Your Martech Stack Strategy Was Never in the Stack

Martech stack strategy is what your team does with the technology to reach a goal, not the technology you wire together. A book on networking drew that line in 1995. The industry still trips over it, and AI agents are about to test it harder than anything has.

A Distinction Drawn Before the Web Had Pictures

I’ll put a stake in the ground. The most useful idea about marketing technology I know is one my coauthor and I worked out before most marketers had heard of a browser.

In 1995, Paul David Henry and I published a book called Strategic Networking (1. Henry & De Libero, 1995). On the surface it was about connecting computers. The argument underneath it was about people. We kept tripping over a single word. “Networking” meant two different things, and the gap between them decided whether any of the technology mattered.

One meaning was the wiring: the cables, protocols, and machines you connect. The other was what people did with that wiring to get somewhere they cared about. Paul put it plainly in an article a few years later. Unless the networks you build are used strategically to reach your goals, you connect people, but not always to those goals (2. Henry, 2002). The network was the thing you made. The strategy was what you did with it.

Swap “network” for “stack” and you have the central confusion of martech in 2026.

Your marketing technology stack is the wiring. Martech stack strategy is what your team does with it to reach a business outcome. The two are not the same, and most stacks get bought as if they were. The platform is selected, integrated, and lit up, and somewhere in that work the goal it was supposed to serve goes quiet. You end up with tools that talk to each other and a team that can’t say what any of it is for.

Why the Stack Alone Never Decides the Outcome

There’s a reason buying technology so rarely produces the result the business case promised, and the reason isn’t the technology.

The psychologist Albert Bandura described a pattern he called reciprocal determinism (3. Bandura, 1978). Behavior, environment, and the person don’t sit in a tidy causal line where one drives the next. They shape each other, all at once. Change the environment and behavior shifts. Change behavior and the environment gets remade. No single corner is the cause.

A martech stack is an environment. It shapes what your marketers do, what they believe is possible, what they stop attempting because the tool makes it painful. Their behavior then reshapes the stack through the workflows they build, the fields they actually fill in, the features they quietly abandon. And both of those are pulled by a third force: the goal you’re trying to reach. Pull any one corner out of that loop and the other two go slack.

That’s why the two most common martech moves both disappoint. “Buy the platform” treats the environment as the whole system and waits for behavior and outcomes to follow on their own. “Rip and replace” swaps the environment and expects new behavior to appear because the logo changed. Both ignore the loop. The stack is one corner of a triangle, and a triangle with one live corner isn’t a shape.

Donald Norman made a smaller version of this point about tools in general. He drew a line between knowledge in the head and knowledge in the world, the way a sticky note on a monitor holds a reminder your memory won’t (4. Norman, 1988). A stack is knowledge in the world at scale. It’s where your strategy is supposed to live once it leaves someone’s head. That only works if the strategy was ever in a head to begin with. Software can hold an intent. It can’t supply one.

When the Tool Starts Supplying the Strategy

Here’s where the thirty-year-old distinction stops being tidy.

For all those years, one thing held steady. The strategic intent came from people. The wiring carried it, stored it, executed it, but a human decided what the goal was and which moves served it. The split between the stack and the strategy held because only one side of it could think.

That assumption is now in play, and the idea behind the change is older than it looks. In 1994, Pattie Maes described software agents that could cut a person’s workload and information overload by acting on their behalf, under their control (5. Maes, 1994). For decades that stayed mostly a research idea. It isn’t one anymore. Agentic systems now take a goal, break it into steps, and run them, inside the same marketing stacks we’ve been talking about. An agent wired into a journey-orchestration tool can watch a live campaign, shift spend across segments, and rewrite send times toward a conversion target you set once and then walked away from.

When the tool starts doing some of the strategic work, the line between the wiring and the strategy blurs. The environment corner of Bandura’s loop begins to behave like the person corner. That’s genuinely new, and I won’t pretend the consequences are settled.

What I will say is that the blur relocates the distinction rather than erasing it. The question stops being “did your team use the stack strategically” and becomes “who set the goal the agent is now pursuing, and who’s checking that it’s still the right one.” An agent optimizing hard toward a goal nobody examined is the old problem wearing faster shoes. The trade-off is concrete. The more strategic work you hand to the system, the more it does for you, and the easier it becomes to lose track of whether the goal still serves the business. Speed and abdication arrive in the same box.

The Distinction Holds. Defend It.

Thirty years on, I’m more convinced of the original point, not less.

The technology has always been the place a strategy goes to live once someone has one. Owning the place was never the same as having something to put in it. That was true of the local area networks Paul and I wrote about, and it’s true of a composable marketing technology stack with agents running through it. The work that decides whether any of it pays off is the same work it always was. You decide what you’re trying to achieve, then keep the wiring, the people, and the goal inside the same loop.

So before the next platform decision, the next integration, the next agent you switch on, ask the question the wiring can’t answer for you. What goal is this serving, who owns that goal, and how will we know if the technology drifts away from it? If you can’t answer that in a sentence, the problem was never your martech stack strategy. It’s that you don’t have one yet.