Noah Weiss on Product-Led Growth and Scaling Slack

Noah Weiss says speed is among the most important features of any product. The Slack chief product officer discusses product strategy, enterprise growth and leadership with Harry Stebbings, drawing on his own company’s experience.

At Google, Weiss encountered a 70/20/10 product allocation: 70% on known customer value, 20% on promising areas to refine and scale, and 10% on more distant bets. Leadership also asked what a tenfold increase in ambition would look like. He sees the portfolio mindset as useful, not a fixed ratio for every company or team.

Slack considers how much of a roadmap goes to maintenance, performance and quality, customer delight and new ideas. Weiss says a mature area such as advertising might reasonably allocate far more to its core than the company-wide Google model suggested.

At Foursquare, Weiss learned that a company facing headwinds can learn things a rapidly growing company may not. Product-market fit is not permanent: a product must renew it as audiences and competition change. He recalls the excitement of early mobile location and social features, followed by the need to serve users beyond novelty-seeking early adopters as Instagram changed how people shared experiences.

Founders often initially build for a problem they know personally. Moving beyond that audience requires humility and curiosity about people unlike the original team. Slack originally imagined serving teams of five to 50, but noticed separate small teams at large companies adopting it and learned that the organization had different needs from its users.

Around 2016–17 Slack spoke with enterprise executives and studied security, compliance, scale and administrative controls. Weiss says the team did not take every request literally; it needed to understand organizations much larger than its own before deciding how to evolve the product.

Product principles capture a team’s beliefs in shared language that helps people make decisions and judge craft. As Slack grew, founder Stewart Butterfield could no longer review everything, so the company articulated principles such as “be a great host,” “don’t make me think,” “protect the path” and “take bigger, bolder bets.”

Stebbings challenges the principles as generic. Weiss says their meaning depends on organizational context: as teams acquire narrow metrics, they can drift toward local, incremental optimization. “Bolder bets” reminds Slack to balance refinement with investments that advance the category.

“Be a great host” means anticipating needs, like placing fresh towels on a guest’s bed so they know which ones to use. Weiss applies that care to Slack as a digital environment where people spend their working day. He thinks many startups wait too long to articulate culture and standards before the founder becomes a bottleneck.

Stebbings favors speed over extensive product debate. Weiss distinguishes reversible two-way-door decisions, which empowered teams should usually make locally, from harder-to-reverse choices such as pricing, brand or a multi-year category investment. The latter justify broader executive discussion; they should not dominate everyday decisions.

Slack’s product, design and engineering organization had around 1,200 people at the time. Weiss says product pillar leaders ran weekly reviews to unblock their teams, while executives met biweekly or monthly to focus on major launches, portfolio choices and difficult decisions rather than review every feature.

Weiss sees one benefit in hybrid reviews: more designers, engineers and product managers can participate and gain context without a large physical meeting becoming a performance. Stebbings notes other leaders find hybrid discussion worse; Weiss agrees it can be less fluid for executives and that hardware teams may need physical access.

Slack moves more preparation out of live meetings: a written brief sets context, while designers and engineers record short demos of prototypes or working software. Participants can review these asynchronously and use the meeting for decisions and feedback.

One unresolved process problem is seeing the current state of all major work. Weiss says Slack channels can be noisy, so the organization still uses a manual spreadsheet and a Monday meeting to review quarterly priorities. He finds the workaround useful but inelegant.

Weiss says Slack began using practices later labeled product-led growth before he had heard the term. Its central aim was a consumer-quality experience for people at work: workplace software should not have to be more painful than what users enjoy outside work.

Slack’s stated mission is to make working life simpler, more pleasant and productive. Weiss says that standard shapes decisions and hiring; many product and design staff came from outside enterprise software. Simple means understandable, he argues, not necessarily fewer clicks: a guided flow with one decision per screen can build confidence.

A major product-led growth mistake was offering a generous free plan for years without giving users a rich trial of the paid product. Research revealed some users did not realize a paid tier existed and assumed the free plan’s limits were inherent to Slack.

Slack built guided in-product trials to let users experience paid capabilities. Weiss says this was one of its most important steps in improving self-service paid conversion. The mindset shifted from assuming users would discover value to actively helping them experience it; he does not supply a numerical lift.

Another lesson came from Slack Connect, which lets organizations share channels. The team worried about breaking users’ mental model of a private workplace and kept it in beta for roughly two years. Weiss believes bringing it to market sooner might have captured more network effects, while acknowledging that it still found value.

