Executive summary
AI adoption pays back when leaders treat AI as a change to how work gets done, owned by the business and measured in results. Most companies treat it as a tool purchase. That is why nearly everyone has AI and so few can say what it has returned.
- 94% of Canadian CEOs say they use AI to some extent, but only 29% have applied it at scale, against 43% of CEOs globally.1
- Only 43% of AI initiatives delivered their expected return over the past two years, according to IBM's 2026 study of Canadian organizations.2
- More than 80% of CEOs in Bain's 2026 survey say they aren't satisfied with what AI is delivering.3
The leaders getting a return make three moves:
- Aim: they put AI on the few workflows where the money is, and fund those properly.
- Redesign: they change the work around the tool, and a business leader owns the result.
- Measure: they track what the work produces before and after, and stop what doesn't pay.
Buying AI is easy now, so the advantage has moved to adoption
I was leading transformation at Walmart Canada when ChatGPT came out. At the time, if you wanted a tool to write product descriptions for the website, you more or less had to build it. Today there is a tool for almost anything. Buying AI doesn't separate one company from another anymore.
Adoption does. IBM found that 90% of Canadian CEOs say they are embedding AI across multiple workflows, and 80% believe they are moving at the pace their goals require.2 Yet fewer than half of AI initiatives paid back as expected. MIT's study of generative AI found it was falling short at 95% of the companies in its dataset, with most pilots showing little to no measurable impact on profit and loss.4
So why do leaders keep investing in AI when, for most of them, it hasn't been making them more money? Because buying the tool feels like progress. A licence is a decision you can make on a Tuesday. Changing how thousands of people do their work is a decision you have to keep making for a year.
Canada is behind on scaling AI, and a cost-only lens keeps it there
PwC's 2026 CEO Survey found Canada trailing on scale: 29% of Canadian CEOs have applied AI at scale, against 43% globally.1 The same survey found only 27% of Canadian CEOs expect Canada's economy to improve over the next year, down from 42%.
When confidence is low, AI gets framed as a cost play. More than 80% of the CEOs Bain surveyed are using AI mainly for cost reduction and productivity.3 Cost and productivity are fair targets. The risk is that a cost-only lens produces small pilots with small budgets and nobody whose job depends on the result. I call those low-expectations experiments. They rarely fail loudly. They just never scale.
AI adoption breaks down in six predictable places
Across more than 80 organizations, I see the same six breakdowns. Two are about where AI is aimed, two are about how the work changes, and two are about how value is measured.
01. AI goes where it is easy. Teams start with whatever a vendor demoed or a keen employee tried first. Many companies have bought AI tools and put them in the wrong places, or bought them and never adopted them. The licence exists. Nobody uses it.
02. Effort is spread across too many pilots. Thirty small experiments each get a sliver of budget and attention. None gets enough to change a whole workflow. Deloitte's 2026 survey found only 16% of chief strategy officers say their organizations are using AI to fundamentally reimagine lines of business.5
03. The work around the tool stays the same. A tool dropped into an old process speeds up one step and leaves the bottleneck where it was. If approvals still need four sign-offs, a faster first draft doesn't change the cycle time.
04. Nobody in the business owns the result. IT owns the platform. The business owns the outcome. In between, nobody owns whether people actually work differently, and the rules are unclear. IBM found 73% of Canadian technology executives say AI adoption is outpacing their IT governance.2
05. Savings are counted, never captured. Business cases count the hours a tool could save. Few track where those hours go. If the saved time disappears into more email, the return is zero.
06. Pilots never have to prove themselves. Without a date to show value, a pilot can run indefinitely, using up attention and making the portfolio look busier than it is.
Three moves turn AI adoption into a return
Aim. Start with the work and the money. Where do people spend hours that don't need their judgment? Where does a slow step hold up revenue, a customer or a regulator? Where do you have more work than people? In many sectors talent is short, and AI can supplement the people you can't hire. Pick the three to five workflows that matter most and fund them fully. This fixes breakdowns 01 and 02.
Redesign. Before rollout, decide what the person does now, what the tool does, who checks the output and what happens to the time saved. Then give a business leader the adoption number, with the rules for data and review published before launch. Plan for the dip: any change to how people work makes performance worse before it makes it better. If you don't plan for it, you stop at the worst point and conclude AI doesn't work. This fixes 03 and 04.
Measure. Set a baseline on the work itself: cycle time, error rate, cost per transaction, revenue per person, response time. Track it after the redesign, decide in advance where the freed hours go, and set a date by which each initiative must show value. This fixes 05 and 06.
The AI adoption test
Six questions, one for each breakdown.
- Can you name the three to five workflows where AI should make the biggest difference to revenue, cost or service this year?
- Does each one have enough budget and attention to change the whole workflow?
