Knowledge-Worker Chatbots: Your Wins Are Real. The Missed Value Is Bigger.

Most firms report chatbot gains—faster drafting, cleaner summaries, less grunt work. Good. But if your knowledge-worker outputs aren’t reusable, verifiable, and measured, you’re still below the line. 

The lens: 

• X: Know how → Don’t know how 

• Y: Know what’s possible → Don’t know what’s possible 

The top-right is don’t know & don’t know—maximum missed value. 

Where Value Actually Shows Up 

• Full Value (bottom-left): Teams apply well-known patterns, produce handoff-ready outputs, and hit business targets. 

• Everywhere else: effort without leverage—polished prose, stalled pilots, or scattered one-offs. 

The Executive Problem to Solve 

It’s not tooling. It’s an operating system for patterns, templates, verification, and measurement—owned by functions, not IT. 

30-60-90, in Business Terms 

30: Make it obvious what’s possible 

• Each function publishes a top-10 pattern gallery (evidence-backed brief, decision path from policy, discovery prep, meeting kit, spreadsheet assistant, risk review). 

• Leaders nominate one high-volume workflow per function to improve in 30 days. 

60: Teach the craft once 

• Five habits: clear role and objective, constraints, structured outcome (table/checklist), progressive refinement, verification before handoff. 

• Convert the best chats into templates with examples; store them where people work. 

90: Prove business movement 

• For each workflow, track one metric that matters and report weekly: 

• Cycle time, error/rework, conversion, backlog, CSAT/ESAT. 

• Keep what moves the number; rework or retire the rest. 

The Metric Menu (steal this) 

• Sales/CS: discovery prep time, proposal turnaround, win-rate lift on templated proposals, reopen rate. 

• Marketing/PMM: campaign brief cycle time, revision count, launch defect rate. 

• Finance: month-end close tasks completed on first pass, variance commentary cycle time. 

• Legal/Compliance: policy decision throughput, exceptions per 100 decisions, review time. 

• HR/People Ops: recruiting slate prep time, offer cycle time, policy question deflection. 

Report three numbers per workflow: baseline → current → target, plus % of outputs passing verification on first review. 

Mini Case (Numbers First) 

Customer Service, 8-week pilot 

• Metric: Average handling time (AHT) for returns. 

• Approach: Standard “policy-to-decision” pattern with references and a ready-to-send draft. 

• Results: AHT −41%, reopens −23%, CSAT +6 pts, reviewer first-pass approvals +31%. 

• Lesson: The model didn’t change; the operating model did. 

Antipatterns to Kill 

• Celebrating time-saved anecdotes without a counterpart metric (quality, rework, throughput). 

• Proudly showcasing one-off “prompt heroes” instead of shared templates. 

• Treating chatbots as a personal assistant, not a workflow asset. 

• Delegating ownership to IT; this is a line-of-business responsibility. 

The Close 

If your teams can’t point to a handful of named templates that others reuse—and a simple weekly dashboard that shows business movement—you’re not in the Full Value quadrant yet. You don’t need a new platform. You need leadership attention to patterns, ownership, and measurement. 

Do this now: pick one workflow per function, publish the pattern, add verification, and move one number in 30 days. Then do it again.