Your Sales Playbook Is Failing—Here’s What Survives AI

“One thing that should be learned from the bitter lesson is the great power of general-purpose methods…We want AI agents that can discover like we can, not that simply contain what we have discovered.”—Rich Sutton, The Bitter Lesson

I can’t pinpoint exactly when business culture shifted from principles to methodologies, but the inflection likely came mid-20th century. The late 1800s were steeped in principled thinking: Ralph Waldo Emerson’s The Conduct of Life emphasized pragmatic rules for the good life; William James’s 1890 Principles of Psychology introduced enduring ideas like the “stream of consciousness." Two decades later, Alfred North Whitehead and Bertrand Russell’s Principia Mathematica attempted to ground mathematics in logic—an explicit bid to found expanding knowledge work on reasoned first principles.

Business and economics mirrored this spirit. W. Edwards Deming and Alfred Marshall anchored practice in principles tested by real-world impact. Post-WWII Japan famously absorbed Deming’s quality principles: measurable improvements in quality and productivity lowered costs and sparked global demand.

Even into the 1990s, principle-centric frameworks thrived—think The 7 Habits of Highly Effective People. The turn away from principles roughly coincided with the internet’s ubiquity. When answers became instantly retrievable, the patience and discipline required to build principled systems of knowledge and behavior began to erode.

A sales lesson from the boiler room

Sales is my home field. In high school, I sold long-distance plans at a telemarketing shop. New hires received a one-week boot camp in methods—objection handling, closing, and scripted dialogue. I promptly crashed and burned. What I saw in top performers wasn’t method magic; it was principles, things like tone, volume, pacing, and mood. Voice modulation did the real work. Sellers used different methods, but they shared the same underlying control of rhythm and feel. That was my first “principles over methods” eureka.

Two decades later, the principles of selling haven’t changed. What has persisted, unfortunately, is a widespread trap: over-indexing on methodologies, tactics, and ad-hoc hacks. Today’s buyers expect more.

Armed with AI, they often know as much—or more—than the seller. Sellers must rethink how they go to market and what value they bring to each interaction. To connect the dots on principles, it helps to review how information asymmetry in sales has evolved.

A brief history of sales asymmetry

Three eras align with technology shifts:

1. Pre-internet

2. Internet

3. AI

Pre-Internet era. Sellers held the power because they held the information—product details, use cases, pricing context. Households had phones, encyclopedias, and perhaps a library down the road. Sales culture adapted accordingly. This was the era dramatized by Glengarry Glen Ross and Boiler Room. Tactics like “always be closing” flourished. Plenty of reputable sellers practiced frameworks like SPIN and Miller Heiman, but the bar for informational rigor was low. Confidence and scripts often carried the day.

Internet era. The web leveled the field. Buyers gained user reviews, competitor visibility, and frictionless discovery. Relationship and consultative sales surged. If buyers would “shop around,” sellers needed to be more useful and more likable. At the era’s peak came The Challenger Sale, which raised the bar with “teach, tailor, take control.”

AI era. November 2022 (the ChatGPT moment) flipped the script again. For the first time, buyers could easily arrive with more information than sellers. The asymmetry is inverted. Combined with an explosion of new capabilities, this shift left many sellers without a reliable playbook. As of August 2025, the profession is still re-finding its footing.

AI has exposed a stark divide:

• Group 1 — Sales Pros (SP): Exemplars who treat selling like a craft and a sport. They excel at thinking and learning—the substrate of success in the AI era.

• Group 2 — Sales Associates (SA): Order-takers anchored in earlier-era methods. They rely on ad-hoc experience, relationships, or a single dimension of skill.

  
  

Knightian uncertainty: making sales robust

The following image was presented by Stanley, Lehman, et al, in a recent research paper entitled, “Evolution and The Knightian Blindspot of Machine Learning.” 

The authors echo Rich Sutton’s Bitter Lesson: overfitting systems with hand-crafted rules underperforms letting general systems learn. Knightian uncertainty (the “unknown unknowns” in economics) offers a useful lens for both AI and sales. Through that lens, evolution—not a single formalism—is the optimal principle-based strategy for robustness.

The two primary sales strategies, like the machine learning strategies above, differ in the following ways:

(a) Diversify and filter → robustness rises. Run many “plays” (messages, offers, sequences, pricing frames, ICPs). Apply periodic KU filters—stress-tests against shifting conditions (new buyer, budget freeze, legal hurdle, competitor move). Keep what still works; prune what doesn’t. Over time the surviving portfolio becomes robust across contexts—achieved via principles, not rigid rules.

(b) One-model training → formalism mismatch.The classic enablement move: collect problems → train everyone on one “optimal” method (SPIN, Challenger, MEDDICC) → deploy. When the world shifts, the method’s assumptions break. Win rates fall because the team was optimized for yesterday.

Deploying an adaptive sales model

A sales system for the AI era should be principle-driven and open-ended, enabling hyper-learning across the cycle. Operationalize it like this:

• Portfolio over playbook. Run parallel hypotheses (ICP, channel, talk track, value metric, offer). Methods are ingredients—not doctrine.

• Selection pressure beats compliance. Instrument leading indicators (reply rate, first-meeting rate, EB [economic buyer] engagement by day 14, stage-to-stage lift, cycle time). Scale survivors; kill the rest—regardless of “the method.”

• Context before content. Teach reps to identify the game they’re in (economic buyer, procurement posture, risk tolerance, problem maturity) and then adapt. Qualification frameworks are guardrails, not rails.

• Outcome-anchored conversations. Center on business value/customer economics (time-to-value, risk reduction, margin/NIM/loss-ratio/cycle-time deltas). Balance the information asymmetry. Methods help only insofar as they expose real outcomes.

• Fast learning loops > “best practice.”Weekly micro-experiments, A/B tests, and post-mortems. Codify survivors into “micro-plays”; retire 20–30% quarterly to avoid overfitting.

• Local autonomy within simple rules. Give freedom to improvise inside a few non-negotiables: ethics, mutual next steps, access to power, quantified value, multi-threading.

• Antifragile enablement. Red-team deals (simulate CFO pushback, budget cuts, security stalls). The goal isn’t perfect calls—it’s resilience under variance.

  
  

Are methodologies obsolete?

No. Keep them in the toolkit as shared vocabulary and checklists, not as doctrine. Use SPIN to deepen questioning, MEDDICC to sharpen qualification, Challenger to reframe—deploy contextually, not universally.

Bottom line: Build and nurture a principle-driven, evolutionary sales system—diversify, measure, select—so your team grows robust to change. In an AI-enabled market, this will consistently enable sales performance beyond any single “optimal” methodology.