LLenny's Podcast
← All episodes
Dharmesh Shah (co-founder/CTO)04 April 2024

Zigging vs. zagging: How HubSpot built a $30B company

9Frameworks
15Insights

Frameworks in this episode

Leadership7 steps

Culture as a Product

Every company builds two products: one for customers, one for the team. Culture is the second one.

Read →
Leadership6 steps

Debate, Decide, Unite: Systematic Decision-Making

Name the decider, list the factors, stack-rank them, then unite behind the call — with calories proportional to consequence.

Read →
Self-Mastery5 steps

Designing Yourself Out of Your Weaknesses

Refuse to spend years becoming passably okay at something — negotiate the constraint at founding, in writing.

Read →
Innovation5 steps

Fight for Simplicity: The Third-Order Cost of Complexity

Entropy always wins unless you fight it — price every feature at its dimensional cost, not its build cost.

Read →
Leadership5 steps

Flash Tags: The Escalating Opinion Scale

Four hashtags that tell people exactly how much weight to put on your opinion.

Read →
Entrepreneurship5 steps

The Four P's: Potential, Probability, Passion, Prowess

Score any idea on four factors — and always compute potential before probability.

Read →
Communication7 steps

The LPM Method: Engineering a High-Stakes Keynote

Decompose public speaking into subskills, then optimise humour with a measured laughs-per-minute score.

Read →
Strategy5 steps

The N+1 Rule: Simplest Thing Until It Breaks

Start with the simplest possible policy, then prove by induction it still works at the next headcount.

Read →
Strategy5 steps

Zig While Others Zag: High-Conviction, Low-Consensus Bets

Pick a small number of contrarian bets, hold them for years, and build a heuristic that tells you when you have drifted.

Read →

Insights & moments

The myth-busts, hot takes, explainers, and tools worth keeping.

Myth Buster· 1

Myth Buster27:30

Most People Use 'First Principles' Wrong

Dharmesh pushes back on the buzzword: first principles doesn't mean the core principles you believe as a company — it means the first principles of the universe, the things everyone collectively knows to be true, i.e. science. Everything layered on top of that (like HubSpot's transparency) is a decision or an assumption, not a first principle. He calls transparency a 'founding principle,' not a first principle.

  • First principles = laws of the universe (e.g. thermodynamics), not your company beliefs
  • It's what everyone collectively knows to be true, not what you believe is true
  • Your beliefs, decisions, and assumptions sit on top of first principles
  • Transparency is a founding principle, not a first principle — it's core to HubSpot but not universal

most of the time first principles is used um it's actually used incorrectly

Dharmesh Shah · 27:30

it's the first principles of the universe it's not what you believe is true is what do we all collectively know to be true to…

Dharmesh Shah · 28:00
#first-principles#thinking#decision-making

Hot Take· 4

Hot Take14:00

Solo Ware: Building Software for Exactly One User (Yourself)

Dharmesh describes 'solo ware' — software built for exactly one person, him. Because it only has to work for a single user, there's almost no testing burden and no UI generality needed. The best part is the freedom to turn it off the moment it stops being useful, with no disappointed users to answer to.

  • He's been building single-user tools for himself for about 30 years
  • One user means minimal testing and a UI that only needs to satisfy one person
  • You can kill it instantly when it stops delivering utility — no users to disappoint
  • The decision to make something for more people ('microware') depends on whether it's worth the extra calories

solo Weare is exactly what it sounds like it's software built for exactly one person and in my case that one person is me

Dharmesh Shah · 14:00

the most important thing is that because it's only for one person if it stops bringing utility you can just turn it off

Dharmesh Shah · 14:30
#software#building#productivity#tools
Hot Take20:30

Founders Overrate How Bad Going Public Is

Dharmesh argues too many founders are terrified of the IPO, treating it as the beginning of the end. In his experience the taxes of being public are real but overstated, and there are underrated benefits: you always know your market value, and valuation oscillates around real value in an efficient market. His social argument: going public lets your customers, partners, and well-wishers share in the upside they can't touch in private markets.

