AI Adoption for Mid-Market Companies: Start with Wins, Not Resistance
AI for Mid-Market Leaders Series – Part 3 of 4
In Part 1, we looked at why AI strategy in mid-market companies has to start at the leadership table, led directly by the CEO. In Part 2, we explored the risk of waiting too long to learn, and why delaying adoption can be far more expensive than moving early.
There’s a third point that builds directly on both, and it gets at the question many CEOs are now asking: How should we implement AI in our company?
In most strategic initiatives (new systems, new processes, new operating models, etc.) senior management defines the scope and objectives upfront. Then change management reverse-engineers adoption: who needs to do what differently, and how do we help them get there?
AI flips that model on its head.
Why the Usual Approach Doesn’t Work
With AI, the value is not fully known upfront. The organization learns what’s possible while it adopts. Scope and priority are shaped by the experience of using it, not by a top-down end-state defined in advance.
The lack of definition around what AI can do is far greater than, for example, an ERP system. And that uncertainty creates a different kind of reaction.
When a new standard platform is announced, say Oracle, SAP, Microsoft Dynamics or Workday, employees may get nervous, but they can still picture how their job might change. They know some manual work will be automated, and they can assess the scope of that change for themselves.
With AI, there is no such familiarity. And humans are much more afraid of the unknown than they are of a familiar threat.
For some employees, that shows up not as the usual anxiety that comes with change, but as deeper apprehension: What does this mean for my role if even the leaders don’t know yet? Will I work myself out of a job and train my AI replacement in the process?
That dynamic isn’t helped by high-profile examples of AI being used explicitly to reduce staff, many of which were later reversed when the promise didn’t match reality. Several of those implementations ended in costly failure, and companies had to rehire the people they released. There are lessons there about overreach and understanding what AI is actually capable of today.
AI Change Management Starts With Anticipated Wins
The entry point for AI adoption is not an energized vision document from the executive suite; it’s the quiet, personal moments of “What’s in it for me?”
AI adoption works when people on the ground get involved in improving their own work using AI. They get to evaluate, in practical terms, what they can gain:
fewer tedious administrative tasks
faster access to information
easier documentation
better first drafts
clearer summaries
fewer meetings spent “recapping”
more time for client work or strategic thinking
Those are small wins, but they matter. They build trust. When individuals experience these benefits, they become advocates for others. They begin to define the vision, scope, and objectives themselves—and they retain a sense of control over the pace of adoption.
Momentum spreads from anticipated wins to experienced wins.
This is why the CEO’s main job is not to “manage resistance,” but to help people see where the likely gains are for them, and create space for them to try it for themselves.
Experimentation Comes Before the “Change Plan”
A second inversion in AI change management is the role of experimentation.
In traditional change programs, experiments come late, as pilots to validate a design. In ERP implementation, there is a clear sequence of testing phases (functional, system, user acceptance) and, in more mature organizations, a sandbox environment for practice.
In AI adoption, experiments ARE the design. Individuals test AI in their work environment, learn lessons, make adjustments, keep experimenting, and eventually arrive at a workable pattern.
Done well, this approach produces exactly what AI strategy needs most in mid-market organizations:
confidence
clarity
a common language
internal champions
and practical knowledge of what works and what doesn’t
That may be more than your competitors have for quite some time!
A Hint of What You Should Do Next
The CEO’s job in this phase is not to predict the end-state, but to create the environment where leaders and teams can learn. That itself reduces pressure.
In Part 1, the work was to align your leadership team on what AI is for in your business, in concrete terms.
In Part 2, the work was to surface risks and turn them into action items in your AI Blueprint.
Now it’s time to bring people into the work.
CEO Hint #3
Bring a small group from your management team and employees together to identify one or two practical areas where AI could reduce friction today, and design a 30-day experiment to test it.
Practically, that means:
supporting small-scale experimentation
giving permission to learn openly
celebrating early wins
treating feedback as input to strategy
keeping the focus on business value, not technology
anchoring everything in the AI Blueprint you drafted as a team
This is where the Blueprint starts to come alive: through real work, not theoretical models.
In the next newsletter, I’ll share a simple way to run these experiments, what to measure, and how to pull those lessons back into your Blueprint so the whole organization benefits.