Uncapped with Jack AltmanPat Grady & Alfred Lin on the Tactics of Great Venture Investing | Ep. 36
CHAPTERS
- 0:00 – 1:15
Conviction beats consensus: how Sequoia votes on investments
Pat explains Sequoia’s internal voting data: whether a deal is consensus or controversial doesn’t predict success, but the presence of strong conviction does. He outlines their 0–10 voting system (no fives) and why a split room with strong yeses can be better than lukewarm agreement.
- •Consensus vs. non-consensus is not predictive in Sequoia’s data
- •Conviction is the key signal; lukewarm ‘6’ votes are a red flag
- •0–10 voting system with clear positive/negative thresholds
- •Strong yes/strong no dynamics can indicate real outlier potential
- 1:15 – 4:30
Stewardship mindset in the generational transition: responsibility without bureaucracy
Alfred and Pat discuss how the new co-leadership feels both heavy (responsibility to the legacy) and light (focus on investing, not admin). They emphasize enabling a talented partner group and keeping the firm oriented toward being “in the field.”
- •Excitement about leading Sequoia into the next generation
- •Stewardship as an honor and responsibility, not an explicit career objective
- •Goal: minimal administrative work; maximize time investing alongside partners
- •Leadership as enabling and getting out of the team’s way
- 4:30 – 6:27
Why venture leadership isn’t ‘CEO’ leadership: operating consistency vs. outlier hunting
Pat contrasts operating companies—where consistency and top-down decisions matter—with venture capital, where the goal is finding rare outliers. To partner with outlier founders, Sequoia aims to be a team of outlier investors who aren’t managed into uniformity.
- •Operating CEOs optimize consistency and quality at scale
- •Venture is an ‘outlier business’: find a few exceptional winners each year
- •Outliers can’t be managed like a standard org; they need autonomy
- •Team must be spiky, competitive, and still collaborative (‘heart of gold’)
- 6:27 – 9:38
Managing inputs with ‘freedom within frameworks’: capabilities and values as the system
They describe how Sequoia creates structure without homogenizing partner styles. The firm uses a value-chain framework (sourcing, picking, winning, building, harvesting) and explicit values, while allowing different authentic operating rhythms across partners.
- •Different partners require different ‘inputs’ to do their best work
- •Framework: sourcing, picking, winning, building, harvesting
- •Values alignment + capability development enables autonomy
- •Examples of varied styles: thematic whiteboarding vs. high-volume market coverage
- 9:38 – 11:52
How Sequoia reviews performance: values, capabilities, and the long-lag reality of outputs
Because venture outcomes take years and interim markups can be mirages, they measure behaviors and skills more than short-term results. Pat outlines a concrete review cadence: values reviews mid-year and capability reviews at year-end, then diagnosing broken links when outputs don’t follow.
- •Outputs lag by ~10 years; markups can mislead
- •Emphasis on input quality: sourcing quality, memo quality, diligence rigor
- •Growth team values: aggressive but humble; strong under scrutiny; high give-a-shit; zero bullshit; demanding & supportive
- •Annual cadence: June values reviews, December capability reviews
- 11:52 – 13:32
Sourcing and coverage: aiming for ‘enough’ without CYA, and avoiding metric gaming
They unpack what “good seeing” means: not maximizing meetings, but finding what’s worth real work. Pat explains Sequoia’s use of coverage targets (e.g., ~70% for growth) and why they avoid individualized metrics that incentivize padding behavior, drawing on Pat’s Summit Partners experience.
- •Seeing everything is not the goal; it can produce false ‘high accuracy’ via excessive passing
- •Coverage targets: growth ~70% of relevant competitor-invested deals; seed lower due to wider waterfront
- •Avoiding individual metrics prevents gaming and misallocated time
- •Early indicator of judgment: how a new investor invests their time
- 13:32 – 17:56
Real-time learning loops: tracking true/false positives, writing down ‘why we passed’
Pat and Alfred describe decision hygiene to make sourcing learnable: tracking decisions in quadrants, updating priors as new info arrives, and documenting rationale in the moment. The aim is to preserve context so the team can evaluate decisions later without hindsight distortion.
- •Maintain running lists of true/false positives/negatives and revisit quarterly
- •Hard part: later you forget the original rationale—so codify it immediately
- •Short ‘pass memos’ to capture last 5% of learning
- •‘Rolling’ process: continuously update priors with new information
- 17:56 – 22:39
Seeing the right early-stage companies: false coverage, network shots, and funnel quality
For seed and early, breadth is unavoidable—but the key is choosing which companies deserve substantive engagement. Alfred introduces “false coverage” (e.g., demo day attendance without follow-through) and describes Sequoia’s ‘three shots’ model: pre-formation, seed, then Series A, all driven by being in the right talent networks.
- •‘False coverage’ vs. substantive engagement with the best few prospects
- •Measure quality from a list: which 5 companies did you pursue and why?
- •Three shots: before formation → seed (if you earned access) → Series A
- •Network placement matters (e.g., being close to top AI labs and talent flows)
- 22:39 – 24:37
Proprietary talent mapping in the CRM: ‘PageRank for people’ and why it can’t be transactional
Pat details Sequoia’s long-running practice of building a talent graph by asking trusted operators who the best people are—then tracking it over a decade. This becomes a proprietary signal (especially around engineering strength) but only works if the firm consistently helps people first and avoids transactional asks.