  • The public-company 'tax' is real but founders over-index on how bad it will be
  • You always know exactly what the market values the company at, any business day
  • In an efficient market, valuation oscillates around value and catches up over time
  • HubSpot created ~$1B in market cap privately, then $29B more that public investors could participate in

I think too many Founders um you know once they get to they are very apprehensive of going public and going through the IPO process…

Dharmesh Shah · 20:30

that first billion of of market cap creation a very small number of people got to participate in the rest of the $29 billion in…

Dharmesh Shah · 23:00
#ipo#public-company#founders#markets
Hot Take48:30

'Reverse Gravity' Pulls Every Software Company Upmarket to Enterprise

Dharmesh argues software has a 'reverse gravity': bigger customers stay longer, pay more, and have higher NPS, so smart teams get pulled up-market until nearly every successful software company becomes an enterprise company. That's exactly why staying focused on SMB is so hard — and so defensible. In SMB you compete only with the few 'stupid enough' to stay there, and once you crack the physics it's a more sustainable, more fun model.

  • Bigger customers retain longer, pay more, and have higher NPS, so the market pulls you up
  • Almost every successful software company ends up being an enterprise company
  • Staying focused on SMB is rare, so competition there is thin
  • SMB gives the best of both worlds: enterprise economics with consumer-scale market and no revenue concentration

in software there's what I call reverse gravity that over time the market will always P you up

Dharmesh Shah · 49:00

every software company ends up being an Enterprise software company so you end up competing with literally everyone in the SMB world you're only competing…

Dharmesh Shah · 49:30
#smb#go-to-market#strategy#saas
Hot Take1:21:00

Culture Is a Product You Build for Your Team

Dharmesh's core insight: every company builds two products — one for customers, one for the team, and that second product is culture. If you accept that, consequences follow: you'd never build a product without talking to customers, so HubSpot runs a quarterly NPS on its culture, treats complaints as 'bugs,' and iterates. The biggest founder mistake is thinking the job is to preserve culture — you should never freeze the code of a culture any more than a product.

  • Every company builds two products: one for customers and one (culture) for the team
  • Treat culture like product: survey it (quarterly NPS), log 'bugs,' fix or mark 'works as designed'
  • Preserving culture is the wrong goal — needs of employees and company change as you scale
  • Never freeze the lines of code of a culture; iterate it like product

culture is a product period and that every company builds two products one is the product they build for their customers and the other is…

Dharmesh Shah · 1:21:00

this is the number one mistake I think um I'm biased uh I think Founders make is that they say oh we have an awesome…

Dharmesh Shah · 1:22:00
#culture#product-thinking#leadership#npr

Explainer· 3

Explainer37:30

Every Business Is Fighting the Second Law of Thermodynamics

Dharmesh frames entropy as the enemy: in a closed system, disorder increases over time unless you intervene — molecules never reassemble a broken egg, and companies and codebases drift toward complexity the same way. He maps three company stages: fighting not to die, fighting not to stagnate, and finally fighting complexity, which kills more reliably (if more slowly) than anything else. Hence 'fight for simplicity' as a cultural principle.

  • In a closed system, entropy (disorder) increases over time unless you intervene
  • Companies drift toward more layers, headcount, and complexity — a slower death
  • Three stages: don't die, don't stagnate, then fight complexity
  • Simplicity requires active fighting because even well-intentioned people add complexity

the second law of Dynamics I'm paraphrasing here is that over time unless you intervene everything goes to crap I'm paraphrasing right

Dharmesh Shah · 38:00

complexity does kill companies uh maybe not as quickly as other things but much more reliably than other things

Dharmesh Shah · 39:00
#simplicity#entropy#complexity#strategy
Explainer41:30

The Hidden 'Dimensional Complexity' Cost of Every New Product

Dharmesh breaks feature/product cost into three orders: first-order is build cost, second-order is ongoing maintenance, and the most important third-order is the complexity a new thing adds to everything else. Going from one product to two isn't incremental — now every decision, hire, campaign, and chart must be split across both products. So the long-term cost of complexity must be factored into every build decision.