Asked whether innovation slowed, Weiss says that criticism was most true around 2018–20, when Slack prioritized enterprise security, compliance and scale. In his account, this diverted attention from product innovation. He says the pandemic renewed demand for ways to connect and that the pace subsequently increased with products including Huddles, clips and Canvas.

Once enterprise blockers such as security controls are addressed, Weiss sees diminishing returns in devoting most effort to buyers alone. Teams also need new capabilities. He frames the pandemic-era shift as a change in customer expectations and Slack’s understanding of its role.

For startups moving upmarket, Weiss recommends hiring people with enterprise domain expertise instead of reinventing security and compliance requirements. At the same time, he warns against over-rotating toward a conservative buyer and losing the speed and product innovation that attracted users.

Some categories, including certain security or infrastructure products, may need to begin with an enterprise buyer; not every company needs product-led growth. For an SMB-first business, independent pockets of adoption inside a large organization can signal demand before a central sales conversation.

Weiss gives the illustrative example of hundreds of employees already using a product on corporate cards. A conversation with the CIO can then expose missing controls needed for an enterprise agreement. That discovery begins the enterprise journey rather than eliminating the work.

When buyer and user needs compete, Slack says it prioritizes the end user while balancing organizational requirements. Weiss reports that more than 85% of its enterprise customers with over 1,000 users began through self-service. This is a company-reported figure and not a universal benchmark.

Weiss places Slack’s platform, workflow automation, Connect and audio or video features among its larger bets. The company debated whether Connect’s benefits outweighed added complexity and whether it could be substantially better than ubiquitous email for particular uses.

For a major new capability, Slack starts with an intentionally rough internal prototype. If employees find it valuable, the team expands use and measures adoption, repeat usage and depth. It then runs progressively larger pilots with diverse customers, combining surveys, conversations and usage data before a broad launch.

A counterexample was Posts, a rich text experience Slack incorporated after an acquisition. Weiss says adoption remained low despite apparent usefulness: slow loading or typing compared poorly with existing document tools. For a ubiquitous task, speed and quality had a much higher bar than the team met.

Slack later introduced Canvas as a different approach to the area. Weiss says the Posts experience was humbling: customers had good-enough alternatives, so apparent value was insufficient when performance did not meet expectations.

On buy versus build, Weiss says Slack struggled to acquire a product and integrate it faster than it could have built one. An acquisition can appear attractive when ambition exceeds a team’s capacity, but product and technical integration are hard. He distinguishes this from talent acquisitions, which he says worked well for Slack.

Large, frequent acquirers can build dedicated integration expertise, whereas Slack’s smaller acquisitions were often one-offs. Weiss does not generalize this experience to a large purchase such as Salesforce’s acquisition of Slack.

His own leadership challenge is making product managers feel connected to one another. Their closest daily colleagues are often engineers and designers on individual teams, leaving the PM organization feeling isolated. He is still looking for ways to create shared context, learning and energy across different areas.

A north-star metric can help unite PMs around impact, but Weiss says enterprise product changes may not map directly to willingness to pay. Metrics can align the head without necessarily creating the emotional connection and energy of a team.

He cautions against rigid definitions of the PM job or copying frameworks as shortcuts. Product management has often filled gaps: understanding customers, defining why a product matters, setting longer-term strategy and helping other disciplines succeed. The role should adapt to what the team and customers need.

In the quick-fire round, Weiss advises hiring a head of product when a product founder becomes a bottleneck to fast, high-quality decisions. He says product and engineering or design should be close partners; in enterprise software, sales can be a source of tension when customer promises and delivery capacity diverge.

Asked whether product is art or science, he estimates a mature, data-rich feature may be 80% science and 20% art, while early products or new capabilities may be 75% art and 25% science. These are illustrative judgments: vision, editing and creativity matter before there is enough scale for optimization.

For a PM seeking promotion, Weiss emphasizes measurable team impact on customers or the business. For a new CPO, he recommends learning the team, culture, customers and strategy before making sweeping changes; start with clear execution improvements, then address team gaps, and change fundamental strategy last.

As a new parent, Weiss is interested in baby technology, and he names OpenAI as a company with unusually interesting product choices. With unlimited resources, he would also like Slack to serve more communities and work-adjacent uses, an area he says the company has repeatedly set aside to maintain focus.

On generative AI, Weiss says product promises and prominence should match the quality of the underlying data and model. He advocates presenting AI as an assistant whose answers should be checked or modified, rather than encouraging unconditional trust in potentially hallucinated output.

He praises OpenAI for packaging language-model research in a simple interface that let users experience a new way to interact with machines. He expects consumer products and infrastructure for other companies to be important areas to watch; his forecast reflects the 2023 moment of the interview.