Lenny's PodcastZigging vs. zagging: How HubSpot built a $30B company | Dharmesh Shah (co-founder/CTO)
CHAPTERS
- 0:00 – 4:16
HubSpot’s earliest “zig”: building many products instead of one wedge
Dharmesh opens with one of HubSpot’s most contrarian early choices: rejecting the classic advice to focus on one thing. He frames this as a deliberate strategy to solve the customer’s end-to-end problem rather than optimize for a single feature area.
- •Classic startup advice: do one thing world-class
- •HubSpot’s deliberate choice to do the opposite early on
- •All-in-one orientation as a strategic zig
- •Customer problem vs. tool-level problem framing
- •Setting the episode’s theme of contrarian, first-principles thinking
- 4:16 – 6:52
Rapid-fire fun facts: no reports, side projects, chat.com, and origin story
Lenny runs through a set of “fun facts” that reveal Dharmesh’s unusual operating style and personal backstory. This segment sets up later deep dives into leadership design, side projects, and how he thinks about leverage and experimentation.
- •Zero direct reports at HubSpot (ever)
- •No 1:1s by design
- •Side projects like Wordplay and building for fun/utility
- •Buying chat.com for 8 figures and flipping it quickly
- •Background: born in a village in India; promised not to start another company (then started HubSpot)
- 6:52 – 11:45
Engineering public speaking: decomposing skills and measuring “laughs per minute”
Dharmesh explains how he approached public speaking like an engineering problem: break it into subskills, practice systematically, and measure outcomes. He details his custom tooling to detect audience laughter and optimize talks using LPM benchmarks.
- •Talent vs. skill: most abilities are learnable
- •Functional decomposition of “high-stakes speaking”
- •Humor as the highest-leverage subskill
- •Custom software to detect laughs and compute LPM
- •Optimizing by adding laughs or reducing time between laughs
- 11:45 – 14:26
Tactical humor for presentations: punchline placement and stacking laughs
Dharmesh shares practical rules borrowed from stand-up: deliver punchlines as the final words, then pause, and reuse setup context to add multiple laugh beats. He also discusses realistic LPM targets for business talks and why comedy metrics correlate with engagement.
- •Punchline must be last words; stop talking to allow laughter
- •Use the same setup to deliver multiple punchlines
- •Business talks optimize for message + entertainment, not just laughs
- •Typical strong business LPM vs. stand-up benchmarks
- •Using measurement to improve talk structure over iterations
- 14:26 – 15:22
“Soloware”: building personal software tools without productizing them
Asked whether he’ll release his laughter-detection tool, Dharmesh introduces “Soloware”—software made for exactly one user (himself). He explains why avoiding users preserves freedom, reduces overhead, and allows tools to be turned off the moment they stop being useful.
- •Definition and appeal of Soloware
- •Minimal UI/testing when you’re the only user
- •Freedom to abandon tools without disappointing users
- •Cost/benefit of turning Soloware into a product
- •A broader philosophy of building for personal utility
- 15:22 – 18:46
Designing a founder role: why Dharmesh has zero direct reports
Dharmesh recounts an early founder conversation where he insisted on never managing directly. He argues it’s better to lean into strengths than spend years becoming “passably okay” at management, and describes how this kept him energized even as HubSpot scaled.
- •Founder agreement: CEO choice + explicit ‘no direct reports’ rule
- •Management as a craft he chose not to optimize for
- •Strengths-based career design inside a growing company
- •Getting scale’s upside without managerial downsides
- •How role design can keep founders engaged long-term
- 18:46 – 20:34
Shaping the universe to your liking: constraints, schedules, and founder-made rules
Dharmesh and Lenny discuss the idea that founders can rewrite default assumptions about how companies operate. Dharmesh shares examples like “no meetings before 11am” (later: meet, but don’t invite founders) as a reminder to proactively design operating norms.
- •Founders often assume norms are fixed; they’re not
- •Example: no meetings before 11am
- •Later evolution: meetings allowed, but not with founders
- •Using constraints to protect energy and deep work
- •Permissioning yourself to try ‘weird’ operating rules
- 20:34 – 24:40
Public company realities: why IPOs aren’t the “beginning of the end”
Dharmesh challenges the fear many founders have about going public, acknowledging added overhead but emphasizing benefits. He highlights clearer market pricing, the separation of value creation from valuation volatility, and the societal upside of letting more people participate in growth.
