AI & strategy
AI and digital strategy: build the plan before you buy the tools.
Almost every business is now trying AI. Very few are getting measurable value from it. The research is consistent about why, and the fix is less about technology than most vendors would like you to believe.
Most conversations about AI in marketing start with a tool. A vendor demo, a new feature in the CRM, a competitor's announcement. The question becomes "how do we use this?" rather than "what problem are we trying to solve?"
That order matters more than it seems. The data from the last two years shows a wide and persistent gap between AI spending and AI results. The businesses on the right side of that gap are not the ones with the most tools. They're the ones that did the planning first.
The gap between AI investment and AI value
Adoption is no longer the question. Statistics Canada's 2026 analysis found that 19.2% of Canadian businesses used AI to produce goods or deliver services in the previous 12 months, up from 12.2% in 2025 and 6.1% in 2024. Among larger organizations, use is much higher: KPMG Canada's 2025 survey of 753 business leaders found 93% said their organizations were using AI in some form.
Results are another story. In that same KPMG Canada survey, only 2% of respondents said their organizations were seeing a return on their generative AI investments. More than half (57%) named understanding how to capture value as a top challenge, and only 38% said they had a clear plan for extracting value.
The global picture is similar. BCG's 2025 research, based on 1,250 senior executives, found only 5% of companies qualified as "future-built" and achieving AI value at scale, while 60% reported minimal revenue and cost gains. The leaders were pulling away, with 1.7 times the revenue growth and 1.6 times the EBIT margin of the laggards.
The most cited data point comes from MIT NANDA's 2025 report, The GenAI Divide. Despite an estimated $30 to $40 billion USD in enterprise investment, it found 95% of organizations getting zero return from generative AI, and only 5% of custom enterprise AI tools reaching production. The authors were clear that the gap was driven by approach, not by model quality.
A note on the data
These studies measure different things. Statistics Canada surveys businesses of all sizes, KPMG Canada and BCG survey senior leaders, and MIT NANDA's report is described by its own authors as preliminary and based on interviews with 52 organizations, a survey of 153 leaders and a review of over 300 public initiatives. Treat any single number with caution. The pattern across all of them is what matters: adoption is high, measurable value is rare.
Why tool-first AI projects fail
Analyst forecasts point to the same causes. Gartner predicted in 2024 that at least 30% of generative AI projects would be abandoned after proof of concept by the end of 2025, citing poor data quality, inadequate risk controls, escalating costs or unclear business value (as reported by Intelligent CIO, 2024). In 2025 it made a similar call about AI agents, predicting over 40% of agentic AI projects would be cancelled by the end of 2027, and estimating that only about 130 of the thousands of vendors marketing "agentic" products were the real thing (as reported by W.Media, 2025).
In marketing specifically, we see four recurring patterns:
- No defined problem. The project is framed as "use AI for content" rather than "cut our time to launch a campaign by half" or "lift qualified lead rate." Without a target, nobody can say whether it worked.
- Data that isn't ready. AI amplifies whatever it's fed. Duplicated CRM records, broken tracking and inconsistent naming produce confident but wrong outputs. Our piece on why your marketing numbers don't add up covers the measurement side of this.
- Visible but low-value use cases. MIT NANDA found roughly half of generative AI budgets went to sales and marketing, partly because results there are easy to show, while some of the clearest savings came from back-office automation.
- No owner and no rules. Tools get rolled out without anyone accountable for outcomes, and without guidance on brand voice, privacy or approvals.
None of these is a technology problem. That's consistent with BCG's 10-20-70 principle, which holds that 10% of AI value comes from algorithms, 20% from data and technology, and 70% from people and processes.
A practical AI strategy framework
This is the sequence we use. It maps onto our Think, Plan, Execute approach, and it works whether you're a five-person team or a national brand.
Think: define the problem and check readiness
1. Start with the business problem. List the three to five marketing constraints that actually limit growth. Slow content production, weak lead qualification, poor reporting, high cost per acquisition. Write each as a measurable outcome. If AI isn't the best answer to a problem, that's a useful finding too. This is the same discipline as any digital strategy.