- For each one, has the process changed, or has a tool been added to the old process?
- Which business leader owns each adoption number, and do people know the rules for using AI?
- When AI saves time, do you know where that time goes?
- Which initiatives will you stop if they haven't shown value by a date you've already set?
What leaders who get a return do in practice
1. They keep a one-page AI portfolio. Three to five bets, each with an owner, a target, a budget and a review date, reviewed by the leadership team on a fixed rhythm.
2. They budget for change as well as the licence. Training, time to learn, the performance dip and the support to get through it are in the business case from day one.
3. They map the workflow before they switch on the tool. The team that does the work redesigns it, with the tool in mind, before anyone is trained.
4. They put adoption on a line leader's scorecard. The leader whose team uses the tool owns weekly use and results, the same way they own sales or service.
5. They decide where the saved hours go. Before launch, they agree whether freed time goes to customers, growth work or lower cost, and they check that it does.
6. They hold a stop-or-scale review. On the date set at launch, each initiative either shows value and scales or stops. Stopping counts as discipline.
Bring The Next Big Thing to your event
The Next Big Thing is my keynote on AI and digital transformation that actually moves the needle. I talk about AI in business language, so every leader in the room can follow it and act on it.
Many leaders feel about AI the way you feel at a party when you've been talking to someone for an hour and forgotten their name. You're in too deep to ask. So I open by building a plain-language foundation in the room, so everyone uses the same terms. Then we cover where AI earns a return, why adoption stalls and what to do on Monday.
I deliver it as a 45- to 90-minute keynote, in person or virtual, tailored to your industry and audience, with a custom tip sheet your attendees can use after the event.
Conclusion: the advantage is in adoption
Nearly every company has AI, so having it is no advantage. The advantage goes to leadership teams that aim it at the right work, change the work around it, give a business leader the result and measure what comes back.
Seven in ten Canadian companies have not yet scaled AI. The ones that make these three moves in the next year will open a gap that the rest will find hard to close.
Read next: Willing and Able
Frequently Asked Questions
What is AI adoption?
AI adoption is the point where people actually change how they work using AI, every week, in the workflows that matter. Buying licences or running pilots gives access. Adoption is when the work itself is different and the result shows up in cost, revenue, speed or service.
Why aren't most companies seeing a return on AI?
Because AI is often aimed where it is easy instead of where the value is, spread across too many small pilots, added to unchanged processes, owned by no one in the business and never measured against the work. IBM found only 43% of AI initiatives delivered their expected return in the past two years.
How do you measure AI ROI?
Measure the work itself. Set a baseline for cycle time, error rate, cost per transaction, revenue per person or response time, track it after the workflow is redesigned, and count the full cost, including training, time spent learning and the short-term performance dip.
Why is Canada behind on scaling AI?
PwC's 2026 CEO Survey found 29% of Canadian CEOs have applied AI at scale against 43% globally, at a time when only 27% expect Canada's economy to improve. Low confidence tends to frame AI as a cost play with small pilots and small budgets, which makes scaling less likely.
What should leaders do first with AI?
Pick the three to five workflows where AI should make the biggest difference this year, fund them fully, redesign the work around the tool, give a business leader the adoption number and set a date by which each must show value.
Is The Next Big Thing a technical AI keynote?
No. It is built for business leaders. Suzanne talks about AI in the language of results: where it earns a return, why adoption stalls and what leaders do next. Each talk is tailored to the audience and industry.
About Suzanne Knight
Suzanne Knight, MBA, MA, is a keynote speaker and the CEO of mBolden, a Toronto-based management consulting firm (mbolden.co). She has led transformation in the Ontario government, at Deloitte for nearly eight years, as Director of Organizational Effectiveness at Philip Morris International in Canada, and as VP, Transformation Services at Walmart Canada.
She has advised more than 80 organizations in 14 countries, including Fortune 500 companies and governments, and has spoken on more than 65 stages in 18 months. Her TEDx talk is The myth of doing it all. Her keynotes are The Next Big Thing, Never Waste a Good Crisis, The Trust Advantage and Create More Time.
Sources and references
- PwC Canada. (2026, January). 29th Annual Global CEO Survey, Canadian findings (133 Canadian CEOs). Source ↗
- IBM Canada. (2026, June 22). Can Canadian organizations deliver in the next era of business AI? Source ↗
- Bain & Company. (2026, June 1). The 2026 CEO Agenda: Where Ambition Outpaces Execution (survey of 100 CEOs). Source ↗
- MIT NANDA initiative. (2025). The GenAI Divide: State of AI in Business 2025. Reported by Fortune. Source ↗
- Deloitte. (2026, February 12). 2026 Chief Strategy Officer Survey. Source ↗