- •Founders over-index on IPO downsides
- •Public markets provide continuous ‘mark-to-market’ clarity
- •Efficient markets: valuation oscillates around value over time
- •Public participation: more stakeholders share in upside
- •Staying long-term oriented with clear communication (Bezos model)
- 24:40 – 27:18
Radical transparency at scale: making every employee a designated insider
To preserve HubSpot’s transparency culture post-IPO, Dharmesh describes a creative solution: designate every employee as an insider. This kept internal financial openness intact, even though it imposes trading restrictions and responsibilities on everyone.
- •Transparency as an early core value and operating mechanism
- •IPO prep revealed ‘insider list’ conventions aren’t strict limits
- •Mathematical induction-style pushing: 5…7…8…no limit
- •Decision: everyone becomes a designated insider
- •Tradeoffs: restrictions and windows, but consistent cultural integrity
- 27:18 – 37:20
High-conviction, low-consensus bets: SMB focus and other contrarian choices
Dharmesh reframes “contrarian” as selecting a small number of high-conviction, low-consensus bets grounded in reality (not vibes). He details HubSpot’s long-running commitment to SMB—resisting pressure to move enterprise—and shares how some contrarian experiments (like no titles) were later reversed.
- •Clarifying ‘first principles of the universe’ vs. ‘founding principles’
- •Contrarianism needs limits: >0 but not too many dimensions
- •SMB as a core bet maintained through board/investor pressure
- •Reverse gravity in software pulls companies upmarket; resisting it is strategic
- •No titles experiment, then reverting to standard titles for signaling/labor market reasons
- 37:20 – 40:28
Fighting entropy with simplicity: transparency, seating lotteries, and constraint systems
Dharmesh introduces his “second law of thermodynamics” metaphor: organizations naturally drift toward complexity unless they spend energy to prevent it. He shows how HubSpot used simple binary rules and algorithms—like universal access and randomized seating—to reduce politics, decision overhead, and creeping complexity.
- •Entropy in companies: disorder/complexity increases over time
- •Company phases: survive → avoid stagnation → fight complexity
- •“Fight for simplicity” as a cultural principle
- •Transparency started as a simplicity choice (“give access to everything”)
- •Seating lottery algorithm to avoid office politics; evolving constraints as scale grew
- 40:28 – 50:29
Product simplicity and the hidden cost of complexity: knobs, dials, and dimensionality
On product strategy, Dharmesh explains a constraint: adding a feature (a knob/dial) requires removing one elsewhere. He expands on third-order costs—how each new product or feature adds dimensional complexity that multiplies coordination costs across hiring, marketing, metrics, and prioritization.
- •Rule of thumb: add one feature, remove one
- •First-order costs: build effort; second-order: maintenance
- •Third-order costs: ‘dimensional complexity’ across the whole org
- •Adding product #2 changes every decision and dashboard permanently
- •Operationalizing simplicity via guardrails/constraints (freemium + SMB as forcing functions)
- 50:29 – 57:11
Zigging when others zag: the ‘wide then integrated’ all-in-one bet
Dharmesh revisits the early HubSpot zig: going broad across multiple marketing tools to solve the SMB integration burden. He shares a surprising heuristic for resource allocation (don’t overinvest in any single module) and explains when founders should choose breadth versus depth based on the true customer problem.
- •Going broad to solve the customer’s end-to-end workflow
- •SMBs struggle to stitch together best-of-breed tools
- •Heuristic: if you’re top-3 in a category, you may be overinvesting (early-stage)
- •“Wide and shallow” as a deliberate tradeoff; later evolution to top-3 in many categories
- •Guideline: fall in love with the problem, not the solution; impose constraints to survive breadth
- 57:11 – 1:02:44
Flashtags: a lightweight protocol to communicate opinion strength without mandates
Dharmesh introduces “flashtags” to prevent founder opinions from being treated as gospel. He defines a small, escalating set (#FYI, #suggestion, #recommendation, #plea) that clarifies intent, response expectations, and how hard he’s willing to push—supporting autonomy while reducing misinterpretation.
