The 90-Day Plan:
How to Book Your First AI Results in Your SME

From diagnosis to action: the approach that works without overwhelming your organisation

You’ve spent the past few weeks thinking about where you stand.

You’ve identified a pain point. You’ve thought about governance, ownership, and tool selection. You know which employee is already working enthusiastically with AI and how to create buy-in.

Now there’s one question left: what do you do on Monday?

And: what do you want to be doing differently in 90 days?

This is the point where most SME directors stop. Not because they don’t want the answer. But because the step from insight to action feels bigger than it is. There are too many possibilities. Too much advice. Too many tools that are supposedly “essential.” The 90-day plan solves that. Not by tackling everything at once, but by deliberately choosing what you won’t do — so you can fully focus on what does work.

Why 90 Days Is the Right Rhythm

Not too short, not too long.

In 30 days you can start a process but not produce results. In 180 days you lose focus and fragmentation sets in — exactly the problem you were trying to get away from.

90 days is the rhythm that works in other change programmes too. Quarterly goals are the standard in well-run companies for a reason. It’s long enough to achieve depth, short enough to stay committed.

Gartner analysed adoption patterns at mid-sized companies and concluded that organisations implementing AI in focused sprints of 60 to 90 days reach measurable results forty percent faster than companies working with an open time horizon. The reason is simple: a deadline forces choices. And choices are precisely what most AI programmes are missing.

The Structure of the First Sprint

A 90-day AI plan for SMEs has three phases. Each runs for one month. They build deliberately on each other.

Month 1: Foundation

This is the phase that gets skipped most often — and that is most decisive for everything that follows.

Choose one process. One.

Not three, not “communication in general,” but one specific, clearly defined process that currently costs time, is prone to errors, or runs inconsistently.

Good starting points are processes that:

• Are repetitive — the same pattern every week
• Cost a lot of time relative to the value they add
• Depend on one person who has “got the knack” for it
• Require a lot of text production that is currently done manually

Poor starting points are processes that:

• Are directly client-facing without an intermediate check
• Depend heavily on nuance and relationship
• Involve strategic decisions that cannot be standardised

Next, designate an owner for this process and this programme. Someone who spends time on it weekly — a few hours, not a full day — and reports to you.

Draw up the three rules that apply to this programme and to AI use in general.

What you have at the end of month 1:

• One clearly chosen process
• An owner with a mandate
• Three rules that everyone knows
• A baseline measurement: how much time does this process currently take?

Month 2: Pilot

Now you actually begin.

Implement AI in the chosen process. Not as an experiment nobody takes seriously, but as a deliberate pilot with a clear goal: finding out what works, what doesn’t, and how much time it saves.

Have the owner keep a simple log. No extensive report — a few lines per week is enough. What went well? What went wrong? Which output was usable without adjustment, which required correction?

This log is more valuable than you realise right now. It will become the foundation for the working method you’ll standardise after this sprint. And it gives you concrete answers when employees ask: “But does it actually work?”

Halfway through month 2 there is a review point. Not to stop or restart the programme, but to adjust course if the direction isn’t right. Is the chosen process actually too complex? Is there a more logical starting point? Those are questions better asked halfway through than at the end.

What you have at the end of month 2:

• First concrete time savings, documented
• A log of what works and what doesn’t
• One or two employees who know the approach and can work with it
• A clear view of what still needs to be adjusted

Month 3: Embedding

This is the phase that determines whether the first two months have lasting effect or disappear once attention shifts elsewhere.

Stop saying “we’re trying it” and make the approach part of the standard working method. Document how the process now runs — not in an extensive manual, but in a workable instruction that a new employee can follow.

Embedding also means: the owner introduces the working method to colleagues who also carry out that process. Not as training, but as a handover. “This is how we do it now. These are the rules. These are the pitfalls I encountered.”

At the end of month 3 you ask one question: which process is now ready for the second sprint?

What you have at the end of month 3:

• An embedded approach for one process
• Documented time savings
• An owner who knows what they’re doing
• An organisation that has proven it can use AI productively
• A clear picture of which next step fits

The Three Pitfalls in the First 90 Days

Pitfall 1: Starting too broad

The temptation to tackle several processes at once is strong. Especially when employees come forward with their own ideas. Especially when the first results look promising. Don’t do it. Starting broad leads to shallow results across the board — and no single process that actually works. Focus on one process until it’s embedded. Only then move on.

Pitfall 2: No measurement

Without a baseline measurement you don’t know what you’ve achieved. “It feels faster” is not a result you can share with your team, your clients, or yourself as motivation to keep going. Measure simply. How long did this process take before the pilot? How long does it take now? What percentage of the output is directly usable without adjustment? Three numbers. That’s all you need.

Pitfall 3: The owner disappears into the daily grind

This is the most common reason things stall after month 1. The owner has good intentions, but day-to-day operations win out over the programme. The solution is banal but effective: block a fixed time in the week. Two hours. Regardless of what else is going on. Treat it like a client appointment — not something you can reschedule when something “more important” comes up. AI adoption only feels less urgent than other things on the agenda because the cost of not doing it is invisible. But it’s there.

What Comes After 90 Days

After the first sprint, you have something most SMEs don’t have: proof.

Proof that AI works in your organisation. Proof in the form of a concrete number, a documented working method, and an owner who can tell the story.

That proof does three things.

It convinces sceptical employees better than any external source. It gives you the foundation to start a second sprint — faster this time, because you know what you’re doing. And it gives you as a director the peace of mind of not having to come up with a new AI strategy every month. The second sprint covers a different process. Or deepens the first one. Or transfers the working method to a larger part of the team. But that’s a decision for September. Right now the question is: which process do you choose for the first sprint?

Three Questions to Answer This Week

1. Which process in your organisation currently costs the most time relative to the value it adds?
2. Who in your organisation is the most logical owner of the first AI programme?
3. When are you starting — and is that already in the diary?

If you have clear answers to these three questions, you’re further along than most SME directors in the Netherlands. The rest is execution.

Ready to get started? At LinkLeaders, we guide the first 90-day programme from A to Z — from choosing the right starting point to embedding the approach in your organisation.

Call +31 (0)182 570927 or email contact@linkleaders.nl.