Todd Jackson on Product-Market Fit Levels, Customer Discovery, and the Four Ps
Todd Jackson, Partner at First Round Capital, in conversation with Lenny Rachitsky. The episode presents First Round’s four-level PMF framework for B2B companies, the three dimensions of demand, satisfaction, and efficiency, the four-Ps pivot model, and a dollar-driven approach to customer discovery. Based on First Round’s PMF Method programme at pmf.firstround.com.
Key ideas
- Product-market fit has four levels, not a binary state. Nascent (L1), developing (L2), strong (L3), and extreme (L4) describe a progression that the best B2B companies typically complete over four to six years. Most founders believe they either have PMF or they don’t; in practice they are somewhere on a spectrum, and the interventions appropriate to each level differ.
- The $100 vending machine fallacy. Extreme PMF requires all three: demand, satisfaction, and efficiency. WeWork and Casper achieved demand and satisfaction but never solved efficiency. A vending machine that dispenses $100 for $1 has perfect demand and perfect satisfaction — and it is not a business.
- The four Ps as a pivot map. Persona, Problem, Promise, Product. When stuck at any level, systematically examine which of these four elements is misaligned. Lattice kept its persona (HR leaders) and changed the other three. Vanta changed all four. Plaid changed three and kept its product infrastructure. The framework forces a structured analysis rather than an emotional response.
- 60% of B2B companies never pass L2. Most plateau at 5–25 customers and never open the demand floodgates required to reach L3. The most common error: mistaking an important problem for an urgent one, or selling a “nice to have” when only “must have” drives consistent close rates.
- Dollar-driven discovery at L1. The signal that a customer is genuinely satisfied is not NPS or verbal enthusiasm — it is a cheque. At L1, the goal is three to five paying customers who would be genuinely disappointed if the product disappeared. Wizard-of-Oz delivery (manual fulfilment, Figma mock-ups, founder-as-the-AI) is fine at this stage; efficiency is the last dimension to optimise.
The Four Levels
Level 1: Nascent PMF. Pre-seed or seed stage, fewer than 10 people, $0–500K ARR. Goal: three to five paying customers who have a specific urgent problem and experience genuine satisfaction from the solution. Efficiency metrics are not applicable. A product that satisfies a promise that unlocks revenue for the customer — Vanta’s SOC 2 clearing an enterprise sale — is the archetype. Signs of being stuck: customers like but do not need the product; each customer uses a different feature as the most critical one (indicating consulting, not product).
Level 2: Developing PMF. Seed or Series A, up to 20 people, $500K–5M ARR. Goal: scale from 5 to 25 satisfied customers by building a repeatable demand source beyond warm introductions. The cold-outreach conversion rate at this level is roughly 10% — this is normal, not a crisis. Signs of being stuck: regretted churn above 20%; customers declining because “the timing isn’t right” (which means no).
Level 3: Strong PMF. Three to five years in, 25–100+ customers, $5M–25M ARR. The marginal customer is getting noticeably easier. Inbound accounts for 10–20% of pipeline. Customer success can be systematically delivered, not heroically improvised. NRR consistently above 100%.
Level 4: Extreme PMF. The ultimate goal: widespread demand, satisfaction of a critical need, delivered repeatably and efficiently. At this stage, the marginal customer arrives with low acquisition cost and high retention. This is the state most B2B unicorns take six or more years to reach.
The Four Ps and When to Pivot
Most founders make pivots that are too small. Jack Altman’s retrospective on the OKR-to-performance-management transition at Lattice: “Most founders do a 10% pivot when they need to do a 200% pivot.” The emotional cost of acknowledging that an approach is wrong — especially after persuading investors, employees, and users — makes founders extend losing bets far beyond the expected-value calculation.
Ironclad’s pivot is the gentlest type: change only the Promise. Jason Boehmig was building an “AI legal assistant” in 2015 when no one was searching for one. A single email asking “Are you a CLM?” prompted a positioning reframe. Playing in an existing category (contract lifecycle management) where buyers were already spending money, and asserting superiority within it, opened the demand floodgates.
The four Ps also clarify what to test: Lattice (persona kept), Vanta (all four changed), Plaid (product infrastructure kept, other three changed). The framework prevents pivot thrashing by focusing effort on the specific dimensions where misalignment is detected.