- •The ‘megaphone problem’: leader opinions get overweighted
- •A discrete, shared taxonomy for intent and urgency
- •Definitions: #FYI → #suggestion → #recommendation → #plea
- •No formal mandates; autonomy supported with DRIs
- •Searchable, trackable communication for learning over time
- 1:02:44 – 1:09:44
Decision-making at HubSpot: DRIs, data-informed choices, and ‘debate, decide, unite’
Dharmesh explains HubSpot’s evolving approach to making decisions: data informs but doesn’t decide, and a single owner (DRI) must be accountable. He emphasizes the cultural importance of alignment after a decision and shares systematic tools for clarifying factors, ranking importance, and matching decision effort to stakes.
- •Data-informed, not data-driven: people decide
- •Assign a DRI: deciding who decides is the first decision
- •Alignment mechanism: “debate, decide, unite” (vs. disagree and commit)
- •Systematizing decisions: define factors like spreadsheet columns; then stack-rank them
- •Calorie budgeting: effort should be proportional to consequences (one-way vs two-way doors)
- 1:09:44 – 1:15:25
Choosing what to invest in: expected value + passion/proximity + prowess
Dharmesh shares a simple scoring framework for evaluating ideas—especially startup or product bets. He stresses starting with potential outcome before probability (to avoid prematurely filtering), then considering whether you care enough to persist, and whether you have an unfair advantage.
- •Score ideas 0–10 to enable comparisons
- •Start with potential magnitude, then probability of success
- •Use expected value thinking (potential × probability) as a sanity check
- •Passion/proximity: you can become passionate after starting
- •Prowess: unfair advantages (distribution, code reuse, market access)
- 1:15:25 – 1:30:56
Culture as a product: the Culture Code, NPS loops, and iterating instead of ‘preserving’
Dharmesh tells the origin story of HubSpot’s Culture Code deck, including initial internal backlash and how he reframed culture as something to articulate and improve. He explains the “culture is a product” model: employees are customers, feedback is continuous (culture NPS), issues are treated like bugs, and leaders iterate rather than freeze culture in time.
- •Founder tasked him with ‘doing culture’; he resisted, then reframed it
- •Culture articulation via a ‘Python function’ thought experiment: attributes predicting success
- •Initial backlash: culture seen as corporate posters/inauthentic signaling
- •Operational loop: quarterly NPS, publish all responses, triage and fix ‘culture bugs’
- •Key principle: don’t preserve culture—iterate it like software; aspirational statements can become self-fulfilling when labeled clearly
- 1:30:56 – 1:37:03
AI as cognition at scale: declarative interfaces and the new product frontier
Dharmesh compares the AI moment to the early web: a platform shift with broad, cross-industry consequences. He describes AI as “cognition at scale,” predicts a move from imperative UI workflows (clicks) to declarative outcomes (describe what you want), and uses GrowthBot/ChatSpot as proof that the tech finally works.
- •AI’s impact feels web-scale; bigger than mobile/social to him
- •From compute (PC) + distribution (internet) to cognition at scale (AI)
- •Imperative → declarative shift: specify outcomes, not steps
- •Analogy to SQL: describe desired result; system figures out execution
- •GrowthBot failed earlier because tech wasn’t ready; ChatSpot works now
- 1:37:03 – 1:39:07
Practical advice to learn AI: solve a real problem, build, and learn in public
Dharmesh’s learning advice mirrors his broader philosophy: don’t learn in the abstract—pick a problem you care about and use AI tools or APIs to solve it. He encourages iterative building and public writing/sharing to accelerate feedback and understanding.
- •Avoid ‘learning because it’s important’; learn by solving a real problem
- •Use existing tools first; build on APIs when needed
- •Iteration beats theorizing; measure progress against concrete goals
- •Learning in public invites fast feedback (including critique)
- •Writing as a forcing function to clarify thinking
- 1:39:07 – 1:41:43
Closing reflections: definition of success and where to find Dharmesh
Dharmesh ends with a personal definition of success centered on others: making believers look brilliant. He then shares where people can follow him, how to engage, and wraps the conversation with gratitude and a final prompt for listeners to share favorite episodes.
- •Success = making the people who believed in you look brilliant
- •Applies to employees, customers, and early investors
- •Online presence: dharmesh.com and social platforms (especially LinkedIn)
- •Invitation to offer feedback and corrections publicly
- •Episode wrap-up and sign-off
