90 days does not transform your company - it proves transformation is possible

90 days won’t change your company. Full stop. But it will tell you whether rollout is worth pursuing at all. That’s the real difference between a sprint and a slog: one proves value, the other assumes it.

I learned this the hard way at Tallyfy, and I learned it the embarrassing way. I’d pitched 12-month rollout roadmaps and watched them rot. Week three brought the first crisis. Week six brought budget questions. By month three, the original vision was buried under “urgent” priorities nobody could quite define. Then I switched to 90-day pilots focused on proving specific value propositions. Renewals became automatic. Expansions became obvious. Proof beats promises. Every time.

What’s the problem with how companies set this up?

Something bugs me about quarterly planning sessions. John Kotter’s change management research keeps showing less than 30% of rollouts succeed. Look closer and you’ll find something interesting: the ones that work start with clear proof points, not grand visions. Current data on AI adoption shows the vast majority of organizations now use AI in at least one function. Only about 6% are generating value at scale. Everyone’s experimenting. Few are transforming. The AI adoption flywheel explains why peer-driven momentum matters more than mandated timelines.

The setup problem is this: companies treat 90 days as a compressed version of a multi-year rollout. They’re trying to finish something instead of prove something. Those are fundamentally different goals, and conflating them is why pilots stall before they ever leave the lab. Whilst it sounds harmless on a kickoff slide, the slip in mindset is rubbish for outcomes. 90 days is long enough to push past the honeymoon phase where everything seems possible. Short enough to maintain urgency without triggering change fatigue. Employees tend to hit peak resistance somewhere around the two-month mark of any major change, once the initial excitement fades and real friction sets in. HBR reported that employee willingness to support enterprise change collapsed from 74% to 43% between 2016 and 2022. By day 90, you’ve either pushed through or you haven’t. No ambiguity. Six months gives doubt too much room to grow. One month doesn’t produce real behavioral change. 90 days hits the sweet spot where iterations produce patterns - three full monthly cycles, each building on the last.

A Nigerian education pilot achieved two years of academic progress in six weeks. If that’s possible in education, one of the slowest-changing sectors, the real question isn’t whether 90 days is enough. It’s whether you’re asking the right question going in.

What you’re actually proving

OK so here’s what’s interesting. Most companies think a 90-day sprint should deliver a mini-change. Wrong. You’re not building the future in 90 days. You’re proving it’s buildable. That’s the core Lean Startup insight from Eric Ries applied to organizational change: validate before you scale.

Four things get tested simultaneously.

Your team’s ability to learn new tools without melting down. Prosci’s research shows 43% of AI adoption failures trace back to inadequate executive sponsorship, while 38% come from user proficiency gaps like learning curves and poor training. Together, people problems account for the majority of failures, outpacing technical issues by a wide margin. In 90 days, you’ll know if your people can adapt or if you need a different approach.

Your processes’ actual flexibility. Can existing workflows bend without snapping? Turns out most “core processes” were really just habits nobody had questioned in years. 90 days forces those questions out into the open, which is uncomfortable but useful.

Your leadership’s commitment when things get hard. The companies that succeed are the ones whose senior leaders demonstrate real ownership of AI work. Day 60 tells you everything about whether that’s real or just performative enthusiasm at kickoff meetings. If leadership stops showing up by week six, you have your answer.

Your actual data readiness. Everyone thinks their data is “pretty good” until they try feeding it to AI. A full 57% of organizations estimate their data isn’t AI-ready, and data quality remains one of the top obstacles organizations face. Better to discover that in a sprint than mid-rollout. Does that mean you need perfect data to start? No. You just need to know how far off you actually are.

Strike that, let me say it better. You don’t need to know how far off you are. You need to know whether the gap is fixable in the next sprint cycle or whether the gap is the whole project. Two very different decisions hide inside that one question.

The sprint framework that actually produces results

After mulling this over across multiple consulting engagements, here’s where I landed. Forget complex methodologies. Most multi-year programs cobble together half a dozen frameworks and call it strategy. What actually drives results is simpler than most consultants will admit.

Days 1-30: Foundation without overthinking

Pick one use case. One. Not three with a backup. Redesigning the workflow around that one use case matters more than the tool you pick. Clarity beats coverage every time.

