「每个问题都是我们的问题」：HubSpot 增长 VP 如何把产品做成增长引擎

Chris Miller leads both product-led growth and, more recently, AI at HubSpot. He describes the AI role as building the foundational technology for AI-powered experiences and then deciding how to use them to make HubSpot's customers, the "B2B business builders," more successful than they could have been before.

Asked why he was given that expanded remit, Miller points first to timing. He joined in 2016, about a year after HubSpot launched its free CRM, when the company had not yet worked out how to turn that free product into enterprise value. He was a growth PM by instinct rather than a feature PM, and he saw the work as product-led growth before the company had a shared vocabulary for it.

He also pushed his way into conversations his title did not entitle him to. He spent time on the sales floor and in other buildings, absorbing how different parts of the business connect, which widened the size of the opportunities he could see. He admits that serendipity is harder to come by now that everything is scheduled from one Zoom call to the next.

His favorite example is a pricing and packaging pitch. At a party in Dublin, he and his designer ran into the COO, who asked what they thought. They pitched on the spot, were invited to present to the executive meeting, and, although the idea was not adopted whole, parts of it later made their way into HubSpot's pricing and he was welcomed back into that room.

Miller's first hiring trait is relentless curiosity: an insatiable desire to understand things, no fear of admitting when you don't, and a refusal to stop until you do. He says the people he has learned the most from share that behavior.

Second is resilience, especially in growth. Done properly, product-led growth mixes hygienic experimentation with ambitious bets, so failure is the norm. He cites a rule of thumb that only 20 to 30% of experiments win, and warns that teams who win 30 to 40% of the time are thinking too small, while teams without resilience start grasping for tiny wins.

The remaining traits are coachability and creativity. Growth PM is still a young discipline, so he does not demand ten years of PLG experience and expects the work to look different from shop to shop. And the best growth people, he says, are "ambivalent" about how sophisticated a solution is, taking no pride in complexity as long as it delivers the outcome.

Miller fell into product management when a startup founder decided the cure for the company's problems was to hire a PM, and Miller googled the job and asked to do it. At the time the function was far from ubiquitous and there was little content or community to learn from, so he describes his first years as painful trial and error.

For people trying to break in today he stresses structure. Smaller companies are easier to get into, but you may give up formal development and the chance to learn from battle-tested leaders. He thinks purely academic upskilling has diminishing returns because it is hard to field curveballs in a classroom, so where you start, who you report to and the company's track record all matter.

If you already work elsewhere in a company, he suggests reaching out to a PM and asking how you can make their day easier. Take work they can offload, or shadow them, to see how a team ideates, prioritizes and ships. At the very least you may win an advocate willing to gamble some capital on you despite a thin resume.

Host Lenny Rachitsky agrees that a PM's minimum job is being useful to the team, and that messing up is essential. Relationships, shifting plans and changing leaders are things you learn to handle only by getting some of it wrong.

Miller's first PM job was on a B2B2C product, where the customers were institutions that white-labeled the product to end users. With so much distance from the end user, he says the team built things to satisfy buyers and close contracts, and he thinks he shipped a lot of bad product in those years.

The inflection point was his second job, at a fitness technology company with a freemium running app, working under Fareed Mosavat. The team did guerrilla research, such as talking to runners outdoors about motivation, and had a large user data set that let them prove causation. He compares it to Dorothy arriving in Technicolor Oz, and says he learned to write a hypothesis and predicted outcome before building anything.

Miller also learns from people who chose not to become customers, including friends who rejected HubSpot. Those decisions are often emotional or brand-driven rather than calculated. Event data, he argues, shows what is happening in a product but never explains why, and Lenny's story of French Airbnb hosts refusing to connect Facebook illustrates the same point.

On careers, Miller distinguishes mentors from sponsors and advocates, who put up professional or social capital to bet on you. He thinks he bombed his first interview with Fareed, who hired him anyway. His own approach is to set ego aside, embrace being bad at things, and take hard feedback gracefully, a habit he credits to growing up playing sports.

Asked what makes HubSpot special, Miller starts with customer obsession that is real rather than marketing. He has watched fierce internal debates in which both sides were really arguing about what was best for the customer. When a decision looks hostile to customers but positive for the business, he says, the company is probably not thinking long-term enough.

In his own teams, this is operationalized by forcing specificity. Every problem statement must say whether it is a business problem, a customer problem or an efficiency problem, and PMs should ask why a business problem hasn't solved itself. They also list their assumptions and the downstream effects of a change, asking "why" and "what" repeatedly to map the blast radius.

Second is staying in the SMB and mid-market rather than crawling up to enterprise. Because revenue is spread across the whole install base, no single customer can hold HubSpot hostage to build bespoke software, and that forces decisions that benefit the largest swath of customers.

