The Selection Matrix: A Rubric for Evaluating Founders Without Bias

A program director I know once described her selection process to me like this:

"We have six reviewers. Each reads the applications independently and scores them 1-5. Then we average the scores and take the top 30 for interviews."

"What are they scoring on?" I asked.

She paused. "Potential, mainly. Team strength. The idea."

"And how do you define those?"

Another pause. "We kind of know it when we see it."

I hear some version of this from nearly every program I talk to. They have a process—ratings, reviewers, interview rounds. But underneath the process, there's no shared definition of what they're actually looking for. Every reviewer is using their own intuition about what "potential" means. And intuition, when it comes to founder evaluation, is heavily influenced by pattern matching against the founders we've seen succeed before.

Which means the process is systematically biased—even when the reviewers are well-intentioned and diverse.

Here's how to build something better.

Why "We Know It When We See It" Is Broken

Let me be direct about what "gut feel" evaluation actually selects for.

It selects for founders who look like successful founders we've seen before. It selects for confidence over competence. It selects for polish over substance—because a well-crafted application or a smooth interview performance triggers the "this person has it" feeling more reliably than evidence of actual founder capability.

It also selects for familiarity: founders from networks we recognize, educational backgrounds we respect, geographic and cultural contexts we understand.

None of this is malicious. It's how human pattern recognition works. But pattern recognition trained on a non-representative sample produces non-representative selections.

The solution isn't to abandon judgment. It's to make your criteria explicit, so your judgment is applied consistently and can be examined and improved.

Building the Selection Matrix

Four Core Dimensions

A rigorous selection rubric evaluates founders on dimensions that actually predict program success—not on dimensions that correlate with "looking like a successful founder."

Here's the framework I recommend, built around four core dimensions.

Dimension 1: Problem-Market Understanding

What you're evaluating: Does this founder have a genuine, evidence-based understanding of the problem they're solving and the market they're entering?

This is the most important dimension at early stages, and the one most commonly overlooked. Founders who deeply understand their problem and market are more likely to iterate effectively, find real customers, and make good decisions when their initial assumptions prove wrong.

Indicators of strength (score 4-5):

• Can articulate the problem from the customer's perspective, not just from their own experience

• Has done primary research: customer interviews, surveys, or domain expertise

• Understands the competitive landscape—not just "there's no solution," but specifically how existing solutions fall short

• Can name specific customers or customer segments, not generic demographics

Indicators of weakness (score 1-2):

• Problem is vague or generic ("small businesses struggle with X")

• Evidence is anecdotal or second-hand ("I read that this is a problem")

• No awareness of existing solutions or why they fail

• Market size claims are theoretical, not grounded in real customer behavior

Dimension 2: Founder-Problem Fit

What you're evaluating: Does this founder have a genuine, specific reason to be working on this problem—beyond general interest or market opportunity?

Founder-problem fit is different from founder-market fit. The question isn't "are they capable of working in this space?" It's "why is this the specific problem they're obsessed with solving?"

Founders with authentic founder-problem fit are more resilient. They don't quit when things get hard, because the problem matters to them personally.

Indicators of strength (score 4-5):

• Has lived experience with the problem or is deeply embedded in the affected community

• Has domain expertise that gives them unique insight into the solution space

• Can articulate why they specifically are working on this—and the answer goes beyond "it's a big market"

• Shows genuine urgency or passion that comes from caring about the outcome, not just the opportunity

Indicators of weakness (score 1-2):

• Working on the problem primarily because they identified a market opportunity

• No direct connection to the problem or affected community

• Could easily pivot to a different problem without losing sleep—suggesting shallow commitment

• Motivation is primarily financial or credential-seeking

Dimension 3: Learning Velocity and Coachability

What you're evaluating: Does this founder learn fast, update their views when given new evidence, and engage productively with feedback?

This is the dimension that most predicts whether a founder will benefit from your program. A founder who thinks they already have all the answers won't get much from mentorship, workshops, or peer learning. A founder who's hungry to learn and quick to integrate new information will extract value from everything.

Indicators of strength (score 4-5):

• Can point to specific moments where they changed their mind based on evidence

• Talks about what they've learned, not just what they've built

• Engages with challenging questions with curiosity rather than defensiveness

• Has actively sought out feedback and can describe how they used it

Indicators of weakness (score 1-2):

• Defensive when challenged on their assumptions

• Can't name anything significant they've changed their mind about

• Dismisses competitor products or customer feedback that contradicts their thesis

• Treats the program as validation-seeking rather than learning-seeking

Dimension 4: Execution Evidence

What you're evaluating: Has this founder actually shipped something, done something, or built something—regardless of whether it "worked"?

Execution evidence is not the same as traction. A founder with zero revenue who has run 50 customer discovery interviews, built a prototype, and tested three different pricing models has strong execution evidence. A founder who has a polished pitch deck and a business model canvas has weak execution evidence.

You're looking for a bias toward action over planning.

Indicators of strength (score 4-5):

• Has built something real, even if imperfect: an MVP, a prototype, a landing page with sign-ups

• Has talked to real customers, not just brainstormed about them

• Has run experiments: A/B tests, pricing tests, distribution experiments

• Moves fast and ships something rather than perfecting before launching

Indicators of weakness (score 1-2):

• Idea is still in the planning or research phase with no tangible output

• Application is heavy on strategy, light on evidence of doing

• Long timelines for basic milestones (6 months to build a landing page, for example)

• Waiting for everything to be "ready" before testing

The Scoring System

For each dimension, score 1-5:

• 5: Exceptional. Well above expectations for program stage. Would be outstanding in any cohort.

• 4: Strong. Meets expectations with clear evidence. A reliable indicator of quality.

