Table of Contents
Gong's Chief Product Officer reveals how working with 6-12 design partners per pod and extreme team autonomy creates a 95% feature success rate in this deep dive into their unconventional product organization.
Eilon Reshef, co-founder and CPO of Gong, shares the secret behind building one of the most ubiquitous B2B products in tech through radical customer collaboration and trust-based team structures.
Key Takeaways
- Gong achieves close to 100% feature adoption by working with 6-12 design partners on every new product before launch
- Their pod structure gives autonomous cross-functional teams complete ownership over outcomes rather than metrics
- Quick decision-making on 51/49 choices prevents analysis paralysis and maintains velocity without significant quality loss
- Early AI adoption taught them that LLMs alone don't solve everything—specialized expertise and measurement remain crucial
- The "spiral method" enables rapid learning of complex topics by iteratively talking to experts until no new information emerges
- Extreme customer focus started with just 5,000 potential companies meeting seven specific criteria before expanding globally
- A dedicated research coordinator manages design partner relationships, removing administrative burden from product managers
- Trust-based autonomy requires leadership willingness to sacrifice some visibility for higher team velocity and engagement
Timeline Overview
- 00:00–06:33 — Pod Model Foundation: How Gong replicated their initial team structure of PM, UX designer, and engineers into autonomous pods responsible for outcomes like sales engagement and conversation intelligence
- 06:33–15:10 — Design Partner Integration: The extreme approach of working with 6-12 design partners per pod, including live product demos with save buttons that don't work yet, coordinated by a dedicated research coordinator
- 15:10–27:15 — Autonomy and Trust Philosophy: Why giving teams complete decision-making authority creates better results, drawing parallels from picnic planning to parenting and the challenges of getting peer leadership buy-in
- 27:15–35:50 — Speed and Decision-Making: The Isaac Asimov-inspired approach to making quick decisions on close calls, recognizing that 51/49 choices rarely create dramatically different outcomes
- 35:50–41:36 — AI Development Lessons: Early machine learning adoption insights, why companies need specialized AI expertise beyond LLMs, and how pod structure enables rapid AI iteration
- 41:36–46:35 — Customer Focus and Failure Stories: Starting with 7 constraints limiting the addressable market to 5,000 companies and lessons from previous company mistakes in horizontal expansion
- 46:35–End — Lightning Round: Book recommendations, life philosophies, and practical insights including the dishwasher cutlery basket productivity hack
The Revolutionary Pod Model: Autonomous Teams Built for Outcomes
Gong's pod structure emerged organically in 2016 when the company faced its first scaling challenge at around 50-60 employees. Rather than adopting traditional functional silos, they chose to replicate their successful initial team composition: one product manager, one UX designer, a fractional writer, a fractional analyst, and five to seven engineers led by a technical team leader.
Each pod operates with remarkable autonomy, receiving broad outcome-based missions rather than specific metrics to hit. For example, a pod might focus on "sales engagement" or "conversation intelligence" rather than increasing a particular usage metric by a specific percentage. This approach allows teams to interpret challenges creatively and develop solutions that address root customer problems rather than gaming measurement systems.
The pod structure extends beyond core product and engineering to include "virtual pod" members from sales, customer success, and product marketing. This integration ensures that product development maintains alignment with market needs and commercial realities without requiring constant executive oversight. Sales team members, while typically focused on their own responsibilities, can provide crucial input about customer conversations and deal dynamics when they have capacity to participate.
This organizational model requires significant trust from leadership, as executives must accept reduced visibility into day-to-day decisions in exchange for higher team velocity and engagement. Pod leaders take responsibility for soliciting feedback and driving review processes rather than having oversight imposed upon them, creating a culture where teams proactively seek input when needed rather than defensively responding to mandatory check-ins.
Design Partner Integration: Customer Collaboration at Unprecedented Scale
Gong's most distinctive practice involves working with 6-12 design partners for every new product or feature development. This extreme customer collaboration goes far beyond traditional user research, with product managers showing half-built functionality to customers and promising specific improvements by their next meeting. In one memorable example, a PM demonstrated a forecast product's save button that generated an error message, then committed to making it functional within a week.
The design partner process varies based on the development context. New products follow a more structured timeline with weekly or bi-weekly meetings to demonstrate progress and gather feedback. Feature enhancements or extensions might involve less formal engagement, with PMs recruiting specific design partner types—such as non-English speakers for internationalization features—and conducting focused sessions before moving to the next development phase.
To manage the administrative complexity of coordinating relationships with hundreds of design partners across dozens of pods, Gong employs a dedicated research coordinator. This role, inspired by talent acquisition recruiting coordinators, handles outreach through their micro-CRM system and schedules meetings based on PM specifications about target customer profiles. This approach eliminates the scheduling and coordination overhead that could otherwise consume PM time and energy.
