Notes — Jason Cohen on Diagnosing Stalled Growth, the Hidden Multipliers Framework, and the Elephant Curve

Source: raw/lenny/Jason Cohen.txt | Jason Cohen in conversation with Lenny Rachitsky | Lenny’s Podcast

Four questions [Adler frame]

Q1 — What is it about as a whole? An ordered diagnostic for why a SaaS product’s growth has stalled, and what to do about it. Five questions asked in sequence — logo churn, pricing/positioning, net revenue retention, channel saturation, and finally whether you even need to grow — because fixing a downstream problem while an upstream one persists yields no durable gain.

Q2 — How is it argued? By arithmetic and story. Cohen leans on a few hard mechanical results (the churn ceiling, the asymmetry of percentage loss vs gain, the false-positive density of A/B tests) and illustrates the soft parts — pricing, positioning, channels — with concrete cases (the SmartBear randomised exit survey, the Double Down repricing, Constant Contact’s workshops, HubSpot’s agencies). The ordering is the argument’s spine.

Q3 — Is it true, in whole or part? The mechanical claims are sound and quantified. The churn-ceiling formula (max customers = new per period ÷ cancellation rate) is arithmetic, not opinion. The pricing and channel claims are framed as one operator’s pattern-recognition across four companies and ~60 investments — directionally strong, not universal; Cohen repeatedly flags that the right move ‘depends.’ The A/B-testing scepticism is correct in its false-positive logic but he and Lenny agree it inverts at large scale.

Q4 — What of it? The sequence is the deliverable: do not flog marketing while the bucket leaks, and do not optimise pricing while customers churn out the bottom. The single unifying thread he names at the close: create value the customer actually perceives, measure it, then split it via price. And the final, deliberately existential check — do you need to grow? — reframes a company problem as sometimes a personal one.

Glossary

Logo churn — the rate at which customers (not dollars) cancel. Cohen’s preferred guard metric because, unlike revenue churn, it cannot be masked by expansion.
Churn ceiling — the maximum customers a business can ever hold: new customers per period ÷ cancellation rate. At 100 new/month and 5% churn, the ceiling is 2,000, and growth crawls as you approach it.
NRR (net revenue retention) — revenue retained from an existing cohort a year on, including upgrades and downgrades. Median at SaaS IPO ≈ 119%; over 100% is effectively mandatory to scale.
Elephant curve — Cohen’s name for the true shape of a marketing channel: an S-curve ‘trunk’ that then sags into decline (the elephant’s rear) as the audience saturates and competitors crowd in. See Elephant Curve.
Hidden multiplier — a small change with outsized downstream effect (churn 5%→4%; fixing onboarding); the title concept of his book.
Root-er cause — his preferred replacement for ‘root cause’: complex systems have many interlocking causes, so dig past the proximate reason (‘too expensive’) toward the actionable ones, without pretending a single root exists.

Key claims by section

Why growth stalls [§ Opening]

Stalls are usually gradual deceleration, not a sudden stop — ‘running through mud.’ Causes range from the economy to AI uncertainty to simple size (you cannot grow 2× forever). [§ Opening]
The questions are sequenced like a funnel: fix the biggest upstream issue first; tuning a downstream step while an upstream one is broken does nothing. [§ Opening]

Step 1 — Logo churn [§ Are customers leaving?]

Churn is the worst problem: the customer is gone (nothing future to recover), and may leave negative reviews — a double hit. Emotionally damning given ‘the gauntlet’ the customer cleared to buy. [§ Are customers leaving?]
Churn ceiling: max customers = new per period ÷ cancellation rate; cancellations scale automatically with size while marketing does not, so cancellations always eventually overtake marketing. The leaky bucket whose leak grows with the water level. [§ Are customers leaving?]
Exit surveys: multiple-choice is noise — randomising answer order at SmartBear made all options equally chosen. Ask ‘what made you cancel?’ not ‘why did you cancel?’ (Groove: 10%→20% usable responses). ‘Too expensive’ is almost never the real cause — they already bought at that price; dig for root-er causes. [§ Are customers leaving?]
Catch at-risk customers before they cancel (idle, over- or under-using support); compare what good customers share that bad ones do not. Highest-leverage intervention: onboarding — most churn is in the first 30–90 days, and small front-of-funnel shifts compound (his YouTube-retention analogy). [§ Are customers leaving?]

