Consensus Is the Riskiest Strategy in the AI Era

Consensus feels safe. It is the riskiest thing you can do. AI regresses to the mean. Your organization does too, approving what many agree on, and the mean was never going to grow.

Consensus feels safe.

It is the riskiest thing you can do.

AI regresses to the mean. Your organization does too, approving whatever the most people agree on. The result never reaches the top of the distribution, the small share of brands that actually grow. And in the AI era, this pull toward the middle is getting stronger, not weaker.

When Everyone Agrees, That Should Scare You

Sit in the room. The strategy gets presented, discussed, softened. Every objection files down another edge. By the time it clears, nobody disagrees. It feels like alignment. It feels like a win.

It is the sound of a strategy regressing to the mean.

What survives a consensus process is the lowest common denominator, the version no single person will fight. That version is defensible. It is comfortable. It is also, almost by construction, the option least likely to win, because a distinctive bet is exactly the thing some people in the room are not yet convinced of. Unanimous approval and competitive advantage rarely travel together.

Why AI Makes This Worse

A large language model is trained to predict the most likely next answer, given nearly everything ever written. So its output is a fluent, confident average of the consensus. That is a feature for most tasks. It is a trap for strategy.

Because everyone is now querying the same models. The safe, plausible answer AI hands your team is the same safe, plausible answer it hands your competitor's team. It pulls all of you toward the same center of gravity at the same moment. The tool that feels like an edge is quietly a homogenizer.

So you get a double dose of the mean. AI hands you the plausible answer. Your organization then polishes it until everyone nods. Two forces, both pointing at the middle, and the middle is a graveyard, roughly 5% of brands grow, 5% of products survive, 5% of campaigns actually move sales. I unpack that pattern in why only 5% succeed. The other 95% did not fail from lack of best practice. Most of them did everything the consensus recommended.

Best Practice Stopped Being an Edge

For decades, competitive advantage often came from knowing the best practice before others did. The playbook, the framework, the benchmark. Access to that knowledge was scarce, so having it was worth something.

That scarcity is gone. Best practice is now universal and instant. Any leader, any competitor, any intern with a chatbot can retrieve the current consensus on pricing, positioning, or go-to-market in seconds. When everyone has the same recipe, the recipe is no longer the advantage.

Growth now comes from owning a customer truth that neither AI nor your own experts can produce. Not a better-worded version of the consensus. Something the models have never seen, because it lives in your customers, not on the internet.

The One Input a Model Can't Give You

AI can retrieve everything that has been written. It cannot tell you why your customers actually buy. That is not published anywhere. It has to be measured.

And it cannot be measured by asking, because people cannot reliably explain their own decisions. The real drivers are subconscious, and the conscious answer is a story written after the fact. Getting to the truth means deep implicit research into what actually moves the choice, and Causal AI to separate the true cause from everything that merely correlates with it. That output is proprietary by nature. A competitor cannot copy it from a public model, because it was never in one.

This is the asymmetry worth internalizing: advice regresses to the mean, measured customer truth does not. One is infinitely copyable. The other is yours alone.

So What Is Consulting For Now?

This is the uncomfortable question for my own industry. How is management consulting different in the AI era?

The old value was aggregation, gathering best practice from many places and importing it into yours. AI now does that part instantly, and for nothing. A consultant who only repackages the consensus is selling what the model gives away.

The remaining value is the opposite move. Producing proprietary, causally validated insight the client could not get from any model, and standing behind the decisions built on it. Not "here is what everyone does." Instead: "here is what is uniquely true about your customers, and here is the bet only you can make because of it." That, honestly, is the gap SUPRA was built to fill, and the whole argument sits in my book THE TOP 5%.

So before your next big decision, ask one question. Is this strategy distinctive because it is true, or agreeable because it is average? If the whole room already nods, you may not be aligned. You may just have regressed to the mean.

How do you keep a strategy from collapsing into consensus? I am reading the replies.

Frequently asked questions

Why is consensus a risky basis for strategy?

Because agreement selects for the average. When a strategy is polished until everyone nods, what survives is the lowest common denominator, the version no one objects to. That version is comfortable and defensible and almost never a winner. Growth comes from a distinctive bet most people are not yet convinced of. If everyone already agrees, the advantage has usually been competed away.

What does it mean that AI regresses to the mean?

Large language models predict the most likely next answer given everything ever written, which makes their output a sophisticated average of the consensus. It is fluent, plausible, and representative of best practice, which is exactly why it cannot hand you a distinctive edge. Everyone querying the same models gets the same center of gravity.

Where does competitive advantage come from in the AI era?

From proprietary truth about your customers that neither AI nor your own experts can produce from memory. Best practice is now universal and instant, so knowing it is no longer an edge. What stays scarce is a validated, causal understanding of why your specific customers buy. That truth has to be measured with deep implicit research, not retrieved.

How is management consulting different in the AI era?

The old value of consulting was aggregating best practice and importing it, which AI now does instantly and for free. The remaining value is the opposite: producing proprietary, causally validated insight the client could not get from a model, and standing behind the decisions built on it. Advice regresses to the mean; measured customer truth does not. The AI Diagnostic is one way to see where your strategy has drifted to the average.

Dr. Frank Buckler is the founder of SUPRA and a pioneer in Causal AI for marketing. He has applied implicit research methods across FMCG, pharma, financial services, and insurance for over 25 years.