The design partner program leverages Gong's core product advantage: since all customer conversations are recorded, the team can easily search their conversation database to identify customers who have expressed specific needs. This capability enables precise targeting when recruiting design partners for particular features or use cases, ensuring that feedback comes from genuinely interested and relevant customers rather than generic user panels.
Autonomy Philosophy: Trust as a Competitive Advantage
Reshef's approach to team autonomy stems from personal philosophy rather than management theory, rooted in the belief that people perform better when allowed to be themselves and make decisions according to their judgment. He illustrates this with a story about school picnics where traditional list-based coordination produces lowest-common-denominator results, while simply asking people to "bring your own thing" results in diverse, high-quality contributions and higher participant satisfaction.
This philosophy extends to parenting, where Reshef avoids installing protective software on his children's devices, instead negotiating time limits and having his daughter request that he install limiting software to help her maintain self-control. The pattern emphasizes personal responsibility and intrinsic motivation over external controls, creating stronger long-term behavior patterns and higher engagement.
In the product context, autonomy means that team members can act on customer feedback without seeking approval, with the expectation that they will exercise judgment about when to escalate or seek additional input. Teams are "not going to be punished" for making decisions based on customer input, but they bear responsibility for determining when they have sufficient information versus when they need management perspective or broader team input.
The autonomy approach requires cultural alignment beyond just the product organization. Sales leaders asking for product roadmap details and finance teams requesting ROI justifications for every feature decision can undermine the trust necessary for autonomous operation. Leaders must accept some loss of visibility and control in exchange for the benefits of higher velocity and team engagement.
The Science of Quick Decision-Making
Reshef's decision-making philosophy draws inspiration from Isaac Asimov's science fiction story "The Machine That Won the War," where a supposedly sophisticated computer system actually provides random outputs, but people's faith in the system enables effective decision-making. The lesson: when facing close-call decisions (51/49 rather than 70/30), the quality difference between options is minimal, making speed more valuable than extended analysis.
This approach recognizes that human decision-making capacity functions like a physical resource—making decisions requires mental energy similar to physical exertion. Extended deliberation on close calls depletes this resource without proportionally improving decision quality. Reshef notes that his decision quality doesn't significantly improve when given two weeks to ponder versus being forced to decide immediately, provided he has sufficient domain expertise.
The key distinction involves one-way versus two-way door decisions, with appropriate time allocation based on reversibility and impact magnitude. Most day-to-day product decisions fall into the reversible category, where the cost of being wrong is lower than the cost of delayed action. This creates a bias toward action that maintains momentum and allows teams to learn from real market feedback rather than theoretical analysis.
Domain expertise becomes crucial for effective quick decision-making. Reshef emphasizes that this approach works within areas of established competence but would require more careful analysis when entering completely new domains. The years of experience in sales technology and customer behavior provide the foundation for confident rapid decisions within Gong's core market.
AI Development: Beyond LLM Hype to Sustainable Advantage
Gong's early adoption of machine learning (before it was rebranded as AI) provides valuable lessons for companies now integrating AI capabilities. When launching, they deliberately avoided the "AI" terminology because customers associated it with unreliable automated decisions. This early experience shaped their understanding that successful AI implementation requires more than just large language model integration.
The company's current AI architecture combines LLMs with specialized models built for specific use cases. Deal prediction, for example, cannot be effectively handled by general-purpose language models because it requires domain-specific training data and highly specialized algorithms. This hybrid approach leverages LLM strengths for text processing and generation while maintaining custom models for prediction and analysis tasks.
Successful AI product development requires maintaining core competencies in data science and machine learning rather than outsourcing everything to foundation model providers. Companies need expertise to determine what problems LLMs can solve versus what requires specialized approaches, how long different implementations will take, and what input data and quality metrics are necessary for success.
The measurement and optimization of AI systems remains crucial for moving beyond initial implementations. While Version 1 of an AI feature might rely entirely on an LLM, subsequent improvements require rigorous measurement frameworks to understand performance gains. Gong employs ELO rating systems (borrowed from chess) and other specialized metrics to track AI system improvements over time.
The Spiral Learning Method: Mastering Complex Topics Rapidly
Reshef's "spiral method" for learning complex subjects provides a systematic approach to quickly developing functional expertise in new domains. The process begins by finding any knowledgeable person and asking basic questions, then requesting recommendations for additional experts to consult. This creates an expanding network of knowledge sources that gradually builds comprehensive understanding.
The method follows a temperature-cooling process similar to annealing in physics, where understanding gradually crystallizes through repeated exposure to similar concepts from different experts. Early conversations might be completely incomprehensible, but by the fifth conversation, learners typically understand 50% of the content with 50% being genuinely new information. The process continues until conversations yield only 5-10% new information, indicating functional mastery at the desired level.
