The Missing Step in Every AI Strategy
Published on June 8, 2026 | 4 min read
Most organizations rush to implement AI by evaluating tools and vendors before they've answered the harder question: who are we, and what are we actually optimizing for? Strategy without self-knowledge is an expensive experiment.
Everyone assumes the hard part of an AI strategy is choosing the right model. The evidence says otherwise. MIT research found that 95% of organizations deploying generative AI saw zero measurable return. The model was not the problem. The organizations were — specifically, they built AI around a version of themselves they had never rigorously examined.
The $7.2 Million Assumption
The average sunk cost per abandoned AI initiative in 2025 was $7.2 million. Not because the technology failed. Because the teams behind those projects could not articulate what success would look like for an organization doing what they actually do — not what they assumed they did.
This is the identity problem, and it shows up at every level. A Charlotte-area law firm wants to use AI to improve client intake. Before they scope the project, they need to answer a harder question: what kind of firm are they? High-volume? Boutique? Relationship-first? Tech-forward? The answer changes every decision — the tool, the workflow, the vendor, the rollout plan.
Teams that skip this step do not discover they skipped it until they are already building. By then, the cost is not just financial. It is organizational trust in the next AI initiative, which is harder to recover than the budget.
Why the Tools Come Last
When organizations treat AI as a technical challenge, they fail at a predictable rate. 80% of AI projects fail overall, with 84% of those failures attributed to leadership — not engineering. The model works fine. The strategy surrounding it does not.
This failure pattern has a clear root cause: organizations start with capability instead of identity. They ask “what can AI do?” before they ask “who are we, and what do we need?” Those questions sound like soft groundwork. They are the highest-leverage work available before a single dollar is spent on implementation.
AI governance research points to a “70-20-10 rule” for successful implementation: 70% of effort goes to organizational readiness and change management, 20% to data infrastructure, and just 10% to algorithm optimization. The technology is a footnote. The organization is the story.
What “Knowing Who You Are” Actually Requires
Organizational identity is not a mission statement. It is not a values slide in a pitch deck. Those documents describe who an organization wants to be perceived as. Identity, in the operational sense, is what an organization actually optimizes for under pressure.
That distinction matters for AI strategy because AI amplifies existing behavior. An organization that optimizes for speed will use AI to move faster. One that optimizes for cost will automate until there is nothing left to cut. Neither outcome is inherently wrong — but both are predetermined by the organization's identity, whether that identity is explicit or not.
A brilliant AI solution built for the wrong organizational vision is just a highly efficient way to go in the wrong direction.
The organizations that get measurable returns spend serious time before implementation on three questions: What decisions are we making today? Which of them slow us down? And what is the consequence if we automate incorrectly? Davidson and Charlotte businesses that work through these questions systematically tend to ship fewer projects — and finish more of them.
The Systematic Starting Point
The path from here to a working AI strategy is not complicated. It is just often skipped. Before evaluating vendors, before piloting a model, before scheduling an “AI workshop” — establish a shared, written answer to what your organization is optimizing for.
Not a vision statement. A decision-making heuristic specific enough that two people on opposite sides of the organization could use it to reach the same conclusion independently. That document is your AI strategy's first input. Every downstream decision — data governance, tool selection, rollout sequencing, success metrics — gets cleaner once it exists.
Only 39% of respondents in McKinsey's 2025 State of AI report attribute any measurable impact to AI. The ones who do are not using smarter models. They are asking better questions before they build — questions that start with who they are, not what the technology can do.
If you are a Charlotte or Mooresville business navigating this decision, this is exactly where we start at Holistic Consulting. Explore our AI strategy and engineering services →