  • First-order cost = implementation; second-order = maintenance; third-order = added complexity everywhere
  • One-to-two products isn't linear — it adds a whole new dimension to every decision
  • Every chart, hire, and campaign now has to be sliced by product one vs product two
  • You can justify building product N+1, but only if it's worth that dimensional complexity

it's like now every decision you make has to be made to the lens of now we have two products we just hired an engineer…

Dharmesh Shah · 42:30

every chart you look at in terms of the the growth uh revenue for whatever like it's all of it now every chart that you…

Dharmesh Shah · 43:00
#product#complexity#strategy#decision-making
Explainer1:32:30

AI Is 'Cognition at Scale' — and Kills the Software Translation Layer

Dharmesh maps the eras: PCs gave compute at scale, the internet gave distribution at scale, and AI now gives cognition at scale — amplifying the human brain in a way that finally works in real products. The big product unlock is moving from imperative interfaces (click, drag, swipe to translate what's in your head) to declarative ones where you just describe the outcome you want, like SQL's WHERE clause, and the software figures out how. He notes his 2017 'growth bot' had the right idea but the tech wasn't there yet.

  • Compute at scale (PCs), distribution at scale (internet), now cognition at scale (AI)
  • Today's interfaces force an 'impedance mismatch' — translating intent into clicks and swipes
  • AI enables declarative interfaces: describe the outcome, not the steps (like SQL)
  • His 2017 'growth bot' proved the concept but failed because the technology wasn't ready

what AI gives us now um is is cognition at scale the ability to kind of literally amplify the human brain right like that's never…

Dharmesh Shah · 1:33:00

to what Engineers would call a declarative model a declarative model is you describe the outcome you want not the steps to get there

Dharmesh Shah · 1:34:30
#ai#product#interfaces#declarative

Story· 5

Story10:00

The Custom Software That Measures His Keynotes in Laughs Per Minute

To get good at humor on stage, Dharmesh built custom software that transcribes his practice talks and marks every point where the audience audibly laughed, computing a laughs-per-minute (LPM) score borrowed from standup comedy. He benchmarks against TED talks, where the most popular ones correlate with high LPM. The tool gives him a visual map so he can find dead stretches and either cut them or inject something funny.

  • LPM (laughs per minute) is a metric standup comedians use; only audible laughter counts
  • Popular TED talks strongly correlate with high LPM
  • Two ways to raise LPM: add more laughs, or cut the words between laughs
  • For a business talk, even 1.0-1.25 LPM puts you in the top decile of non-professional speakers; 2+ is hard because you also have a message to deliver

I have custom software that I've written that will say okay here are the points at which the audience laughed like actually audibly laugh that's…

Dharmesh Shah · 10:00

there are two ways to improve it one is to add more laughs two is to decrease number of words between laughs

Dharmesh Shah · 10:30
#public-speaking#humor#measurement#solo-software
Story16:30

Why He's Had Zero Direct Reports in 18 Years at HubSpot

At HubSpot's founding meeting, Dharmesh volunteered that Brian should be CEO (he'd been CEO twice and knew he was bad at it) and then, on impulse, said he wanted zero direct reports — ever. His reasoning: he'd sink years just becoming passably okay at management, calories better spent on things he's actually good at and enjoys. He made Brian promise never to hand him even an interim report, and calls it one of his best decisions.

  • 7,000+ employees and exactly zero direct reports since day one; no one-on-ones either
  • He'd rather not spend years becoming 'passably okay' at management
  • You tend to be good at what you enjoy and enjoy what you're good at
  • He got all the upside of scale (big bets, long-term plans) without the management downside

I think I could become possibly okay at management with some training with some coaching or whatever I don't want to spend any years of…

Dharmesh Shah · 17:00

right now I can honestly say I'm having a better time at HubSpot now it's 7,000 people than I was having at 70 people

Dharmesh Shah · 18:00
#leadership#management#strengths#founders
Story25:30

Making Every Single Employee a 'Designated Insider' to Keep Transparency

HubSpot shares all information internally (except illegal-to-share items and salaries). When IPO bankers asked for their list of 'designated insiders' and suggested five or six, Brian pushed to seven and Dharmesh to eight — and they discovered by 'mathematical induction' there's no legal cap. So on the day they went public, they made every employee a designated insider, and still do at 7,000 people, so the financials can stay open to everyone.