- •Talent map built from repeated ‘top 5 smartest’ referrals over many years
- •Integrated into CRM with paid, public, and proprietary data sources
- •Used as a signal: e.g., engineering team percentile correlates with outcomes
- •Requires trust and giving-first behavior; transactional requests degrade data quality
- 24:37 – 29:07
The impact of great engineers: when one 1,000× builder is enough—and when it isn’t
They explore how engineering quality relates to company success, emphasizing that changing company DNA is hard. Examples like ServiceNow and Palo Alto Networks show the ‘one extraordinary engineer’ model, while HubSpot illustrates upgrading product/engineering via acquisition; deeply technical companies (e.g., foundation models) require elite teams from day one.
- •Company DNA is hard to change; you can augment more easily than transform
- •ServiceNow: largely one-person codebase (Fred Luddy) even through IPO
- •HubSpot: acquired Performable to rebuild platform and unlock product excellence
- •More technical products demand more technical founders/engineering depth
- 29:07 – 32:22
Picking winners: venture math, ownership, and why being ‘half right’ can be great
Alfred frames picking as assembling enough high-multiple outcomes to make fund economics work, accepting high write-off rates. He underscores the necessity of meaningful ownership and conviction—small checks into big funds won’t move returns even when right.
- •Fund math: ~45–55 shots; need a handful of 10×+ outcomes and a few massive gains
- •Great funds can still have ~50% write-off rates
- •Risk-taking is the job; avoiding losses isn’t the objective
- •Ownership matters: low-ownership wins often don’t matter to fund returns
- 32:22 – 36:26
Hot rounds vs. non-consensus: Sequoia’s data says conviction matters more than heat
They discuss how deal ‘hotness’ varies by round and cycle (DoorDash A/B hot, C less wanted, later rounds consensus). Pat shares examples (Okta, HubSpot, Zoom, Snowflake) and returns to the core lesson: consensus doesn’t predict results; strong internal conviction does.
- •DoorDash: A/B competitive; C less wanted but highly attractive in hindsight
- •Examples of non-consensus wins: Okta, HubSpot, Zoom; Snowflake second-chance bet
- •Sequoia tracks votes for a decade: consensus vs. non-consensus not predictive
- •Volatility is necessary: truth is rarely ‘in the middle’
- 36:26 – 44:35
Coaching for asymmetry: building courage, confronting biases, and ‘front stabbing’ feedback
Pat explains how Sequoia coaches investors who produce ‘base hits’ without outlier upside: force earlier decisions, do deals together, and normalize being wrong. They highlight psychological barriers—fear of missing out and fear of looking stupid—and how post-mortems reveal errors are usually emotional biases, not calculation mistakes.
- •Direct feedback culture (‘front stabbing’) to address risk aversion
- •Practical coaching: push the last 30% → 20% → 10% of hesitation down over time
- •Investor profile trap: A-student perfectionism clashes with VC’s error rate
- •Cataloging ~40 decision biases; example: ‘separation of church and state’ (thrill vs. clinical judgment)
- 44:35 – 53:18
Founder-market fit as a picking lens, and avoiding ‘force-fitting’ your vision onto founders
Alfred describes his core heuristic: the founder must be uniquely suited to the market’s problems and the company’s multi-act arc. He warns that experienced operators may over-impose their own solution vision; instead, investors should stay curious, go deep with questions, and let the founder’s authentic ambition and adaptability emerge.
- •Founder-market fit: ‘Was this person made for this company and market?’
- •Legendary companies require multiple acts beyond the initial wedge
- •Avoid operator trap: don’t force-fit your vision onto the founder
- •Use deep questioning (why/what/how) to co-clarify vision with the founder
- 53:18 – 1:02:59
Winning and onboarding: authenticity, great ‘passes,’ and building trust through competence + intention
They argue winning isn’t performative—founders can feel genuine conviction if you’ve done the work throughout the process. After investing, onboarding is about earning trust: deeply learning the business (even doing employee onboarding) and demonstrating pure intent, so that by year one the founder sees you as a reliable partner.
- •Winning comes from consistent engagement and real homework, not ‘salesy’ tactics
- •Passing well preserves relationships; ‘not now’ framing and useful diligence feedback
- •Onboarding: listen first; pattern recognition as a board member, not an operator
- •Trust components: competence (context, decks, VP 1:1s) + intention (help founder become best self)
- 1:02:59 – 1:09:38
Proudest board moments and 2026 priorities: stable partnership enables volatile individual bets
They reflect on what makes board work meaningful—being helpful without needing to be ‘right’—and the pride that comes from founder-level trust. Looking to 2026, they emphasize continuity: keep executing the core craft, compound improvements, and maintain partnership stability so individual partners can take bold, outlier-shaped swings.
- •Best board dynamics come from humility, curiosity, and updating priors
- •Pride comes from founder trust and being treated as part of the journey
- •2026: ‘business as usual’—investing and working with founders remains primary
- •Stability at the partnership level enables volatility and risk-taking at the partner level