Get the right people involved from day one. Not a committee. One person owns it. They pull everyone else in. This isn’t democracy; it’s delivery. Set up basic measurement - not perfect dashboards, just enough to know if you’re winning or losing. You can refine the metrics later.

Days 31-60: Reality meets resistance

This is where most pilots die. The novelty wore off. Real, messy problems surfaced. Change fatigue kicks in hard. If you can push through this phase, you’ve proved more than any presentation deck ever could. This is where you find out if your organization has the stomach for real change or just likes talking about it at off-sites.

Double down on what’s working. Kill what isn’t. No sentimentality. Done.

Days 61-90: Scaling signals emerge

Patterns are clear by now. You know what scales and what doesn’t. You’ve identified your champions and your skeptics. The compound effect starts showing.

Document everything that worked. Pull it out of people’s heads before they forget the details. Turn it into repeatable processes. This is your blueprint for the real change that follows.

Choosing where to run your first sprint

People don’t give this step enough weight. Pick wrong and you’ve wasted 90 days. Pick right and you’ve bought years of momentum. I think this choice matters more than most people realize. (The choice is usually made in about 20 minutes, in a meeting where nobody pushed back.)

Will the perfect candidate process announce itself? No. You have to go find it. Next question.

Look for processes with measurable outcomes within 30 days, willing participants who won’t quietly undermine the work, enough complexity to be worth the effort but not so much that it’s overwhelming, and clear before/after comparisons that you can show to skeptics.

I watched one company try to change their entire customer service operation in 90 days. Failed spectacularly. Another focused only on routing tickets 20% faster. That 20% improvement bought them executive support for the next five sprints. The scope difference was the whole game.

Mid-size companies have real advantages here. You’re not fighting enterprise bureaucracy. You’re not constrained by startup chaos. You can pick a battlefield where winning is actually possible, which is probably more important than most guides acknowledge. If you’re keen on building durable momentum, this is where it starts. A focused 3-day audit can surface the right candidate process before the sprint clock starts.

If you want to apply this thinking to your firm, Blue Sheen handles work like this.

Converting sprint wins into lasting change

What I appreciate about this stage is that the work gets easier, not harder. In conversations I’ve had with mid-size leadership teams over the last couple of years, the conversion gap is where most of the trust is gained or lost. The majority of challenges in AI rollout relate to people and processes, not technical issues. Fewer than 20% of employees have heard from their direct manager about AI’s impact on their job. People arrive anxious and exhausted before you’ve even started. That’s the real environment you’re working in.

90 days works partly because it’s defined. There’s an end date. People can see the finish line. Set expectations early: this is about learning, not perfection. We’re testing feasibility, not delivering final products. Success means clear signals, not complete solutions.

Build in breather moments. Week 6 should be lighter. Week 11 should consolidate, not accelerate. Communicate progress weekly - not lengthy updates, just “this week we learned X, next week we’re testing Y.” Simple. Clear. Forward motion.

Most companies drop the ball right here. They run a successful 90-day sprint, celebrate, then nothing. Six months later they’re running another “pilot” because they never converted the first one into sustained change. Treat each sprint as a building block, not a standalone event. Your second sprint should build on the first. Your third should scale what the second proved.

Document patterns, not just outcomes. What worked and why? What failed and why? These patterns become your actual playbook. Build your coalition gradually. Each sprint should create more believers. By sprint three, you should have enough advocates that resistance becomes futile. Systematic strengthening beats random improvement.

One client ran four consecutive 90-day sprints, each building on the last. By month 12, they’d achieved more than the original 3-year plan had promised. They proved each step worked before taking the next one. That sequencing was everything.

One sprint is the unit. The cadence is what compounds.

The handoff arrow at Day 90 is the part most teams skip. Sprint 1 produces a result and the team disbands. Sprint 2 starts from scratch six months later. The loop only works if the team rolls into the next assessment with the previous sprint’s data, advocates, and scar tissue still warm.

90 days won’t change your company. But it will tell you if rollout is possible, who will drive it, what will break, and what it’s actually worth.

Will 90 days guarantee success? No. But it will guarantee clarity.

In a world where 95% of GenAI pilots fail to deliver measurable returns and analysts predicted 30% of AI projects would be abandoned after proof of concept, intelligence beats ambition. Start your 90 days to prove something. Anything.

Because proof builds momentum. And momentum is what reshapes companies.