How HubSpot built a $30B company by zigging while others zagged

Executive overview

Most startup advice says focus on one thing and be world-class at it. HubSpot did the opposite — and built a $30B company. Dharmesh Shah, co-founder and CTO, shares how he and Brian Halligan made a series of high-conviction, low-consensus bets across product, culture, and personal design that defined HubSpot's trajectory.

The through-line: simplicity is worth fighting for, complexity kills companies slowly, and you can shape your company around your strengths rather than defaulting to convention.

Culture is a product — and your job is to iterate it, not preserve it.

Leaning into strengths: no direct reports, ever

• Dharmesh has had zero direct reports across 18 years and 7,000+ employees.
• He had been CEO twice before HubSpot and concluded he was bad at management; he didn't want to spend years becoming passively okay at something he didn't enjoy.
• He negotiated this explicitly with co-founder Brian Halligan at the founding meeting — including a promise it wouldn't change when someone left and the team needed interim coverage.
• Result: he gets all the upside of scale (long-term bets, global ambition) without the downside (managing people, org mechanics).
• Talent controls the slope of the learning curve; most things are acquirable skills — but that doesn't mean you should acquire all of them.

Engineering humor: laughs per minute

• Public speaking doesn't come naturally to Dharmesh; he treated it as an engineering problem.
• He decomposed the skill into subskills and focused on one per year (slide design, visual expression, humor).
• He records practice talks, transcribes them, and runs custom software that maps every audible laugh to measure LPM (laughs per minute).
• For business talks, 1.25+ LPM puts you in the top decile; stand-up comedians target much higher because entertainment is their only goal.
• Two tactical rules from stand-up: (1) the funny bit must be the last words in the segment — stop talking immediately after; (2) leverage a single setup story to squeeze multiple punchlines from the same context.

Going public: an underrated move for founders

• Founders over-index on the downsides of being public; Dharmesh argues the benefits are underappreciated.
• Public markets give you real-time, daily price discovery — vs. private companies getting marked to market only when raising a round.
• HubSpot's first billion in market cap creation was accessible to a tiny group; the next $29B was available to any public investor — customers, partners, employees.
• Bezos's first shareholder letter is the model: clearly state how you will run the company, and implicitly tell investors who don't agree not to buy your shares.
• HubSpot made every employee a designated insider on IPO day — there is no legal cap on insider count — preserving full financial transparency companywide.

High conviction, low consensus bets

• "You need to be right about something other people think you're wrong about for a long time" — Peter Thiel.
• HubSpot's core zig: focus on SMB when every VC said it was impossible. They held this for all 18 years despite constant IPO roadshow pressure to pursue enterprise.
• SMB advantages: no revenue concentration, you own your roadmap, short feedback loops, easy to experiment — with the scale of a consumer market.
• Reverse gravity: software markets always pull companies upmarket toward enterprise; staying in SMB means competing only with the few willing to do the hard thing.
• Secondary zig: build a broad all-in-one product from year one, when advice said focus on one thing. The insight: SMBs didn't lack individual tools, they lacked integration.
• Internal heuristic: if you're in the top three in any individual product category, you've overinvested in that category — your value prop is the whole, not any part.

Fighting the second law of thermodynamics

• The second law: within a closed system, entropy increases over time. In companies: everything trends toward complexity, slowness, and eventual death.
• Three stages of a company: fighting to survive → fighting stagnation → fighting complexity.
• Complexity is the slowest and most reliable killer of companies at scale.
• HubSpot's cultural principle: "Fight for simplicity" — three words that convey both that simplicity matters and that it requires active effort.
• Product rule from the early years: every new feature added requires removing one — a crude but effective constraint.
• Third-order cost of new features: not just development time or maintenance, but the dimensional complexity added to every subsequent decision (hiring, marketing, reporting, strategy).
• Systems and imposed constraints outlast cultural slogans; find mechanisms that enforce simplicity structurally.

Flashtags: calibrating the weight of your opinions

• Founders have megaphone problems — any offhand thought gets treated as a mandate.
• Flashtags are hashtags on an escalating scale of conviction:
  1. #fyi — interesting, no response expected
  2. #suggestion — I'd consider this; no response needed, no obligation to act
  3. #recommendation — I've researched this, I'd do it; response expected if you don't
  4. #plea — I believe this deeply; still not a mandate, but please consider it seriously
• Self-descriptive, searchable, and reusable across HubSpot daily by many people.
• HubSpot has no mandates; autonomy is preserved even at the plea level.

Decisions: debate, decide, unite

• Data informs decisions; people make them. Don't let a chart mandate an action.
• Designate one person (the DRI — directly responsible individual) to own each decision.
• Amazon's "disagree and commit" reframed as debate, decide, unite — the unite step is critical and frequently skipped.
• Once a decision is made, even those who argued the other side must unite behind it.
• Calories spent on a decision should be proportional to the reversibility and magnitude of its consequences — not every decision needs deep analysis.
• Default position should be no; force yourself to accumulate reasons to say yes before committing.

Evaluating ideas: a four-factor framework

• When assessing any new startup, product, or feature, evaluate four dimensions:
  1. Potential — what's the magnitude of the outcome if successful?
  2. Probability — what are the realistic odds? (Use expected value, not gut filter.)
  3. Passion — do you care enough to work on this for years? Passion can develop over time.
  4. Prowess — what unfair advantage do you have? Existing code, market access, domain knowledge?
• Common mistake: filtering on probability before assessing potential — this discards asymmetric opportunities.
• Every yes requires identifying what you will say no to in order to make room for it.

Culture as product

• Culture already exists the moment you have two people in a company; the job is to articulate and iterate it, not manufacture it.
• Dharmesh's framing: every company builds two products — one for customers, one for the team. Culture is the product you build for your team.
• Implications: run quarterly NPS surveys on your culture product, publish all results transparently, categorize feedback as bugs, commit to fixes or explain why something is "working as designed."
• The job is not to preserve culture but to iterate it — the needs of your people change as the company scales.
• Distinguish core values (federal law — hold constant) from cultural practices (state law — teams can adapt).
• Aspirational items in a culture document can become true over time by being stated: new hires adopt them as fact, and the culture self-fulfills.

AI: cognition at scale

• The internet gave distribution at scale; AI gives cognition at scale — the first technology to demonstrably amplify the human brain.
• Software has always claimed to be intuitive but forced an impedance mismatch: users translate intent into clicks, drags, and menus.
• AI enables a shift from imperative interfaces (step-by-step instructions) to declarative interfaces (describe the outcome, the software figures out the steps) — the same leap SQL made for databases.
• Best way to learn AI: don't study it abstractly. Find a real problem you care about and solve it with the tools available today.
• Build in public — the internet is uniquely good at telling you where you're wrong.