Building company culture like a product: HubSpot's iterative approach

Executive overview

Most companies treat culture as a static declaration — words on posters that nobody believes. HubSpot co-founder Dharmesh Shah applied software engineering discipline to culture instead: define narrowly, ship early, iterate on feedback.

The result is the HubSpot Culture Code, grown from 16 slides in 2013 to 128 today. It is built around four core components — solving for the customer, extreme transparency, bold risk-taking, and diversity — each treated as a living feature, not a fixed rule.

Culture is the second product every company builds, and no product is ever done.

Starting with hostile user research

• Dharmesh reluctantly took on culture after co-founder Brian Halligan was challenged by a CEO peer group.
• A simple company-wide email about culture drew the harshest feedback Dharmesh had ever received.
• Employees associated "culture talk" with performative posters and toxic misrepresentation.
• He treated this blowback as actionable product research, not personal criticism.
• First output: 16 slides answering one question — what attributes predict success and happiness at HubSpot?
• Framing the culture deck as "culture Python" let him apply coding principles: narrow scope, launch early, iterate.

Iterating with customers in the loop

• Version 1.0 was validated quickly; teams said it helped but didn't go far enough.
• Subsequent versions added decision-making frameworks based on direct team feedback.
• Surveys, focus groups, and one-on-ones drove each iteration cycle.
• Publishing the deck publicly — inspired by the Netflix culture deck and Patty McCord's encouragement — unlocked feedback at scale.
• Netflix CEO Reed Hastings called it brilliant; Dharmesh treated that as validation, not a finish line.
• The deck now has 128 slides and is still being updated.

Component 1: Solve for the customer

• HubSpot holds monthly exec meetings where an unhappy customer joins live — the team collectively experiences the pain, not just hears about it.
• Engineers are pushed (sometimes "draconian-ly") to talk directly to customers, not just their product area.
• Solving for the customer means the collective customer base, not individual requests.
• Direct customer contact is self-reinforcing: exposure builds empathy, which drives better decisions.
• Dharmesh's ISP example — an outage email that never said "sorry" — crystallised what customer-hostile culture looks like in practice.

Component 2: Extreme transparency

• Transparency started by default: the first employee got access to everything, and the rule was never changed.
• At IPO in 2014, lawyers suggested five to seven "designated insiders" for sensitive financial information.
• Dharmesh and Brian designated every employee an insider — ~800 at IPO, ~7,000 today.
• Limiting insiders would have broken a core cultural value; the unconventional move preserved it.
• Transparency empowers individuals to make good decisions on the company's behalf without escalation.

Component 3: Bold risk-taking

• Every employee is expected to allocate some percentage of their time to experiments that could be big bets.
• If everything succeeds, it wasn't a real experiment — some failure rate is required by design.
• Decisions are not made by consensus; one person is designated responsible for each risky initiative.
• Everyone has a voice, but not a vote; middle-ground compromises are explicitly avoided.
• Once a decision is made, the team commits fully — "sailing the ship" — rather than second-guessing.

Component 4: Diversity as a product requirement

• HubSpot's early team was dominated by MIT MBA graduates — efficient, but narrowly monocultural.
• A hallway photo wall of all employees made the lack of diversity visually undeniable.
• "Hire for culture fit" is high-risk without a written, defined culture — it defaults to "hire people like us."
• Changes included appointing a director of diversity, inclusion, and belonging, and tracking diversity metrics.
• A team that doesn't reflect the diversity of its customer base cannot genuinely empathise with those customers.

Culture as a permanently evolving product

• The mistake is trying to preserve startup culture; the goal is to evolve it.
• Every company builds two products: one for customers, one for its people.
• HubSpot treats culture as iterative feature development — retiring things that no longer fit, adding new "feature requests."
• Preserving culture in amber — the "10 commandments in stone" mindset — misunderstands what companies are: living, evolving communities.