Overview of 20VC: The 8 Moats of Enduring Software Companies (with Gokul Rajaram)
This episode of 20VC (host Harry Stebbings) features Gokul Rajaram — operator-turned-investor and founder of Marathon — on how to evaluate software durability and defensibility in an AI-driven world. The conversation centers on Gokul’s “eight modes” of moat-building, lessons from his time at Google, Facebook, Square and DoorDash, the practical effects of AI on product & pricing, and tactical advice for founders and investors (seed → growth). The episode also touches on why some remote early-stage companies are struggling and why many young dropouts are now “AI-maxed” founders.
Key takeaways
- Durable software companies combine multiple defensibility modes — one strong mode alone is rarely enough.
- The eight modes (data, workflow, regulatory, distribution, ecosystem, network, physical infrastructure, scale) are a practical rubric to score defensibility; 4+ modes = strong.
- AI shifts the product landscape: bolt-on AI often has limited upside unless it redefines the experience and economics (agentic workflows).
- Early-stage remote-first teams struggle with iteration speed and alignment; partial in-person cadence is increasingly important.
- For investors: be thesis-driven, mix seed/incubation with A rounds, prioritize ownership and the ability to double-down, and evaluate IRR (not just MOIC) when deciding whether to sell.
The eight modes of moat (short rubric + examples)
Gokul reframes Hamilton Helmer’s ideas into eight practical defensive “modes” to score companies:
- Data mode — proprietary, hard-to-replicate datasets that improve with time (example: Spotify Discover).
- Workflow mode — deeply embedded tools that operate inside a company’s processes (strong: NetSuite; weak: lighter tools like Zendesk).
- Regulatory mode — licenses, regulated infrastructure and certifications that raise barriers (example: Coinbase with MTLs and compliance).
- Distribution mode — exclusive distribution channels or entrenched partners (example: QuickBooks + CPA network).
- Ecosystem mode — third-party apps and integrations that make the platform sticky (example: Shopify).
- Network mode — marketplace/network effects and liquidity (example: DoorDash).
- Physical infrastructure mode — atoms, logistics, facilities that are costly to replicate (example: fulfillment centers).
- Scale mode — cost advantages and supply chain/production scale (examples: Amazon, TSMC).
Practical scoring: assign 1 point per mode present. Companies with 4+ points are generally durable; 2–3 points signal weakness; 0–1 points is dangerous.
Lessons from Gokul’s operator background (impact on investing)
- Google: product-first mindset — “a remarkable product” is foundational. GTM can’t save a non-remarkable product.
- Facebook: distribution mastery and multiplayer/network effects — multiplayer products create unique defensibility.
- Square: the power of being multi-product — products should be adjacent and serve different roles (retention vs profit pool). Don’t confuse objectives.
- DoorDash: operations excellence in the physical world — operational rigor and talent matter when software touches atoms. Tough short-term choices (e.g., waiving restaurant fees in COVID) can be right long-term.
Notable quote (paraphrase): “If there is not a remarkable product, all the go‑to‑market distribution in the world will not save you.”
AI, product strategy and pricing implications
- Bolt-on AI often has a ceiling. The AI upgrade works when it reframes what the product does and changes UX/economics (not just a search box).
- Agentic workflows (AI doing work on users’ behalf) shift value from seat/access to outcomes — this drives consumption/outcome-based pricing rather than seat pricing.
- Seat-based pricing remains valid for access products (predictability for enterprise buyers). Work-products should be priced on output/consumption.
- Data portability and improved tooling will lower switching costs; brand becomes less of a defensive moat in B2B.
- Verticalized AI businesses can succeed if they own full-stack for the vertical (cover product + services spend). First they attack BPO budgets, then attrition/attrition replacement, then potential layoffs.
- Rapid model improvement shortens product roadmaps — companies must iterate faster and fine-tune models for their customers.
Examples: document processing becomes qualitatively different with new models (instant contract insights). Notion and Harvey illustrate different approaches to AI integration.
Market durability, metrics and sector calls
- Core indicators of business quality: gross retention and net revenue retention (NRR). Durable revenue > high short-term growth.
