Zig When Others Zag: HubSpot Co-founder Dharmesh Shah on Simplicity, Culture as a Product and Building a Company Around Your Strengths

Shah says public speaking does not come naturally, but he could not delegate stage time as HubSpot's annual event grew from 150 people at a Marriott to around 10,000 in the live audience. He believes talent controls the slope of the learning curve, but most abilities are acquirable skills. As an engineer, he functionally decomposed public speaking into sub-skills and worked on one each year.

The one with the most impact was humor, which he approaches in a data-driven way. He practices before increasingly large live audiences, transcribes the talks and runs custom software that marks where people audibly laughed, computing laughs per minute (LPM), a metric stand-up comedians use. Business talks have a low LPM, he says, and for non-professionals a rate above about 1.2 or 1.25 is top decile. The software maps each talk by minute so he can see, for example, a stretch of one minute 17 seconds with no laughs.

He shares two tactical tips. Make the funny bit the last words of a segment and then stop talking, because the audience needs about half a second and permission to laugh. And once you have invested words in setting up a story, add several punchlines, because a laugh is a laugh.

Asked about releasing the software, Shah explains his idea of SoloWare: software built for one user, himself. It needs little testing, and if it stops being useful he can turn it off, unlike a product with users who would be disappointed.

Shah tells how it began at the founders' meeting where he and co-founder Brian settled who would be CEO. Having been CEO twice before, he concluded he was not good at it, partly because the role requires management as a craft, so Brian took it. Shah then added that he wanted no direct reports, an idea that popped into his head, and that Brian would have to promise not to make him interim head of anything.

His reasoning is that he could become passably okay at management with coaching but does not want to spend years on that, preferring to build on what he is good at and enjoys. He says the decision was one of his best for himself and for HubSpot, and that he is having a better time at 7,000 people than at 70, with the upside of scale, such as making big long-term bets, without the downside of managing many people.

Rachitsky says the story shows you can design a company the way you want. Shah agrees that people, especially first-time founders, assume things must go a certain way, and gives another example: he and Brian are night people, so for years there were no meetings before 11 a.m., later relaxed to allow them as long as neither co-founder was invited.

Shah says founders over-index on the idea that going public is the beginning of the end. There is a tax in obligations, he says, but also benefits, including knowing what the market values HubSpot at on any business day instead of being marked to market only when raising a round. He argues the public market is relatively efficient, so valuation oscillates around value, and if you focus on creating value the two move together over time.

There is also a social reason, he says. HubSpot went public at about eight years old with roughly a billion dollars of market cap, and the later growth to about 30 billion is something customers, partners and well-wishers could take part in, which is not possible with private companies.

On staying motivated, he cites Bezos's first shareholder letter as the model: being public only requires transparency about what you do, and those who disagree need not buy the shares. HubSpot's own mechanism is that all information is shared equally with everyone in the company, with exceptions only for things legally unshareable, and salaries, which he says are co-owned by the company and the individual.

In IPO prep the bankers asked for a list of designated insiders, typically five or six. Shah and Brian pushed to seven, then eight, and found there is no legal limit, so on the day they went public they designated every employee an insider, mandatory rather than opt-in. They still do this at 7,000 people.

Shah says people misuse the term first principles. In his view they are what we collectively know to be true, like the laws of thermodynamics, while a company's own beliefs, such as transparency, are founding principles. Being contrarian must be limited: the number of contrarian bets must be above zero but not too high, and he describes them as high conviction, low consensus, citing Peter Thiel's idea of being right about something others think is wrong for a long time.

HubSpot's first such bet was targeting SMBs, which it kept for all 18 years, even when roadshow investors asked for the path to the enterprise. He says it was not a bowling-pin strategy of conquering SMBs first. Conviction on that bet let the company make secondary decisions more easily.

Not everything has to be reinvented, and the company reverted on some choices. In the early years nobody had a title, but as HubSpot scaled employees explained that titles signal progression outside the company, and that going without them meant HubSpot should pay more in other ways. After a daylong meeting with three options, the company chose classic titles because made-up titles lack meaning, much like having none.

