Last year I wrote about the gap between martech promises and results. The questions I raised about revenue allocation, process dysfunction, and team preparation became more pointed. Because there is now a costly new variable: AI that promises to fix everything but has trouble proving that it fixes anything.
Agentic AI is the buzzword of the moment. Vendors offer autonomous systems that plan campaigns and optimize spend without human intervention. The demos look amazing. The reality of production is different.
Gartner predicts that more than 40% of agentic AI projects will be canceled by the end of 2027. Not because the technology is disappointing in controlled environments, but because costs rise, risks surface, and the business cases never materialize. We’ve watched this pattern repeat itself with every generation of martech sophistication.
Agentic AI adapts your strategy, for better or worse. When your workflows are disrupted, adding autonomous agents helps amplify the disruption at machine speed.
Your customers are searching everywhere. Make sure your brand introduces himself.
The SEO toolkit you know, plus the AI visibility data you need.
Start free trial
Start with
The collapse of confidence in return on investment
Share of marketers who say they can prove AI ROI went from 49% to 41% in a single year. In retail, the fall was steeper: from 54% to 38% despite steady adoption.
The bar has moved. The first successes of AI focused on speed: faster content production, automated segmentation. These gains were real but superficial. Now, executives want revenue growth and pipeline contribution.
Most teams can’t connect these dots because they never built the measurement infrastructure to track them. They layered AI on top of the same flawed attribution models and manual reporting processes that were already failing.
Proving ROI requires internal muscles to define what success looks like and instrument it correctly. Then you need to report it in a language that finance understands. No AI tool does this for you.
Your people problem has a new layer
Marketing organizations are still structured around tools rather than results. Campaign managers cannot access customer data. Analysts create reports without understanding marketing strategy, while strategists plan campaigns they cannot measure. None of this has improved since last year. AI has made the situation worse.
AI is eroding mid-level marketing roles faster than most leaders want to admit. When an agent can write positioning, pressure test messaging, and generate campaign variations before lunch, what does human expertise mean?
The marketers who thrive in this environment are the ones who can look at what AI has produced and know which 20% is wrong, why it’s wrong, and how to fix it. Role confusion and discreet disengagement spread within teams that have not yet made this shift.
The latest research by Scott Brinker and Frans Riemersma describes a a split emerges between “The Laboratory” and “The Factory” in marketing operations. The Laboratory takes care of the experimentation. The Factory runs large-scale, revenue-critical programs. Organizations that attempt to execute both with a single process and set of KPIs fail in both cases.
Clicking buttons on platforms has never been a lack of skill. Your people must have the ability to drive business results and use the judgment to interpret results when the data surprises them. Most companies spend more on unused SaaS features than on developing these skills.
Process Dysfunction Meets AI at Scale
Your marketing processes seem solid and paper-based, but fall apart in practice. This automated campaign workflow still requires manual intervention at every step.
AI has not solved process dysfunction. He revealed it: the secret solutions, the undocumented spreadsheets that hold everything together. Agentic AI can’t navigate all of this. This requires clear input and clear decision-making authority. Most marketing organizations offer neither.
Attribution presents a new problem: your buyer’s AI assistant has already made the shortlist before your analytics records a visit. The lead nurturing sequences and funnels you spent years building? Buyers completely ignore them.
Adding complexity to keep up with the chaos still doesn’t work. Measure what you can prove and build from there.
The year of capacity building
If 2025 was the year of AI experimentation, 2026 is the year experimentation meets responsibility. Accountability reveals the gap between organizations that have invested in capabilities and those that have invested in tools.
Organizations with strong operational power extract real value from the right platforms. Lean organizations underutilize sophisticated systems because teams can’t make them work. AI is widening this gap faster than any previous technology cycle.
Capacity determines what you get from the technology. This has always been the case. Choose a workflow where your team relies on workarounds rather than the platform. Correct this before purchasing anything else.
