The Research Gaps No One Talks About

Research programs are designed around known questions. That is, by definition, their limitation. You build an interview guide, you recruit experts, and you ask the questions you came in believing were the right ones. The process is rigorous. The output is often incomplete.

There are three types of research gaps worth naming. The first: known unknowns — questions you know you need to answer and have built into your research design. The second: unknown unknowns — questions you don't yet know you need to ask, because your current frame of reference doesn't accommodate them. The third: structural blindspots — questions you can't ask, not because of oversight, but because of how your expert network is constructed.

This article is about the second and third types. The first type takes care of itself — if you know you don't know something, you can design around it. The gaps that produce real research failures are the ones you never thought to look for.

Unknown Unknowns: The Questions You Don't Know to Ask

Unknown unknowns form through a predictable mechanism. The research team arrives at a project with an existing model of the market — built from public data, prior deals, and internal conviction. That model shapes the interview guide. The interview guide shapes which questions get asked. Which means the entire research program is, at its foundation, a test of what the team already believes.

This is not a failure of intelligence. It is a structural feature of how knowledge works. You can only ask questions from within a frame. The problem is that the most important answers often exist outside that frame.

Consider a team researching the competitive position of a mid-market SaaS company. The interview guide is built around the company's known competitors — the ones named in analyst reports, the ones that come up in sales calls, the ones the team has already profiled. Every expert call confirms or challenges the team's view of that known competitive set.

The unknown unknown: a smaller competitor — not yet on anyone's radar — is quietly piloting a consumption-based pricing model with a cohort of mid-market buyers. None of the experts the team spoke to mentioned it, not because they were withholding information, but because it genuinely hadn't reached the level of market visibility where people were talking about it in the contexts those experts inhabited.

The fix is not to ask more of the same questions. It is to build open-ended discovery questions into every call — questions designed to surface what isn't in the interview guide. Two that consistently perform: "What's the most important thing about this market that most outsiders get wrong?" and "What's changed in the last six months that hasn't been publicly reported yet?" These questions create the space for experts to surface information that doesn't fit neatly into your existing model.

That question — "what would I need to believe to be wrong" — is structurally different from most interview questions. It invites the expert to think adversarially about the thesis, rather than as a witness confirming or denying specific claims. It consistently surfaces unknown unknowns because it asks the expert to step outside the frame the interviewer has constructed.

Structural Blindspots: When Your Expert Pool Is the Problem

A structural blindspot is different from an unknown unknown. It isn't a question you haven't thought to ask — it is a category of answer that cannot enter your research because of who you are talking to. The gap is in the network, not the guide.

The most common structural blindspots follow predictable patterns. An expert pool drawn exclusively from Western multinationals will systematically miss local and regional market dynamics — not because those dynamics are hidden, but because they are simply not part of the professional experience of the experts being consulted. An expert pool drawn exclusively from former C-suite executives will miss ground-level operational reality — because the view from the thirty-fifth floor is genuinely different from the view on the distribution floor.

Other common structural blindspots: expert pools that skew toward English-language speakers in sectors where multilingual market participants hold important information; expert pools drawn entirely from large incumbents, which systematically miss the challenger and startup perspective; expert pools recruited through a single professional network, which tend to reflect the consensus views of that network rather than the full distribution of market opinion.

That quote describes a program that was executed well by conventional measures — 14 calls is a substantial sample, the experts were credible, the research was systematic. The problem was not process failure. It was that the expert pool shared a structural characteristic — proximity to multinational corporate networks — that made it constitutionally unable to describe what was actually happening at the local market level.

The practical tool for catching this before it matters: an end-of-program expert pool audit. After every program, map the experts by five dimensions — geography, function (finance, operations, sales, technical), company type (large incumbent, mid-market, small/startup), language, and seniority. If the distribution is heavily skewed on any dimension, that skew is a structural blindspot. It tells you what the research probably got right and what category of answer is almost certainly missing.

