Gainsight, and MIT Sloan/BCG have all been published. These reports aren’t the complete picture (I am sure I have missed some other reports) — AI moves so quickly that better insights may be just around the corner — but taken together, they offer a useful snapshot of where we are today. Apologies in advance, this article contains quite a lot of numbers but I have still kept it quite high-level. Obviously, if you are interested, reading these reports in full is very informative and insightful.
My starting point was naturally through the lens of customer experience. That’s why the first report I read this year was the Gainsight State of AI in Customer Success. From there, I broadened out to see how AI adoption looks across sales, marketing, operations, and the enterprise as a whole. The findings are fascinating, especially for SMEs like mine.
Table of Contents
Starting in Customer Experience: The Gainsight Report
Because I lead in customer experience at my company, my personal interest was always in how AI could elevate CX and improve processes. The Gainsight report gave me a reality check: AI adoption in customer success is U-shaped. The smallest companies (under $5 million in revenue) and the largest enterprises (over $250 million) are leading adoption — 63% and 55% respectively. The mid-market ($20m–$250m) lags behind at just 39–42%.
This struck me as pretty relevant for my work and company. In my own experience, I’ve seen that smaller (and younger) companies can move quickly with AI because they don’t face layers of compliance and legacy systems. Large enterprises, meanwhile, can throw budget and teams at experimentation and running pilots. Mid-sized companies like mine (possibly) get stuck in between: too complex to be without processes, but without the organisational set-up to scale AI adoption.
Where are companies seeing most value? According to Gainsight, the strongest AI use cases in customer success are onboarding (58%), product engagement (75%), and increasingly in data analysis and pre-call preparation. In other words, areas where repetitive data work slows down the team, and where AI can add immediate leverage. To be honest, this didn’t come as a surprise, as AI use cases cover CX/Customer Experience extensively, and there are plenty of case studies to read about already.
Arguably, this makes CX a good starting point. If you want to prove AI ROI quickly, start in CX processes where there are clear customer touchpoints, measurable KPIs, and a strong need for better insights.
Broader Lessons: McKinsey and Deloitte
The next two reports widen the view beyond CX to the whole enterprise.
McKinsey’s State of AI 2025 survey found adoption is now mainstream: 78% of organizations use AI somewhere. Yet only 21% have redesigned workflows to embed it, and only 28% report CEO-level oversight of AI governance. That disconnect explains why over 80% of firms still haven’t seen a tangible EBIT uplift.
I currently work for a medium-sized company, but I have also had the benefit of working for large international corporations. I’ve seen how governance and workflow matter. It’s not enough to have a new tool or technology — you need to redesign the process around it. AI is no different. The McKinsey data makes the point: workflow redesign and senior ownership are the two strongest predictors of value.
Deloitte’s State of AI in the Enterprise comes to a similar conclusion. Adoption is high — 79% of firms now fully deploy three or more types of AI — but outcomes lag. In fact, the proportion of “Underachievers” has risen to 22%. What separates leaders from laggards isn’t technology but post-launch support, model lifecycle management, and ROI tracking.
ROI Potential vs Barriers: IDC and PwC
IDC’s late-2024 study found the average ROI from generative AI was 3.7×. That’s a compelling number, and it suggests that AI can more than pay its way. But IDC also found that the biggest barriers senior management worries about are data security (41%), data governance (41%), privacy (39%), and regulatory compliance (39%). Looking at these barriers, these are all potential reputational and financial risks and shouldn’t be taken lightly.
For mid-sized firms, this is critical. We may not always have the budgets for dedicated risk and data teams (it is often covered by non-specialists and not as the main job), but without at least a minimum governance model, we won’t get past the pilot stage.
PwC’s AI Predictions 2025 drives this point home: companies that make AI core to business strategy are 2.6× more likely to see value. In my own work, I’ve argued that customer experience can’t be an afterthought in AI adoption — AI has to be part of the strategy if it’s going to change outcomes. (If you are interested, have a look at my other articles)
The Learning Company Edge: MIT Sloan / BCG
Finally, the MIT Sloan/BCG research makes a different point: AI isn’t just about tools — it’s about how organizations learn. Their survey found that “Augmented Learners” (about 15% of firms) are 2× more prepared for disruption and 1.6–2.2× more confident handling uncertainty than others.
One stat stood out: in 2023, 61% of leaders said AI was core to strategy; by 2024, that had dropped to 38%. That drop reflects hype fatigue — but it also shows why many firms stall.
For mid-sized companies, the lesson is powerful: even if we can’t invest at the same level as the big players, we can create learning checkpoints. In my own team, we’ve used AI tools for multilingual customer training for our new Learning Management System (LMS). The key wasn’t the tool itself — it was setting up the discipline of testing, reviewing, and sharing what worked. That kind of lightweight “learning mindset and checkpoints” makes all the difference.
If You Had to Start Small…
Across all six reports, the same answer emerges: if you’re proving ROI, start small in functions where data quality is manageable, processes are repetitive, and KPIs are clear. For most SMEs, that means:
- Customer onboarding and engagement (CX)
- Sales & marketing personalization
- Service operations (product support, chat)
- Internal reporting and knowledge bases
Some of these are not big projects; they’re practical AI first use cases to prove ROI. Importantly, if successful, they build credibility and momentum for bigger moves.
Why Mid-Sized Companies Lag — and How to Catch Up
So why do mid-sized companies struggle, when small and large firms are ahead? The reports highlight four gaps:
- Governance gap — CEOs not directly involved, risk processes immature.
- Talent gap — fewer in-house AI/data experts. (Closing this talent gap and embedding a continuous learning culture to support AI will be a future article)
- Integration gap — AI adoption isn’t embedded into redesigned workflows.
- Clarity gap — lack of KPI baselines, so ROI can’t be demonstrated. To go beyond AI pilots, you need to prove ROI!
And why do some SMEs get ahead?
- Startups benefit from agility.
- Larger companies and corporations benefit from resources.
- But mid-sized companies can benefit from focus: fewer layers of bureaucracy than large corporations, and more stability than a startup.
In my company, we’ve already seen that having a more mature governance structure (thanks to our international operations and ownership model) is actually a strength. It gives us a base to adopt AI responsibly without fear of regulatory or reputational damage. Not every SME has that — but those that invest in lightweight governance, training, and a learning mindset and checkpoints will be the ones that move up the curve in my opinion.
Conclusion
The last 12 months of AI hype have been overwhelming at times (again, I have written an article about this topic). It’s often hard to separate pilot buzz from proven ROI. But looking across six of the most credible reports gives a clearer picture. Adoption is widespread, but real value is patchy. Larger firms are scaling faster, while mid-sized companies often lag.
Yet that’s not a reason to be pessimistic. If anything, it’s an opportunity. SMEs can lean on their governance maturity, start small in CX and service operations, and create learning cultures that turn AI from experiment into embedded capability.
As I wrote in my earlier articles on GEO vs SEO and AI in Fire Safety CX, the real challenge is less about the tools, and more about how we change the way we work. AI isn’t just about products — it’s about customer experience, workflows, and culture. And that applies just as much to mid-sized SMEs as it does to global giants.
Further Reading
- McKinsey & Company — The State of AI: How Organizations Are Rewiring to Capture Value (2025)
- Deloitte — State of AI in the Enterprise, 5th Edition
- IDC / Microsoft — The Business Opportunity of AI: Generative AI Delivering New Business Value and Increasing ROI (2024)
- PwC — AI Business Predictions 2025
- Gainsight — The State of AI in Customer Success 2024
- MIT Sloan Management Review & BCG — Artificial Intelligence and Business Strategy: 2024 Global Executive Study
