Reading Notes

Jag Duggal on Building Fanatical Customers, the Sean Ellis Score, and Nubank's Product Strategy

Source: Jag Duggal on Building Fanatical Customers, the Sean Ellis Score, and Nubank's Product Strategy

Notes — Jag Duggal on Building Fanatical Customers, the Sean Ellis Score, and Nubank’s Product Strategy

Four questions [Adler frame]

Q1 — What is it about? How Nubank — the Latin American neobank larger than Coinbase, Robinhood, Affirm, SoFi, and Lemonade combined, growing 80–90% by word of mouth — turns ‘we want our customers to love us fanatically’ into a concrete product discipline. Duggal (CPO, ex-Facebook, ex-Google, ex-strategy-consultant) lays out the operating machinery: a hard Sean Ellis PMF gate before scaling, the bullseye-cohort method for iterating to love, a ‘talk to ten customers’ research habit, and a strategy-as-concentrated-bets view drawn from Rumelt and Roger Martin.

Q2 — How is it argued? From Nubank’s operating practice plus Duggal’s earlier career. He gives hard numbers: NPS of 94–95 at Mexico launch; the Payments Assistant moving from hundreds of thousands to 10M+ monthly actives in ~15 months once the bullseye was found; Ultravioleta launched July 2021, scaled only 2–2.5 years later. The strategy argument is explicitly framed against the Valley dogma that ‘strategy is easy, execution is everything’, using Rumelt’s definition and a Kevin Systrom line.

Q3 — Is it true? The PMF-gate discipline is internally consistent and backed by Nubank’s word-of-mouth growth — but the causal claim (the gate produces the growth) is correlational, drawn from one company’s run. [?] The 50% threshold (vs Sean Ellis’s 40%) is justified by ‘Brazilians are culturally more optimistic and polite’ — plausible but asserted, not measured. The ‘fundamentally different, not incrementally better’ mantra is honest about its own limits: Duggal concedes most of the hundred things launched in a year are incrementally better, and that category creation is partly a Bay Area ‘luxury good’.

Q4 — What of it? The Sean Ellis gate, the bullseye-cohort method, and ‘call ten customers yourself’ are all directly transferable to any product team. The deeper transferable discipline is refusing to scale a small problem into a big mess — holding the line against organisational pressure to scale before PMF, which Duggal frames as the core, lonely job of the PM (to say no).

Glossary

Love us fanatically — Nubank’s first cultural value, and the active decision filter when trade-offs arise: will this make customers love us enough to tell their friends? ~80–90% of employees can recite the five values.

Sean Ellis score (Nubank gate) — the share of users answering ‘very disappointed’ if the product went away. Nubank raises the threshold from Sean Ellis’s 40% to 50%, adjusting for Brazilian optimism, and treats it as a near-hard gate: products are rarely scaled until they clear it.

Bullseye cohort — the small sub-segment within a borderline product that already scores very high (≥70%). Reverse-engineer what they share, then expand the bullseye until the whole base resembles it. The opposite of optimising for the average.

Good enough isn’t good enough — is it great enough? — a lead designer’s phrase, now a Nubank review standard. The bar for tentpole products especially.

Fundamentally different, not incrementally better — Duggal’s signature mantra (on the Nubank mugs). Only a fundamentally different experience compels word of mouth; incremental improvement does not. Applies at company and product level.

Leaky bucket — a product with good acquisition but poor retention; growth via a treadmill of new users replacing churned ones. Duggal is ‘maniacal’ that the bucket has no hole, because the word-of-mouth loop only works if acquired customers are retained.

Strategy (Rumelt, via Duggal) — not an ambitious goal, an aspiration, or a set of financial targets, but a coherent plan to apply your strengths in a leveraged way against a specific important problem for a specific customer. ‘Great execution multiplied by a poor strategy is a waste of everyone’s time.’

Concentrated bets — concentration builds wealth; diversification preserves it; a startup has nothing to preserve. Hedging feels smart and safe but produces diffuse execution and unreadable results.