Step 2 — Pricing and positioning [§ Is pricing correct?]

Patrick Campbell (4,200 data points): ‘your prices are way too low because you just guessed and you haven’t changed them.’ Raising price often leaves signups flat or higher. [§ Is pricing correct?]
Pricing selects the market: the textbook demand curve only holds at the very low end; a mid-size buyer reads $2–$100/month as ‘can’t be good enough.’ Demand is a mesa — it rises into the right range, plateaus, then falls. Raising price enters a different, often better market. [§ Is pricing correct?]
Positioning can move price ~8× on the same product: the Double Down example — ‘cut your AdWords cost in half’ (worth ~$5K, capped by savings) versus ‘double your leads for the same ROI’ (worth ~$40K, because you sell growth, which unlocks budget). Sell more of what the buyer already values. [§ Is pricing correct?]
Caution: price is not a standalone knob. Moving up-market drags in new demands (SOC 2, governance, integrations, professional services) and may destroy the very differentiation that made you valuable — or clash with culture (Buffer choosing ‘the little people’). [§ Is pricing correct?]

Step 3 — Net revenue retention [§ NRR]

To counter cancellations you need a force that also scales with size: expansion revenue. Hence NRR. But NRR has a hidden flaw — a 20% loss needs a 25% gain to recover (percentage-loss asymmetry), so NRR flatters the picture; logo churn stays the stricter honesty check. [§ NRR]
Over-100% NRR is effectively mandatory at scale (only ~2 of 100+ public SaaS firms are below; IPO median ≈ 119%). [§ NRR]
The right frame for expansion: measure the value the customer perceives (a number if you are lucky, else proxies/qualitative), grow that, then split it via price — ‘price doubles, but I get 5× the value.’ Land-and-expand has a ceiling: a 10K→100K jump triggers ‘are we getting 10× the value?’ (per Jen Abel). [§ NRR]

Step 4 — Channel saturation and the elephant curve [§ Channels]

Channels have inventory limits (only so many searches; one slot per keyword). They look like S-curves but sag into decline — the elephant curve — as the audience saturates (the seven-exposures rule cuts both ways), the channel itself declines (magazines inflate circulation then fold; conferences likewise), and competitors crowd the discovered alpha. See Elephant Curve. [§ Channels]
Implication: you cannot ‘flog AdWords’ or bolt on one feature to reignite growth. Ask which channels are saturated right now; if you don’t know, the answer is effectively ‘all.’ [§ Channels]
Reignition usually needs a genuinely new channel or product: Constant Contact ran in-person small-business workshops (recruiting power-user agencies); HubSpot sold through agencies (→ ~50% of revenue in 4–5 years); WP Engine sells through WordPress agencies. Expansion rule: ‘plant one foot in a strength, move the other foot.’ [§ Channels]

Step 5 — Do you need to grow? [§ Existential]

The deliberately philosophical close. If every prior question checks out and growth is still flat, ask whether growth is the right goal — some firms (37signals, bootstrappers) rightly optimise profit over revenue. [§ Existential]
‘If you’re not growing, you’re dying’ may apply to you the person more than you the company: stagnation can be a signal to start a new chapter, not to torture the metrics. He withholds a universal answer — but insists it is the right question. [§ Existential]

A/B testing [§ Contrarian Corner]

A/B testing cannot decide strategy or vision, and on details it is mostly false positives: even a 95%-accurate test, hunting a rare true effect, yields more false positives than real wins. Stacked ‘winners’ fail to compound — a year on, conversion is unchanged. Valid only at massive scale with rigorous method; ‘if you don’t know who the patsy is, it’s you.’ Even Shopify finds ~⅓ of wins evaporate on re-check. [§ Contrarian Corner]

The common thread [§ Close]

If one thing fixes most of it: the customer genuinely getting value — promise the right thing, deliver it, onboard them to it, make them realise it, and measure that it rises. Pricing, retention, and expansion then largely follow. [§ Close]