When everyone has the same AI, proprietary data becomes the advantage

By Felipe Ochoa, CEO of IncQuery

For a long time, one of the most valuable skills in consulting was knowing how to find information others couldn't. I remember working on a Bain project where finding one useful industry data point meant digging through obscure sources and tracking down the person who actually had the answer.

There was even a story in Bain lore about a team trying to estimate beer brewery market share in a country lacking reliable data. Since the information wasn’t readily available, the team visited multiple bars and counted the individual bottle caps from their dumpsters to proxy the true shares. Whether the challenge involved tracking down a hard-to-find source, manually building a dataset, or literally counting physical evidence, creating an information advantage often required considerable ingenuity and effort.

Today, AI can often surface that same information in minutes. That is a major productivity gain, but it also creates a new challenge: if you can find the information that quickly, so can everyone else.

AI is commoditizing access to existing information. The competitive advantage is shifting from finding what is already known to generating evidence that does not exist yet.

‍

Finding information is no longer the moat

A significant portion of traditional knowledge work has always been information retrieval. Consultants searched analyst reports, databases, filings and trade publications, while experts often created value by knowing things that were difficult to find elsewhere.

I remember one project where I needed to estimate growth in miles of train track with certain characteristics. Cobbling together that timeseries required scraping back issues of a trade publication and fishing out individual data points from dozens of PDFs. It was painful work over multiple days, but it was valuable precisely because the information was not readily available anywhere else. Today, a capable AI system could likely achieve the same thing with just a couple of prompts from a consultant and some tokens to burn.

Generative AI is collapsing much of that work. Teams can now understand unfamiliar markets, synthesize public information and identify relevant sources far faster than before.

That is good for the industry. But as information becomes easier to access, it also becomes less differentiated.

If multiple firms are evaluating the same market using similar AI systems and similar public sources, they are increasingly starting from the same information base. The internet is not a proprietary research strategy.

That said, LLMs currently struggle with effective source criticism and reconciliation of conflicting data points. Those are core consulting skills where AI lags human performance, but that is an analytical limitation, not necessarily a permanent one. As models improve, they will likely become better at comparing conflicting evidence, identifying inconsistencies and reasoning about source quality; there are benchmarks for this already. Expect analysis and synthesis of secondary data to become less differentiating over time.

‍

The advantage is what you can create

Many of the most important business questions cannot be answered by searching harder.

An investment team may want to know how customers would react to a price increase. A strategy team may need to understand what actually drives switching behavior. A consultant may need to quantify how purchasing criteria differ across customer segments.

AI can summarize what has already been written and help identify hypotheses. But it cannot retrieve an answer that has never been measured.

Someone has to create the evidence.

That is where proprietary primary research becomes more valuable. A survey or structured research program can generate information specific to the decision at hand, producing findings that competitors cannot simply retrieve from the same public sources.

In that environment, proprietary data becomes a competitive asset.

‍

Better questions become more valuable

As information retrieval gets cheaper, knowing what to ask becomes more important.

The higher-value work is identifying what remains uncertain after AI has synthesized everything that already exists. Which assumption is still untested? Where is the investment thesis relying on anecdote? What evidence would actually change the decision?

AI can help formulate those questions, but deciding which uncertainties matter still requires judgment.

In that sense, AI does not reduce the need for primary research. It can make the need more obvious. Once everything that is already known has been summarized, the gaps become much easier to see.

‍

Proprietary data matters even more in an agentic world

This dynamic will become more important as organizations rely more heavily on AI agents to support analysis and decision-making.

Give an agent the same public information available to every competitor, and there is little reason to expect a differentiated conclusion. Give it proprietary evidence about your customers, market, or investment target, and the calculation changes.

The advantage may not belong to the firm with the biggest team or even the best general-purpose AI model. It may belong to the firm that can continually supply its systems with better evidence.

From information advantage to evidence advantage

AI is making research dramatically faster, and that is a good thing. Consultants and investors should use it to understand markets, synthesize existing information, and accelerate analysis.

But speed alone will not remain a durable advantage when everyone has access to similar tools.

The more capable AI becomes at finding, evaluating and synthesizing what the world already knows, the more valuable it becomes to know something the world does not.

For years, the advantage belonged to the people who knew where to find the answer. Increasingly, it will belong to the people who know which new questions are worth answering.

‍