2. Check data readiness. For each problem, ask what data the solution depends on, where it lives, how clean it is and who can access it. Check consent and privacy obligations at the same time. A use case with great potential and poor data should wait until the data is fixed.
Plan: prioritize, then pilot
3. Prioritize use cases. Score each candidate on value, feasibility and risk. A simple grid works:
| High feasibility | Low feasibility | |
|---|---|---|
| High value | Pilot first | Fix the foundations, then revisit |
| Low value | Quick wins if cheap and low risk | Park it |
4. Pilot with a success metric. Pick one or two use cases. Set a baseline, a target, a time frame (usually 60 to 90 days) and a decision rule for what happens if it hits or misses. Consider buying before building: MIT NANDA found purchased, customized tools from external partners reached deployment about 67% of the time, compared with about 33% for internal builds, though the authors note these figures are self-reported.
Execute: govern, then scale
5. Set governance. Decide who owns each AI use case, what data it can touch, what needs human review before it goes public, and how you'll handle errors. Keep it short enough that people will actually read it.
6. Scale what works. Only expand a pilot once it has hit its metric. Scaling means redesigning the workflow around the tool, training the team and updating reporting. It doesn't just mean buying more licences.
What this looks like in practice
A useful test: could you explain, in one sentence each, the problem your AI project solves, the data it depends on, the metric that proves it works and the person accountable for it? If not, the project probably isn't ready to fund. For more on how Canadian teams are using AI day to day, see our look at AI in digital marketing in Canada. If much of your opportunity is in repetitive operations, our guide to what to automate first is a good companion read.
How Three Dots Brands approaches this
We help brands in Toronto and across Canada build AI into their marketing plans using our Think, Plan, Execute approach. We start with the business problem and the data, not a vendor shortlist, then help you prioritize, pilot and measure. We don't resell AI tools, so our recommendations aren't shaped by what we'd earn from them. If you're not sure where AI fits, or whether the real constraint is somewhere else, our marketing diagnostic is a good place to start.
Frequently asked questions
What is an AI marketing strategy?
An AI marketing strategy is a plan for where AI will improve specific marketing outcomes, what data and processes it depends on, who owns it, how success will be measured and what rules govern its use. It starts with business problems and only then chooses tools, rather than starting with a tool and looking for somewhere to use it.
Why do so many AI projects fail to deliver value?
The research points to the same causes again and again: unclear business value, poor data quality, rising costs, weak risk controls and tools that don't fit how people actually work. MIT NANDA's 2025 report found most generative AI pilots had no measurable profit impact, and attributed the gap to approach and implementation rather than model quality.
Do we need an AI strategy consultant or can we do this in-house?
Many teams can run the framework themselves if someone senior has the time and authority to own it. An outside AI digital strategy consultant is most useful when nobody internally can step back from day-to-day work, when vendors are driving the agenda, or when you need an independent view on data readiness and priorities.
How long does it take to see results from AI in marketing?
A well-scoped pilot can show whether it moves its target metric within 60 to 90 days. Broader value takes longer. In KPMG Canada's 2025 survey, six in 10 business leaders expected a return within one to five years, and three in 10 within the year. Set a clear success metric before the pilot starts so you can tell the difference.
Sources
- Statistics Canada, Analysis on artificial intelligence use by businesses in Canada, second quarter of 2026, 2026.
- KPMG Canada, Canadian businesses adopting AI, but few are seeing ROI, 2025.
- BCG, AI leaders outpace laggards with double the revenue growth and 40% more cost savings (The Widening AI Value Gap), 2025.
- MIT NANDA, The GenAI Divide: State of AI in Business 2025 (hosted copy of the report PDF), 2025.
- Intelligent CIO, Gartner predicts 30% of generative AI projects will be abandoned after proof of concept by end of 2025, 2024.
- W.Media, Over 40 percent of agentic AI projects could face the axe by end of 2027: Gartner, 2025.
- BCG, What's driving AI value, 2026.
Gartner figures are cited through trade press coverage of Gartner's own press releases. Survey figures are self-reported and samples differ by study, so compare them with care.