Once upon a time, Arthur Laffer sketched a curve on a cocktail napkin, or so the story goes. The idea behind it was simple but enduring: Tax revenue isn’t a straight line against the tax rate. At zero percent, the government collects nothing, but it also collects nothing at one hundred percent, since nobody works for free. The peak is somewhere in the middle. Once you’ve passed the peak, continuing to raise the rate will actually lower revenue instead of raising it. The incentive to produce collapses faster than the rate can extract value from what’s left. 

It’s a powerful illustration about how to turn diminishing returns into negative returns.  

As it turns out, it’s also a useful way to think about the modern marketing stack. 

The pattern is the same 

Marketing teams have their own version of the zero-percent problem. An organization with no research discipline, structured audience definition, or shared data layer is forced to operate on instinct. Decisions are quick, often wrong, and indefensible when a campaign underperforms. That’s the low end of the curve — plenty of activity with little return because there’s nothing to ground it all. No foundation.  

The intuitive fix has been to add, whether it’s a research tool, a CDP, or a new DSP. Add an attribution platform, creative testing tool, or social listening dashboard. Build a stack! Each addition looks a lot like progress. The industry has spent the last fifteen years running that experiment at scale, and honestly? It looks like a lot of organizations have driven straight past the peak of their own curve without even noticing. 

The numbers don’t look good. The martech landscape now catalogs more than 15,500 distinct commercial tools. Tellingly, that count grew less than one percent last year, delivering the flattest growth in 15 years of tracking. That suggests that the market may sense that it has overshot. The average enterprise runs around 90 martech tools. According to Gartner’s Marketing Technology Survey, only about 49% of marketing tools are regularly used. The sad truth is that, on average, more than half (and possibly up to 2/3) of what marketing organizations are paying for is sitting idle. 

Meanwhile, budgets are tightening around this fragmentation. In 2025, Gartner’s CMO Spend Survey found martech’s share of the marketing budget actually fell from 30 percent in 2023 down to 22 percent, even as the tool count kept climbing, and a majority of CMOs say they don’t have enough budget to execute the strategy they’re accountable for. And that’s exactly what this side of the curve will look like: more inputs, shrinking returns, and less capacity use the tools that have already been bought. 

Why adding more can make it worse 

The least intuitive lesson is that once you pass the highest point of the curve, additional inputs do more harm than good. In tax policy, that happens because the incentive to work is destroyed faster than the higher rate can capture. In marketing, the mechanism is different, but the shape is the same. Every new, disparate tool means another handoff, and at every handoff, an audience definition gets reinterpreted, a data model gets reconciled, or some of the context gets lost. 

An audience built in a research platform is rebuilt by a strategy team creating a brief. That brief is rebuilt by the creative team, who turns it into concepts. It’s rebuilt again by the media team as they set targeting parameters. By the time the campaign is live, the “audience” carries just a hint of the original research described. The definition of that audience decayed one handoff at a time, the marketing equivalent of a rising tax rate eating the incentive to produce. 

This is also the trap waiting inside the current AI buildout. It’s projected that more than 40 percent of agentic AI projects will be abandoned by 2027 due to “escalating costs, unclear business value or inadequate risk controls.” So, this failure has nothing to do with the models themselves, but everything to do with the fact that they’re being layered onto data infrastructure that was never built to support them. That’s the curve again: Adding a layer of intelligence on top of an unstable foundation moves an organization past the peak, not to it. 

Finding the peak, not abandoning the climb 

I’m not saying structure is bad, nor doesLaffer Curve conclude that taxation is bad.  

But having no process is as productive (or unproductive) as maximum fragmentation. There’s a peak; a point where an organization has enough shared structure to make decisions defensible and repeatable, without so many disconnected handoffs that the structure itself becomes the tax. 

Simply buying the next point solution isn’t likely to get you to that peak. Recent data from The State of the Stack report shows 62 percent of marketing teams are already using more tools than they were two years ago, and nearly two-thirds cite data integration as their biggest stack-management challenge. Adding another tool to fix that problem is, almost by definition, a slippery slide down the wrong side of the curve. 

The really durable solution is having fewer handoffs between the moment an audience is understood and the moment it’s acted on. Without all those handoffs, the audience stays intact. The audience discovered via research is the same one that gets targeted, and not a fourth-generation translation of it. That’s more work (and harder work) than building a new dashboard or another level to the tech stack. 

Every single marketing organization around the world is somewhere on this curve right now. The strategic question for 2026 is whether anyone has summited the peak, or whether the team is still climbing past it, one well-intentioned, misguided tool at a time.