- Non-consumption markets (products people didn’t previously buy) are highly valuable — Shopify, Figma, Gamma are cited as examples.
- Vertical SaaS: to reach very large outcomes you often must own the full stack and capture services/payroll budgets as well as software budgets.
- Margins: early negative/low margins are okay; focus on ability to increase prices (pricing power) and lower costs over time rather than short-term margins.
Atlassian vs Monday example: using the eight modes, Atlassian scores higher defensibility (proprietary data, workflow, ecosystem) and may be oversold; Monday scores weaker.
Investing strategy & practical advice for VCs and LPs
- Seed vs Series A: at very early stages price matters less if you have conviction. At Series A and beyond, pricing and ownership become more critical to returns.
- Fund construction: smaller early-stage funds should mix seed/incubation bets + later-stage deals; mega funds play a different game (deploy widely at A and double down later).
- Reserves & concentration: doubling down on winners (exploit) can outperform broad diversification — but balance concentration risk. Founder/partner conviction around which companies to double-down on matters.
- Founder access: “proprietary access” claims are common; real differentiation is demonstrable value-add (talent placement, go-to-market, customer intros) and founder relationships. LPs should call founders in the GP’s portfolio to validate claims.
- Selling/liquidity: evaluate go-forward IRR, not just MOIC. If a single asset will return a large chunk of the fund and IRR is low, consider selling some. Fred Wilson’s heuristic: sell 1/3, hold 1/3, trade 1/3 for fully liquid assets.
- Preemptive mega-rounds: mega funds enable fast liquidity but can orphan founders if the partner leaves; they play a different strategic game than smaller VC firms.
Quick-fire — notable opinions & practical rules
- Remote early-stage teams: changed mind — pure remote early-stage companies often fail due to slow iteration and founder misalignment. Recommendation: in-person at least a few days per week.
- Young founders / dropouts: bullish. Younger founders are “AI-maxed” (adopt AI tools faster) and Gokul has been investing more in dropouts recently.
- Advice to young founders: get 2–3 years of work experience at a good company before founding — it builds skills and relationships.
- Pricing rule of thumb: distinguish “access” products (seat-based) vs “work” products (outcome/consumption-based).
- Biggest angel win: Figma (estimated 500–1,000x at IPO). Biggest regret: passing on Quince.
- What excites Gokul: ambitious entrepreneurs tackling hard societal problems enabled by AI.
Notable quotes
- “If there is not a remarkable product, all the go‑to‑market distribution in the world will not save you.”
- “One mode alone is not enough — score across the eight modes; 4+ modes and you’re pretty damn secure.”
- “Bolt‑on AI has a ceiling unless you change the product experience and economics.”
Actionable checklist (for founders & investors)
For founders:
- Build a remarkable core product (10x if possible).
- Map which of the eight modes you currently have; prioritize building at least 3–4.
- If adding AI, ask: does this change the product experience and economics or just improve an existing feature?
- For vertical plays, aim to own the full stack and target services/BPO budgets, not just software budgets.
- Think early about pricing: access vs outcome.
For investors:
- Score startups across the eight modes; weigh data and workflow heavily for pure software early-stage deals.
- Prioritize founders with demonstrated execution velocity and a plan to compound their data or workflow integration.
- Mix seeds/incubation to maintain exposure to early founders and reserve capital to double down on winners.
- When evaluating liquidity choices, compute go-forward IRR and don’t focus solely on headline multiples.
Episode highlights / useful examples
- Spotify Discover = data moat (decade+ listening data).
- Coinbase = regulatory moat (money transmission licenses).
- QuickBooks = distribution moat (CPA network).
- Shopify = ecosystem moat (apps & third-party integrations).
- DoorDash = network + operations moat.
- NetSuite = deep workflow moat vs Zendesk’s lighter workflow positioning.
This episode is a concentrated playbook on measuring defensibility in a rapidly changing tech landscape where AI both threatens traditional moats (data portability, cloning) and creates new opportunities (agentic workflows, services displacement).