Martin Gontovnikas on Product-Led Growth and Selling to Enterprises

Martin Gontovnikas, known as Gonto, tells Harry Stebbings that he entered growth by chance. He joined Auth0 for developer advocacy and was put in charge of self-service after the team saw that much of its revenue came from people who signed up and tried the platform. With no marketing background, he applied engineering thinking to improving the numbers.

After seven years at Auth0, he says he learned that psychology and branding matter even when selling to developers. Emails with less data and a more coherent story converted better in his experience. He also learned that rapid growth involves problems and consequential bets, not a smooth exponential curve.

When Auth0 was trying to grow annual recurring revenue from $40 million to $80 million and was unlikely to hit its goal, Gontovnikas says the company redirected half the marketing team and much of product to signups and activation. They paused product marketing and launches. He considers it a risky but necessary choice, though the interview does not independently establish its causal effect.

Courage, he argues, is acting despite fear. To enable measured risks, he lets people run a well-researched, measurable experiment costing under $10,000 without approval. If a properly designed attempt fails, he wants to know what the team learned rather than punish the failure.

A postmortem should identify what did not work and include interviews to understand why people behaved as they did. Gontovnikas keeps searchable notes and tags in a spreadsheet, and says he made a GPT from them to retrieve lessons when considering new ideas.

He defines growth as applying the scientific method to key performance indicators. Data supplies information, but in his view it does not generate the creative options to test: understanding a persona, interviewing customers and imagining different experiences are essential.

Stebbings points out that A/B tests can clearly identify a better thumbnail. Gontovnikas agrees that tests offer information, but says B2B samples are smaller and decisions often involve teams. A winning variant might still be far worse than an unimagined alternative. He favors sufficiently different tests to detect meaningful effects, and sometimes uses a test as a guardrail against making a metric worse.

A big swing often costs time from people rather than a separate cash budget, so the real question is what work they will stop doing. He suggests a portfolio of small, medium and large bets: incremental gains can help the team meet near-term goals while it waits for a larger change to work. Every bet should have a time limit.

At Auth0, an outbound campaign targeted by industry and role failed. Gontovnikas contacted prospects to ask why, without pitching them. They said they lacked a current reason to change authentication providers. The team then looked for hiring signals suggesting a technology-stack change and targeted accounts with a plausible initiative; he says this worked better.

He argues large companies must also keep making big bets, using Amazon’s failed phone and AWS as examples of a willingness to experiment. Ideally a growth mentality runs throughout a company, he says, but as product teams accumulate maintenance, support and enterprise requirements, separate growth or innovation teams can protect room for experimentation.

From the beginning, founders should think creatively, inspect results and iterate. He describes making growth work competitive by showing each person’s pipeline contribution on a dashboard. A growth manager needs creativity and an understanding of the target persona, supported by someone with deeper customer knowledge.

Gontovnikas would not hire a dedicated growth person before product-market fit; he sees finding that fit as the founder’s responsibility. Even a product-led company should begin with a closed group of six to eight design partners from its network, learn from them, and build a good initial experience before opening signups more broadly.

A blank dashboard with no useful data can make a poor first impression. So can pushing advanced, monetizable features before users understand the basics, or guiding someone into the wrong use case. In his view, first impressions are hard to recover from and onboarding should reflect the user’s real goal.

At Auth0, the team tested a step-by-step onboarding flow against one explaining the main components and leaving users to explore. The aggregate results looked similar, but interviews revealed junior developers often wanted detailed directions while experienced developers found them condescending. The team created different experiences for different needs.

For a horizontal product such as Notion, he suggests building templates and use-case content. A user’s choices before signup can help infer what they want to accomplish and tailor the ordering or options in the product. The template library is useful both to users and as a source of information about their intent.

In a bottom-up motion, people trying a product usually want to explore without a sales interruption. The challenge is recognizing when they need help or are ready to talk. Gontovnikas describes using behavior—such as moving between documentation and a dashboard without making progress—as a possible signal that a user is blocked.

At Vercel, he says, product advocates with junior technical backgrounds contacted blocked users to help them, then could connect them to sales. The team recruited people from technical support and boot camps. The idea was to make the initial interaction useful and relatable to developers rather than immediately selling to them.

To design free and paid tiers, he has clustered features by usage and placed one or two useful features from a cluster in the next plan. That can create a reason to upgrade as users grow. He describes this as a way to think about packaging, not a universally correct free-tier boundary.

He also looks for anti-retention patterns: actions that precede users dropping off. At Auth0, he says, users implementing multifactor authentication in their first week often left, while those implementing it around a month to six weeks in were more likely to stay. His interpretation is that the feature was too complex before users had learned the product.

The team compared feature adoption paths of users retained after a year with those who were not, including when each feature was first implemented. These are observational patterns in his account, not proof that timing alone caused retention. He measures retention by recurring use of the product’s core value at the frequency relevant to that product.

He values personal-project use as well as work use: a developer who knows a product personally may remember it at a future employer. He also favors letting people experience features before asking for payment. In his Auth0 example, features were not technically cut off by plan; the team periodically contacted accounts using more than their allowance and asked them to pay.