This approach proved effective when Reshef took a sabbatical to study deep learning, ultimately leading to early Nvidia stock purchases and Gong's AI-first product strategy. The spiral method also applies to understanding customer segments, where talking to multiple account managers eventually reveals consistent patterns about their unique needs compared to new business salespeople or contact center representatives.
The spiral method works particularly well in technology domains where practitioners are generally willing to share knowledge with genuinely interested learners. The key is managing the ask appropriately—requesting reasonable amounts of time and demonstrating progressive learning that makes subsequent conversations more valuable for both parties.
Extreme Customer Focus: The Power of Radical Constraints
Gong's initial customer focus demonstrates the counterintuitive power of extreme market narrowing. Instead of pursuing the largest possible addressable market, they applied seven specific constraints: companies selling in the US, in English, over video conference (specifically WebEx), selling software worth $1,000 to $100,000. This reduced their potential market to just 5,000 companies.
This extreme focus enabled several crucial advantages. First, it created a "small pond" effect where customers talked to each other, enabling viral spread that is rare in B2B markets. The company heard about job candidates who refused offers from companies not using Gong, directly leading to new customer acquisitions. Second, it allowed for much more focused product development, avoiding the complexity of serving multiple industries with different terminology and technology needs.
The constraints followed Geoffrey Moore's "Crossing the Chasm" methodology but taken to an extreme level based on lessons from Reshef's previous company. That earlier venture attempted to serve customers across completely different industries—L'Oreal, American Express, and Cisco—making it impossible to develop focused expertise or create scalable sales processes. The 20 salespeople they hired all failed because the company lacked both product-market fit and a repeatable customer profile.
This focused approach enabled Gong to develop deep expertise in their chosen market segment before expanding. The viral effects within their narrow customer base created momentum that would have been impossible with a horizontal approach. Only after establishing strong product-market fit within their constraints did they begin expanding to adjacent markets and use cases.
Building High-Performance Teams: Lessons from Early Failures
Reshef's previous company provided crucial negative lessons that shaped Gong's approach. The earlier venture made classic startup mistakes: attempting horizontal market expansion without product-market fit, scaling the sales team before establishing repeatability, and pursuing growth based on investor pressure rather than customer validation. These experiences created a determination to avoid similar patterns at Gong.
The hiring of 20 salespeople before achieving product-market fit illustrates the dangers of premature scaling. Without a focused customer profile and repeatable sales process, individual salesperson success became impossible regardless of their capabilities. This reinforced the importance of achieving genuine product-market fit within a narrow segment before attempting to scale customer acquisition.
The experience also highlighted the importance of resisting external pressure to scale before achieving fundamental business metrics. Investors encouraged rapid hiring despite weak customer validation, leading to expensive lessons about the sequence of startup development. Gong's methodical approach to expansion reflects these hard-learned lessons about the importance of foundational strength before pursuing growth.
These failures also shaped Gong's approach to product development, emphasizing the importance of design partner validation before committing significant resources to new features or products. The close customer collaboration model directly addresses the risk of building products that don't find market adoption, ensuring that development resources focus on validated customer needs.
Conclusion
Gong's success demonstrates the power of extreme customer collaboration combined with autonomous team structures. Their approach challenges conventional wisdom about product development by showing that deep customer integration and rapid decision-making can create sustainable competitive advantages. The combination of design partner programs, pod autonomy, and quick decision-making creates a learning organization that adapts faster than competitors while maintaining strong customer relationships.
The model requires significant cultural commitment from leadership, demanding trust in team judgment and willingness to sacrifice some control for higher velocity. However, the results speak for themselves: near-universal feature adoption, strong customer loyalty, and rapid growth in a competitive market. This approach offers valuable lessons for any B2B company seeking to build products that consistently meet customer needs while maintaining development speed.
Practical Implications
- Design Partner Programs: Implement systematic customer collaboration with 6-12 design partners per major product initiative, using dedicated coordination resources to manage complexity
- Autonomous Team Structure: Organize cross-functional pods around outcomes rather than metrics, giving teams decision-making authority while expecting them to drive their own review processes
- Decision-Making Speed: Accelerate close-call decisions (51/49) while investing appropriate time in one-way door decisions, recognizing that extended analysis rarely improves decision quality significantly
- AI Development Strategy: Maintain specialized ML/AI expertise rather than relying entirely on LLMs, focusing on measurement frameworks and hybrid approaches for different use cases
- Learning Systems: Adopt spiral learning methods for quickly developing functional expertise in new domains through iterative expert conversations
- Market Focus: Start with extremely narrow customer segments to enable viral effects and focused product development before expanding to adjacent markets
- Cultural Requirements: Build leadership alignment around trust and autonomy, accepting reduced visibility in exchange for higher team velocity and engagement
