For a long time now, we’ve had a clear, clean taxonomy for enterprise software. Systems of record store data. Systems of engagement put it in front of people. The framework works and gives vendors a ladder to climb and buyers a map to navigate.
IBM added a third rung, systems of insight: platforms that monitor, analyze, and surface meaning from everything the first two layers captured. Qualtrics and Medallia climbed that rung and built empires on it. Survey programs; voice-of-the-customer (VoC) dashboards; Net Promoter Score℠ (NPS) tracking: the whole apparatus of “listening to the customer” industrialized, licensed, and baked into an annual recurring revenue model that’s pure shareholder-value steroids.
Insight Is Dead. Long Live … Erm, More Insight?
I know, I know. Everyone keeps claiming things in CX are dead, so now it’s my turn … except, let’s not get ahead of ourselves …
Insight Is Now A Depreciating Asset, Not A Value Proposition
The insight layer isn’t failing because Medallia was badly run or Qualtrics was overleveraged. The deeper problem is that AI is eating the entire value proposition on two key dimensions.
AI tools are rapidly becoming your insights engine. If an LLM can synthesize every support ticket, review, transaction, and conversation into a coherent picture of customer sentiment without a survey, a lengthy software implementation, or a six-figure license, then what exactly are you paying for? And while you may not trust Copilot to do all this (yet), any future enterprise insight investments will need a clear impact ROI story.
Robot customers won’t answer your surveys. Machine-mediated interaction will be one of the defining CX shifts of the coming decade. We can already see solid signals of the rise of a world where AI agents plan, purchase, escalate, and switch providers with minimal human involvement, and where dissatisfaction is expressed by algorithms rather than people. Your website is probably feeling the early symptoms: The discovery and feedback loops that fed insight platforms are collapsing as AI answers replace clicks. When agents become the dominant customer, your surveys have no recipient.
Welcome To Systems Of Action
Most CX teams are built around insight production, and that’s a depreciating asset. The measurement apparatus, the survey programs, the enterprise vendors — none of that disappears tomorrow. But the value is migrating to whoever owns the action layer. The vendors are already naming the next layer. Medallia, Forsta, Qualtrics, and more all want to move “from insight to action.”
But many CX leaders and teams haven’t caught up yet.
The terminology is nascent but resonant. Salesforce published it in February as an architectural principle: moving enterprise software “from systems of record into systems of action,” where agents don’t surface recommendations; they execute decisions within governed parameters. Zoom positioned its entire agentic platform around the same idea in March, calling it a “system of action” designed to turn “every meeting, call, and customer interaction into a trigger for workflow automation.” Gong thinks that AI systems of action will replace traditional CRM.
Investors are backing outcome-oriented platforms, which will influence what you buy — and how you pay for it. Sierra, founded in 2023, valued at $10 billion, and growing at 400% year on year, doesn’t sell insight. It sells resolutions. Decagon handles three-quarters of customer contacts without a human and charges you nothing for the ones it doesn’t resolve. TheyDo raised $50 million on the proposition that the journey is the operating model, not the map on the workshop wall.
Predictive will be the next wave. The stickiest CX software and the most backable CX teams of the next decade won’t be the ones that tell you what your customers used to feel. They’ll be the ones that spot the signals that preindicate dissatisfaction, risk, churn, revenue loss … and do something about it autonomously, within human-defined guardrails, and with a clear audit trail, all before you’ve finished reading the dashboard.
Tomorrow’s question won’t be whether your platform gives you good insights. It will be whether you can act on them — and, increasingly, whether it needs you in the loop at all.