Gontovnikas says top-down enterprise sales should complement bottom-up usage. If a developer at a large company is already using a product, sales can approach relevant leaders with that context instead of starting completely cold. Ads that make the brand familiar and materials that help an internal champion explain the value are ways to support that motion.

A message to a champion’s manager should begin with a specific benefit to the user, then connect it to the manager’s goals. His example is a developer using Vercel previews to collaborate and ship faster, which may help an engineering leader deliver projects on time. He says enterprise buyers may value analyst material, while startups may care more about what makes them distinctive.

A company can package a working demonstration so a user can show it internally and lend it their own credibility. At Auth0, Gontovnikas says enterprise architects joined sales calls and whiteboarded how services fit together with prospects’ architects. That offered a conversation in the customer’s technical language, though it was initially labor-intensive.

Rather than organizing all expertise by industry, he recommends specializing in four or five common use cases. Different industries can share the same problem, and product usage can reveal which cases merit attention. Specialists can be added as new use cases become important.

Stebbings asks whether a company needs enterprise customers at all, noting the size of the small-business market. Gontovnikas agrees an SMB business is possible, but says many product-led companies eventually derive most revenue from midmarket and enterprise customers. His 80–85% estimate is a generalization from his experience, not a demonstrated industry benchmark.

He looks for midmarket or enterprise users already signing up as a signal to invest further. Selling to small businesses generally means lower revenue per customer and therefore less room for costly high-touch acquisition. Larger accounts can justify more sales investment, but require a different operating model.

Stebbings describes portfolio companies with enterprise-level acquisition and support costs but SMB-level revenue, and asks when an expected expansion is credible. Gontovnikas suggests questioning the timeline and the concrete expansion plan; he considers expansion beyond a year less likely. At Auth0, he says, a $30,000 initial deal could be followed by workshops and outreach to other teams inside the account. His timeline is a rule of thumb, not a guarantee.

The biggest enterprise mistake, he argues, is treating the move upmarket as a switch and forgetting the users who brought the product into the company. Enterprise buyers may ask for long feature checklists, but the end user still needs a good experience. He cites New Relic as his interpretation of the danger of adding feature bloat while neglecting developers; Stebbings notes that it was acquired, not shut down.

In his view, enterprise buyers balance the desire for promotion against the fear of making a career-damaging purchase. Security, compliance and analyst endorsements can reduce perceived risk, while a distinctive product may promise professional upside. This is his model of buyer psychology, not evidence that every buyer acts this way.

Gontovnikas proposes a chief technology marketing officer concept for product-led businesses. He argues that marketing’s promise and the product’s delivered experience must match: a signup can move directly from the website into the product without a salesperson explaining the gap.

His practical structure has a marketing growth team focused on activated signups and a product growth team focused on retention and conversion. They jointly own onboarding and activation, with changes requiring agreement from both sides. At Auth0, marketing wanted an immediate sales prompt; the compromise showed it to certain non-developer directors or users from very large companies while letting most users proceed into the product.

Overpromising on the website creates disappointed users when the product cannot deliver or is hard to understand. Product managers can also lose touch with users, he says; marketing’s conversations at events and online should feed back into product decisions.

For an early growth hire, Gontovnikas values what he calls hunger for glory: the drive to prove oneself. He looks for creativity, familiarity with the user persona, competitiveness and resourcefulness, sometimes valuing these traits over extensive experience or a famous employer.

He cautions that celebrated growth frameworks can be explanations written after a success rather than the process teams consciously followed at the time. For interviews he prefers open-ended exercises that give candidates a problem, research and data. Their questions, investigation and reasoning matter more to him than landing on a predetermined answer.

His own hiring mistakes went in both directions: first giving too much weight to impressive resumes when he lacked marketing experience, and later hiring too many people like himself. He now seeks different perspectives, even from people he finds personally difficult, provided they are capable and get work done.

One campaign spent $300,000 on ads and generated many signups, but few activated, stayed or converted. Gontovnikas says the team had optimized a board-friendly signup number instead of an outcome tied to revenue. His lesson is to select a KPI connected to the business result, rather than a vanity metric.

In another experiment, the team sent onboarding emails to 90% of users and no emails to 10%. He reports that the no-email group had 20% higher retention, despite improving open and click rates. The team stopped the emails and concluded that confusing content was driving people away. The interview gives no sample size or other details needed to independently assess the result.

In the rapid-fire round, Gontovnikas says he has become more convinced of the importance of marketing operations and engineering. He describes using scraped intent signals, personalized AI-generated outbound and even CEO deepfake videos. Stebbings questions whether accessible tools make the technical advantage less durable; Gontovnikas responds that custom execution may still be needed to stand out.

He expects more outbound based on inferred intent, such as hiring patterns suggesting a change in technology. He warns founders both against hiring for growth before product-market fit and against waiting too long after strong fit, when building the necessary team and operations may take several quarters.

For his client Clerk, Gontovnikas describes a sponsored YouTube experiment. A dedicated video about the product had little effect, he says; a series in which creators naturally used Clerk while building apps, with sponsorship identified at the end, performed better. He attributes the result to viewers’ curiosity and the appearance of organic use, while acknowledging the videos were sponsored. He gives no conversion figures.