  • All information is equally shared with everyone, except illegal-to-share items and individual salaries
  • Bankers proposed ~5-6 designated insiders; there's actually no legal limit
  • They designated every single employee an insider on IPO day
  • Insider status is mandatory (with trading windows), not opt-in — everyone's in it together

there's no actual legal limit to how many insiders a company can have

Dharmesh Shah · 26:00

the day we went public is we designated every single employee to be a designated Insider so that allowed us to maintain the transparency thing

Dharmesh Shah · 26:30
#transparency#culture#ipo#hubspot
Story34:30

The Seating Lottery That Scaled to Hundreds and Killed Office Politics

HubSpot's first office had four chairs, two by a window. Instead of founders grabbing the good seats, they ran a lottery where winners chose from remaining seats, re-shuffling every new hire. Investors called it cute and said it wouldn't scale — it worked at 25, 50, 100, 200 people (later quarterly, with light optimization). Dharmesh's point: one simple decision eliminated an incalculable amount of the desk/office politics that eats corporate America.

  • Winners pick from remaining seats because people value different seats differently
  • Re-run every new hire, later every quarter; it held for hundreds of people
  • Investors insisted it wouldn't scale — it did
  • One simplifying decision avoided huge amounts of office politics and had chemistry side-benefits

our investors are like oh that's cute and that's funny it it's not going to work at 15 20 people it's not going to work…

Dharmesh Shah · 35:30

so the moral of the story is that one decision of simplifying can you imagine the amount of politics we avoided like in Corporate America

Dharmesh Shah · 36:30
#simplicity#culture#scaling#algorithms
Story51:00

The Zig: Going Broad When Everyone Says Focus — and the 'Top Three' Heuristic

The classic advice is to be world-class at one thing; HubSpot's early zig was the opposite — from year one they built SEO, analytics, blogging, and CMS tools, each category already dominated by strong incumbents. The 'one thing' they wanted to be best at was solving the whole customer problem, since SMBs couldn't stitch the tools together themselves. Their discipline heuristic: if a category ranked top three in the market, they'd over-invested, because the value prop was integration, not category dominance.

  • Standard advice: focus and be world-class at one thing — HubSpot deliberately did the opposite
  • The 'one thing' was solving the customer's whole problem, not any single tool category
  • SMBs couldn't assemble the point tools themselves, so all-in-one was the real value
  • Heuristic: being top three in an individual category meant they'd invested too much there

some of the best startup advice I've heard and I've ever given is startups should focus on one thing and be really really exceptionally world-class…

Dharmesh Shah · 51:30

are we in the top three in the market in that category if the answer is yes that means we invested too much in that…

Dharmesh Shah · 53:00
#zigging#product-strategy#all-in-one#smb

Takeaway· 2

Takeaway07:30

Talent Only Sets the Slope — Almost Everything Is an Acquirable Skill

Dharmesh reframes the talent-vs-skill debate: talent is real but it only controls how fast you learn and how high your ceiling is, not whether you can learn something at all. He treats public speaking (which doesn't come naturally to him) as proof that even a weak starting point can be trained up. The practical move is to functionally decompose a skill into sub-skills and improve one at a time.

  • Talent controls the slope of the learning curve and the ceiling, not whether learning is possible
  • Most things are acquirable skills, not innate gifts
  • He couldn't delegate or talk his way out of on-stage founder keynotes, so he chose to learn it
  • Break a hard skill into underlying sub-skills and improve each incrementally

Talent basically controls the slope of the curve but most things are actually acquirable skills

Dharmesh Shah · 07:30
#skill-building#public-speaking#learning#mindset
Takeaway1:39:00

His Definition of Success: Make the People Who Believed in You Look Brilliant

Dharmesh closes with a definition of success he's held for years and that HubSpot has adopted: success is making the people who believed in you look brilliant. He likes that it's outward-facing — the employees, customers, and investors who took a chance on you early — and about repaying their belief.

  • Success = making the people who believed in you look brilliant
  • It's an outward-facing definition, not a self-centered one
  • The 'believers' are early employees, customers who backed a rough product, and investors

success is making the people who believed in you look brilliant that's it

Dharmesh Shah · 1:39:00
#success#philosophy#leadership