The news
Martech.org published a sharp piece arguing that most organizations have confused an AI portfolio with an AI strategy — and that the difference between a stack of pilots and a real strategy is whether it's tied to specific outcomes and operating changes. The article offers four diagnostic questions to help leaders figure out which one they actually have.
Our take
This is the right diagnosis, and it's more widespread than most GTM leaders want to admit.
The pattern looks like this: a vendor demo lands well, a team member finds a tool that saves them twenty minutes on a task, a CMO feels the competitive heat, and suddenly there are six active pilots, none of which connect to each other or to a documented business outcome. The portfolio looks busy. Nothing in the pipeline has actually changed.
The article is correct that strategy starts with the experience you want to create and the operating model required to deliver it. But here's the operator-level mechanic that usually goes missing: before you can automate toward an outcome, the process that produces that outcome has to be documented. Not vibed at. Not understood by one person on the team. Written down, step by step.
This is why so many AI pilots plateau. The tool works fine — the workflow it's supposed to accelerate was never actually mapped. The automation hits the first undocumented decision point and either stalls, errors out, or hands off to a human who wasn't expecting it. The pilot gets chalked up as "not quite ready" when the real issue is that the team handed an AI a job description instead of a process.
A portfolio of point solutions without a shared outcome is also compounding debt, not compounding wins. Every disconnected tool is a new context switch, a new login, a new prompt someone has to remember to run. The teams pulling ahead aren't the ones with the most tools — they're the ones chaining small, reliable automations toward a single measurable end state.
The so-what
Ask the four questions the article poses, but start with a fifth one first: can you draw a straight line from your AI spend to a specific metric that was already on your team's scorecard before AI entered the conversation? If the answer is no, you have a portfolio, not a strategy.
- Audit your active AI tools and map each one to a single owned outcome — if you can't, the tool shouldn't survive the next budget review.
- Before adding anything new, document the manual process it's meant to replace. No documentation, no automation.
- Look for two or three tools that are already working and ask whether they could be connected rather than supplemented with something new.
Activity is easy to generate. Durable change requires knowing what you're building toward before you buy the next thing.