• 3: Adequate. Some evidence, some gaps. Proceed with questions in this area.

• 2: Weak. Significant gaps. Major concern. Would need interview to determine if this is a red flag or a presentation issue.

• 1: Disqualifying. Evidence of the opposite: no customer contact, defensive to feedback, can't articulate the problem.

Total score: 4-20

• 16-20: Strong admit candidate

• 12-15: Interview with focused questions on gaps

• 8-11: Probable decline, but worth a second reviewer if something interesting

• Below 8: Decline

The threshold for each stage will depend on your cohort size and application volume. But the rubric should be calibrated consistently across reviewers.

Bias Checks:

Building Them Into the Process

A rubric reduces bias—but doesn't eliminate it. Here are the structural checks to build into your process.

Blind first-round review

In the first review pass, remove or obscure identifying information: founder name, university attended, previous employer, geographic location. Reviewers evaluate the substance of what's in the application, not the signals we associate with "credentialed founders."

This isn't always practical, but even partial blinding (removing names and schools) significantly reduces bias in first-pass scoring.

Calibration sessions before each cycle

Before reviewers score applications, run a calibration session. Review 3-5 sample applications together, score them independently, compare scores, and discuss the gaps. This surfaces differences in how reviewers interpret criteria—and creates shared understanding before the real review begins.

Multi-reviewer scoring with structured discussion

Never let a single reviewer's score determine an outcome. Require at least two independent scores. When scores diverge significantly (3+ point gap on a dimension), require a discussion before finalizing.

The discussion isn't about who's right—it's about surfacing what each reviewer saw and why. Often, one reviewer noticed something the other missed.

Track your selection patterns over time

After each cohort, analyze who you selected and who you didn't. Look at patterns:

• What's the demographic breakdown of admits vs. declines at each stage?

• Which criteria are highest-scored in your admit pool? Which are lowest?

• Are there criteria where your highest-scoring founders consistently outperform, or where scores don't predict outcomes?

This data lets you refine your rubric and catch systematic biases before they become entrenched.

Common Evaluation Mistakes to Avoid

Mistake 1: Evaluating the idea instead of the founder

Your program isn't investing in ideas. You're investing in founders. A mediocre idea in the hands of an exceptional founder will be pivoted into something better. An exceptional idea in the hands of a mediocre founder will be executed badly.

Evaluate the person—their understanding, their learning, their execution—not the brilliance of their concept.

Mistake 2: Treating polish as a signal of quality

A beautifully designed pitch deck, a confident video, a grammatically perfect application—these tell you the founder is good at presenting. They don't tell you the founder is good at building.

Some of the best early-stage founders present rough applications because they're spending their time building, not crafting applications. Don't penalize real work to reward presentation skill.

Mistake 3: Letting the interview carry too much weight

Interviews are good at evaluating charisma and communication skill. They're poor at evaluating most of the things that actually predict founder success.

Weight the application evidence heavily. Use the interview to explore specific gaps or concerns, not to do a fresh evaluation from scratch.

Mistake 4: Comparing founders to each other instead of to your rubric

Comparative evaluation ("this founder is better than that one") sounds logical but produces strange results when cohort composition matters. Score each founder against the rubric independently. Then use the scores to compare.

Mistake 5: Making exceptions without documenting why

Every selection process has borderline cases. The problem isn't that exceptions get made—it's that exceptions made by intuition embed bias without accountability.

When you make an exception—accepting a lower-scoring founder or declining a higher-scoring one—document the reasoning. Over time, review your exceptions. Are they producing good cohort additions? Or are you consistently making exceptions that favor particular demographics?

What to Do in the Interview

The interview isn't a fresh start. It's a focused follow-up to what you've already learned from the application.

Before each interview, the reviewer should have:

• The application score on each dimension

• The specific questions raised by the score (usually in the 2-3 areas with lower scores)

• Two or three evidence-based questions designed to probe those areas

Good interview questions for each dimension:

For problem-market understanding:

• "Walk me through a specific customer conversation that changed how you think about this problem."

• "Who is your most skeptical potential customer, and what would it take to convince them?"

For founder-problem fit:

• "What would have to happen for you to give up on this problem and move on to something else?"

• "Why you, specifically, and not someone with deeper domain expertise?"

For learning velocity:

• "What's the most significant thing you believed six months ago that you've since changed your mind on?"

• "Walk me through a decision you made that turned out to be wrong—and what you did next."

For execution evidence:

• "What have you built or done in the last 30 days, and what did you learn from it?"

• "What's the fastest you've ever moved on a decision? What made you move that fast?"

The Bottom Line

Selection is the highest-leverage decision you make as a program manager. The cohort you build determines the program you run.

A gut-feel selection process will produce inconsistent cohorts, systematic bias toward familiar founder profiles, and missed founders who could have been exceptional with the right support.

An explicit, rubric-based process won't eliminate bias—but it makes your biases visible, creates accountability, and gives you data to improve over time.

Know what you're looking for. Define it explicitly. Score against it consistently. And check your patterns to make sure you're actually selecting for the things that matter.

That's how you go from "we know it when we see it" to "we know what we're looking for, and we find it."

.

.

.

Want a head start on building your selection process? I've built a Selection Rubric Template with scoring criteria for all four dimensions, a calibration guide, and a set of bias-check prompts for your review team. Download it here.

You might also find the Interview Question Bank useful—it's a library of evidence-based questions organized by founder dimension and stage. Grab it here.

This post is part of a series on founder experience for accelerators, incubators, and startup studios. If you found this useful, you might also like: "The Sourcing Problem: How to Build a High-Quality Pipeline Without Getting Crushed" and "Assessing Founder Potential When There's No Traction (Yet)."