Self-driving / social bank — Nubank’s phase-three vision: a ‘personal banker in everyone’s pocket’ built on a single global code base, AI, and the insight that financial lives are inherently social (‘customers everywhere live harshly unoptimised financial lives’).

Key frameworks

The Sean Ellis gate at Nubank [§ Measuring fanatical love]

The Sean Ellis question — ‘how disappointed would you be if this product went away?’ (not / somewhat / very) — used as an iterate-toward-launch instrument. Nubank’s threshold is 50% ‘very disappointed’ (raised from 40% for cultural optimism). NPS is the post-launch instrument; Nubank skews 70s–90s and treats a one-to-two-point dip as cause for alarm. Churn is watched to guard against the leaky bucket. The discipline this enforces: do not scale a small problem, or you get a big mess — you can keep any product looking good for months on a large base (‘bouncing a dead cat’), then face a far larger untangling.

The bullseye-cohort method [§ Finding the bullseye]

When the overall Sean Ellis score is borderline, do not optimise the average. Find the small cohort already scoring ≥70%, reverse-engineer what they share, and expand it. Worked example — the Payments Assistant: borderline ~40% overall, but a sub-segment with 4+ commitments registered across at least 2 of Brazil’s 4 payment rails scored 70%. The fix: make it easy to onboard multiple bills across multiple rails so automation takes over. Result: from hundreds of thousands to 10M+ monthly actives in ~15 months. Repeated with Ultravioleta (the cohort that hit the spend-based fee waiver loved it).

Talk to ten customers [§ Voice of the customer]

Reject the slow path (a researcher surveys 1,000, cross-tabs, summarises, reports — many degrees of distance). Instead the PM picks up the phone and calls ten customers; by the fifth call you can predict calls six to ten. A phone call captures tone (Brazilians and Mexicans are expressive) that statistics never show. Backed by the Bezos line Lenny offers: when the data and the anecdote disagree, trust the anecdote.

Customer research with a hypothesis [§ Research discipline]

Duggal’s own weakest-area-admitted, with four rules:

  1. The anecdote usually trumps the data — never lose the customer’s actual words.
  2. Be crisp about the hypothesis first; without one, research is just an interesting conversation that neither validates nor invalidates anything.
  3. Do not fall in love with the hypothesis — be its judge, not its lawyer.
  4. When researching, observe more than ask; ask indirect over direct questions; make the customer sell you (play devil’s advocate against your own idea). The Swiffer (P&G) is the canonical observe-the-unspoken-pain example; Nubank’s live case is reinventing the 120–150-year-old joint bank account for shared modern financial lives.

Strategy as concentrated bets [§ Strategy]

Against the Valley dogma that strategy is easy. The legendary companies got strategy and execution right; the ones that got strategy wrong multiplied great execution by zero. Rumelt’s definition (a coherent plan applying strengths against a specific problem, not a goal or aspiration) plus Systrom’s ‘we may not be right, but at least we are clear’ — clarity lets you read whether you are off course. Nubank’s 2014 strategy stated concretely: middle-class urban Brazilian millennials who hated credit-card fees and couldn’t get credit, served a no-fee card enabled by a branchless, digital cost structure. Concentration over hedging; courage to make hard choices (against MBA instinct); guard against the scaled, profitable company turning conservative — ‘a slow path to death’.

Category design — reinvent even if you don’t invent [§ Category design debate]

Duggal takes the contrarian, pro-category-design side (against the ‘go where customers already spend’ view of a prior guest, Todd Jackson), with a synthesis: at company level, the legendary winners are in the new-category business (Google, Netflix, Airbnb, Salesforce); Nubank built the branchless-bank category and made the harder, riskier bet of credit-first (10× harder than a bank account). But even entering an established category (Consignado — secured lending for government employees), the job is to be fundamentally different (direct-to-consumer, undercut pricing, no middlemen). As an insurgent you cannot fight where incumbents stand firm; you must find a disruptive vector. See Category Design.

Connections