Doing this audit after the program feels too late — but it isn't. The findings from this program become the recruitment brief for the next one. And doing it consistently, across programs, builds an institutional awareness of which types of experts your network tends to over- and under-index on.

The Recency Gap

Primary research has a structural recency problem that is easy to overlook: expert knowledge decays. A former CFO who left a company three years ago carries a detailed and accurate model of that business — as it existed three years ago. The customer mix, the cost structure, the go-to-market motion, the technology stack: all of it reflects a point in time that may no longer bear much resemblance to the present.

The recency gap widens in fast-moving sectors. In consumer technology, e-commerce, and fintech, a three-year-old view of a company's unit economics may be as misleading as no view at all — because the underlying market, competitive dynamics, and cost structures have shifted substantially in the interim. An expert who was deeply informed about a business in 2022 may be describing an entity that no longer exists in any operationally meaningful sense.

The practical framework: for every claim that comes from an expert who left their role more than 18 months ago, treat it as historical intelligence rather than current intelligence. Require corroboration from a more recent source — a current employee, a customer with a recent renewal conversation, a competitor with current visibility into the market — before treating the claim as a live data point. The 18-month threshold is not arbitrary; it roughly tracks the cycle time for meaningful operational change at most businesses.

This does not mean excluding experts with older tenure — historical context is often exactly what a research program needs. It means tagging claims by recency and not allowing historical intelligence to stand in for current intelligence when the two are being treated as equivalent.

The Confirmation Gap

The confirmation gap is the most socially awkward gap to name, because it implicates the research team directly. It occurs when teams — unconsciously, usually — select experts who are likely to agree with the thesis. The result is a research program that produces a high volume of confirmatory signal and a low volume of challenge.

The mechanism is partly framing and partly network. On the framing side: experts are sometimes selected based on their public views. If a team is building conviction on a thesis and discovers an expert who has written or spoken favorably about the thesis topic, that expert gets called. The expert pool drifts toward people who already agree. On the network side: teams call back the experts who were helpful on previous programs. "Helpful" often means "agreed with us and gave us good quotes." The most challenging expert — the one who pushed back hardest — sometimes doesn't make the callback list.

The diagnostic signs of confirmation gap are recognizable after the fact: every expert in the program is bullish; no call produces a genuine contradiction; every call summary notes that the expert "reinforced the thesis." A program where every single call goes one direction is not a well-designed program — it is a program that was, perhaps unconsciously, designed to go one direction.

The structural fix is simple to state and requires institutional discipline to execute: require every research program to include at least one "bear case" expert — someone who has publicly or privately argued against the thesis, or who has a reason to be skeptical of the bull case. This is not about achieving artificial balance. It is about ensuring the research program has genuinely stress-tested its conclusions rather than just accumulated support for them.

In practice, the bear case expert is often the most valuable call in the program — not because they are right, but because the best experts make you defend your thesis in ways that either strengthen your conviction or reveal cracks you hadn't noticed.

Surfacing Gaps Before They Become Mistakes

Research gaps are not embarrassing oversights. They are inevitable features of how knowledge is structured. Every research program begins with a frame. Every frame has edges. The gaps live at the edges — in the questions the frame didn't accommodate, the experts the network didn't include, the time periods the roster didn't cover, the perspectives the expert pool was constitutionally unable to represent.

The teams that produce better research are not the ones who eliminate gaps — no research program eliminates gaps. They are the ones who surface gaps explicitly, before those gaps have a chance to become investment mistakes. A gap you know about is a manageable risk. A gap you don't know about is a liability.

A practical end-of-program ritual that takes less than twenty minutes: list the three most important things you didn't know at the start of the program that you know now. Then list the three most important things you still don't know. The first list is evidence that the program worked. The second list is your next research brief.

Research programs that end with a clean summary of confirmed findings are satisfying. Research programs that end with an honest accounting of what remains unknown are useful. The difference between those two orientations is, over time, the difference between research that builds conviction and research that generates intelligence.