NO. 024 · OCTOBER 13, 2026 · THE GREAT SORTING
This week, The Great Sorting looks at two scorecards, one that ranks feedback platforms and one that grades the AI agents now talking to your customers. Both measure something real, and neither measures the outcome.
🟣 THIS WEEK IN THE FIELD
Qualtrics Spent $6.75 Billion and Was Marked Down. Medallia's Owners Got Nothing and Won.
Forrester demoted Qualtrics after its $6.75 billion purchase of Press Ganey Forsta, and named Medallia sole Leader weeks after its owners handed it to creditors. Both placements are defensible. The criteria reward the best version of the feedback model, while the Leader's own strategy chief says most of his customers never got past basic surveys. Renewing this quarter? You're choosing the cleanest dirty shirt in the drawer.
Governance Tells the Agent What Not to Do. Nothing Tells It Whether It's Working.
OpenAI's new agent platform leads with 75% of support issues resolved without a human. Its policies, permissions, thresholds and graders all govern what the agent does. None of them measures whether the customer renews. Governance is the brakes. Steering needs a forward read of each account, because an agent retuned weekly will get very good at the wrong number long before the renewal shows it.
🟣 THE BREAKDOWN
Put this week's two articles side by side and they describe the same mistake at two scales. Forrester scores feedback platforms on ingestion, data models, identity stitching and governance, all of them skills of collecting and analyzing feedback. It rewards the best version of the existing model, then marks down the vendor moving its north star from NPS to customer lifetime value, which is a move toward Forrester's own first rule for a top-level metric: it should lead financial results.
Agent platforms make the same mistake faster. Resolution rate, policy compliance and grader scores are recorded the moment the agent acts. Renewal, expansion and price arrive on the contract cycle, months later, and nothing in the stack connects the two. An improvement loop that retunes weekly will optimize whatever it is shown, and right now it is shown the action.
Inside an enterprise account base, the test is simple. For every system that touches customers, ask what it is scored on and when that score ever meets the renewal. If the answer is never, it is measuring activity, not the business.
Is your agent scored on the action or the outcome? Reply to this email.
🟣 THE FIELD TACTIC
Three ways to put an outcome behind every customer agent
- Write down the number each agent is optimizing. Resolution, deflection, handle time: name it, then name the business outcome it is meant to serve. If nobody can draw the line between the two, the improvement loop has nothing worth learning from.
- Check the agent against the renewal clock, not the weekly one. Keep a comparable group of accounts outside the agent's offers or workflows, and compare renewal and expansion between the two groups over a full contract cycle.
- Ask your vendors where the outcome lives. At renewal, ask each platform how its output connects to retention and revenue, and what it has changed in the past year to measure that. A ranking won't tell you; the answer will.
🟣 FOR ENTERPRISE TEAMS
OCX Cognition's Customer AI Analyst gives you a forward read of every account in your base, built from what customers do as well as what they say. That gives your teams, and your agents, something to be measured against besides ticket closure. If you'd like to see it on your own data, we'll walk you through it.
🟣 GO DEEPER
The Customer AI Field Guide is a practical guide for CX, CS and RevOps leaders on turning customer intelligence into revenue action.
The Customer AI Masterclass is a self-paced certification, 8 modules over about 8 hours, where you earn both NPS® Certification and Customer AI Certification. Learn more.
Learn more about certification →
Until next week,
Richard Owen
Co-founder and CEO, OCX Cognition
Joined and then led the team that co-created NPS.
Know someone who should read this? Forward it, or they can subscribe here
Responses