Third is culture, notably the culture code that co-founder Dharmesh published publicly. Miller says it creates internal alignment and helps attract people who want that culture, which he summarizes as humility, empathy, adaptability, remarkability and transparency. The company regularly pressure-tests it, and he cites PEER Week, an in-person product and engineering gathering in Cambridge and Dublin, as a ritual he looks forward to every year.

Miller credits Brian Balfour with injecting the first dose of PLG into HubSpot; after he left, the effort was start and stop. When Miller joined and the team took another stab, their charter was a narrow sales tool activation mission, but they threw it out and decided to fix any opportunity nobody else was looking at, asking forgiveness rather than permission.

At the time only a small share of subscription revenue was self-service, and most product-driven leads went to sales. The team found a pricing page where customers could buy that nobody was maintaining, and asked its owners if they could have it. The answer was essentially "one less code base for us to maintain."

They redesigned it around discoverability, desirability and usability, removing friction at each step. When it launched it produced a step-function change in the funnel and was a catalyst moment for the company. That success made the business more comfortable putting more on the team's plate, expanding its remit over time.

Miller says the pre-existing motion, with self-service pieces but a salesperson in the loop, still counted as product-led, though the company did not talk about itself that way yet. The team's data, experiments and research gave leadership the conviction to double down, alongside a broader desire for non-linear growth when revenue is tied to go-to-market headcount.

He calls it a fallacy, common among early founders, to treat PLG and fully self-service as interchangeable. Most large PLG companies still have many humans in go-to-market. The right mix depends on the customer, the product, how complex the billing is, whether decisions are top-down or bottom-up, and how comfortable the market is with the category. Loom and Slack need no salesperson, while other products do.

So HubSpot takes a modular approach across the customer journey. Ask at each stage whether a human is needed, and if so, put them where the cost structure is defensible. Customers with IT, security or data-migration worries may want to talk to someone, while a small team replacing post-it notes may never speak to anyone. Miller's first metrics included how much demand he sent to sales, so there was no turf war.

When founders ask how to become more product led, Miller first asks why they want it and what they mean by it. His definition is a go-to-market approach where the product's job is to grow revenue and humans are the backstop. Reasons vary widely, from top-of-funnel demand to constrained resources to revenue efficiency, and the goal determines where to start; self-service checkout is a silly first move if the problem is demand.

The most common mistake, he says, is hiring a head of growth, giving them no tooling, engineers, designer or data access, and handing them a big number. Another is expecting the quick payback of a sales hire, when PLG is still R&D and needs patience to see through.

Bad data hygiene is the third. Without properly instrumented products and self-serve data access, analyst bottlenecks will slow everything. The last mistake is giving up for lack of data, though Miller says quantitative data is just one input. Talking to ten customers can reveal a lot, and PLG at its core is the product selling its own value, which you can do without a sophisticated experimentation practice.

Miller says tactical loops are hard to achieve in B2B SaaS, since the best ones come from user-generated content in consumer products. HubSpot instead has a macro flywheel of attract, engage and delight, guided by the principle of giving value before extracting it, which he says sat at the core of inbound marketing.

Before PLG, that meant content, white papers and eBooks filling the top of the funnel. Now it takes the form of free software designed not to run out of value on day one. Customers who engage deeply hit its limits, buying becomes a no-brainer, and delighted customers become advocates who bring peers in. He says a large share of revenue now flows through the product.

He argues for diversifying channels because an algorithm change at Google or Apple, or the effect of generative AI on search, can disrupt a funnel overnight. HubSpot is experimenting with microapps, an idea going back to Website Grader, and has built a brand kit generator and an email signature generator. Dharmesh's AI copilot ChatSpot, he says, drew an unexpected number of signups.

Asked what most accelerated growth recently, Miller names Covid. HubSpot expected the worst but caught a tailwind as companies suddenly needed digital marketing. Following the internal phrase "never waste a good crisis," it used goodwill pricing and temporary leniency, and the reduced friction accelerated growth in his free and starter business.

In the lightning round he recommends *Everybody Lies*, about people being honest with Google but not on surveys, and *Chop Wood Carry Water*, about loving the mundane parts of the process. His interview favorites are estimation questions, such as how many people cross the Longfellow Bridge in a week, to watch how candidates reason, and asking how recent colleagues would describe the candidate when they are out of the room.

His life motto is "the details matter." He ties it to Stripe's notion that PMs should have taste, meaning going deep enough on something to hold strong, informed, even polarizing opinions. Miller also mentions that he advises founders, does some angel investing, and is hiring a group product manager for HubSpot's AI platform team.