Shah confirms that early on everyone at HubSpot, founders included, was paid $5,000 a month for the first year or year and a half. He stresses solving for simplicity in the early years. Transparency, he says, did not start as a moral position: when hiring the first employee, Brian said to give them access to everything, and Shah noted that binary decisions are easier than non-binary ones.

The logic of 'if it works for employee three, why not four, five or fifty' kept the policy going until it broke. He says the same approach shaped HubSpot's very first office in a co-working space in Cambridge, where there were four chairs, two by the window and two not.

To avoid favoritism, they ran a lottery in which whoever draws first picks from the remaining seats, and they reshuffled every time a new employee joined. Investors said it would not work beyond 15 to 20 people, but it worked at 25, 50, 100 and 200, with tweaks such as keeping engineers away from sales and shuffling quarterly instead of per hire.

Shah says that one decision avoided politics over office size and desk height, a savings he calls incalculable, along with side benefits like people getting to know each other.

Shah paraphrases the second law of thermodynamics as: unless you intervene, everything goes to crap. He sees a company's life in three stages: first not dying, then not stagnating, and third fighting complexity, which he says kills companies more slowly but more reliably. More layers, headcount and margin pressure follow, and it is never too early to plant the seeds of simplicity.

HubSpot's culture includes the principle 'fight for simplicity,' which means simplicity is worth fighting for and that it takes a fight, because even well-intentioned people add complexity such as more pricing tiers and product knobs. In the early product years, adding a feature required removing one elsewhere, a coarse rule that at least forces thought.

He describes three orders of thinking about a feature's cost: implementation, maintenance, and the third and most important, the added dimensional complexity. Going from one product to two means every decision, hire, campaign and chart must be sliced by two products. Shah says that cost should factor into everything you do.

On operationalizing it, he says it is hard. HubSpot's early constraints helped: serving SMBs with a freemium product limits how much complexity you can add. His lesson is to build guardrails and mechanisms, as Amazon would say, because even a reasonably good system outperforms values on a page, which deteriorate over time.

Rachitsky, an angel investor, notes that small-business startups are hard because customers are slow to adopt and expand. Shah's advice is nonetheless to do it. He and Brian had both worked in enterprise software, where startup life suffers from long sales cycles, slow feedback and revenue concentration that lets the biggest customer drive the roadmap.

Consumer startups, he says, tend to have bimodal outcomes: huge or zero. SMB offered the best of both: a product people pay for, an incrementally more valuable measurable business, millions of potential customers, no concentration, short feedback loops and room to experiment. When HubSpot began it was about a hundred times harder, with investors citing only one global SMB software brand.

He describes a pattern he calls reverse gravity: bigger customers stay longer and pay more, so smart teams get pulled upmarket, and almost every successful software company ends up enterprise, competing with everyone. In SMB you only compete with those stubborn enough to stay, which makes it harder but more sustainable, and he urges startups to start there and keep trying even when it is painful.

Shah says he and Brian both ask whether they have considered the alternative path before following the normal one. His advice is not always to zig but to know what the zig would have been. He calls the advice to focus on one thing and be the best at it among the best he has heard and given, and says HubSpot chose the opposite in year one by building SEO, web analytics, blogging and content management tools.

The reasoning was that SMBs did not lack tools but the wherewithal to assemble them, since for most it is a science project. To solve the actual customer problem, they had to go broad even though it felt wrong. They accepted being miles wide and inches deep and used a heuristic: if HubSpot was in the top three in a category, it had over-invested, because the value proposition was that everything worked well together.

Shah adds that today HubSpot is in the top three in most of its categories, but at the time, with scarce resources, it was a way of staying disciplined. Asked when going wide makes sense, he says founders fall in love with the solution instead of the problem. If talking to customers shows you need more than one thing, impose a constraint on how broad you go, since you cannot be the best in all dimensions.

Shah describes the megaphone problem: as a founder with an opinion on everything, a throwaway comment can be treated as gospel. His fix is flash tags, a ladder of hashtags: FYI (no response expected), suggestion (no response needed), recommendation (a response is expected if you decide not to act) and plea, the strongest. None is a mandate, and he says he can count on one hand the times plea reaches that point.

The tags are searchable, on the company wiki, and used daily at HubSpot, and the explanation is at flashtag.org. He ties the absence of mandates to HubSpot's culture of autonomy and its directly responsible individuals, or DRIs.

On decisions, Shah says HubSpot used to decide poorly and has improved. It is data-informed but data does not decide, people do; one designated decision-maker owns each decision, who is not necessarily the executive over the team; and the company practices 'debate, decide, unite,' a softer take on Amazon's disagree-and-commit.

Personally, he lists the factors of a decision, as if writing a Python function or spreadsheet, and ranks them, which he calls an 80% win. He also scales effort to consequence, extending Amazon's one-way and two-way doors, and defaults to no on new things, since every yes means a no somewhere else.

Shah scores startup, product or feature ideas on a zero-to-ten scale across four columns. First is potential, how big the outcome could be if successful, and second probability of success. He warns the common mistake is to look at probability first and throw out ideas with a one-in-ten chance that might be worth it, given a big enough payoff. He explains expected value with a fifty-fifty shot at 10 million.

The third is passion or proximity: do you care about the problem and can you work on it for years? He is wary of the word passion, since many successful companies did not begin with their founders' passion, and it is fine to become passionate once you dig in. Rachitsky offers Zip, whose founders became passionate about procurement after spotting the opportunity.

The fourth is prowess: do you have assets or an unfair advantage, such as reusable code or market access? No single factor decides, he says, but the discipline of asking how big it could be, what the chances are, how much you care, and why you and your company should pursue it makes the choice clearer.

Shah says the word culture had never been mentioned at HubSpot until three or four years in, when Brian suggested he take it on. He saw himself as the bottom decile for the job, but took it as describing the existing culture rather than creating one. He framed the question as what factors he would use in a Python function to predict whether a new hire would succeed at HubSpot.

His first email announcing a culture survey produced one of the worst reactions he has had, including the line that next there would be posters about excellence. People had seen culture done badly elsewhere. He went back and wrote a 16-slide deck, the Culture Code. The survey asked an NPS-style question about recommending HubSpot as a workplace, and the top reason was the other people there, which led him to define the attributes of those people.

His idea is that every company builds two products, one for customers and one for its team, which is culture. Following from that, he repeats the NPS survey quarterly and publishes every anonymous response; the results are categorized and discussed at the all-minds meeting, with bugs identified, fixes committed and some items declined as 'works as designed.'

He says founders' job is not to preserve culture, since employee needs change, and a culture should be iterated like code. Some things are core values, like transparency, while others can vary by team in a way he compares to federal versus state law. He also says cultures become self-fulfilling prophecies, so aspirational items, flagged as such, tend to become true as people join.

Shah says that after nearly 30 years in commercial software, AI is the most excited he has been since the web in the mid-to-late 1990s, beyond mobile and social. In his framing, PCs gave compute at scale, the internet gave distribution at scale, and AI gives cognition at scale, the ability to amplify the human brain.

He argues that software called intuitive still forces users to translate their intent into clicks and drags, an impedance mismatch. AI moves us from an imperative model of step-by-step instructions to a declarative one in which you describe the outcome, similar to how SQL let you describe data instead of looping over records. He imagines asking for a report on European customers signed last quarter above 100K ARR, in text or voice.

He says it was not possible earlier: six and a half years ago he built GrowthBot, a chatbot for HubSpot software that, in his words, did not work because the technology was not there. ChatSpot is essentially its successor, now handed to a team. He does not think web and mobile interfaces will vanish, but a class of use cases can drop the translation layer.

For learning, he advises against studying something because it is worth learning. Find a problem you care about and solve it with the new tool, use the APIs, iterate and learn in public, since the internet is good at telling you why you are wrong. He closes by defining success as making the people who believed in you look brilliant.