Second-order thinking
And then what?
- And then what?
- Operating: Many SaaS companies that aggressively cut their free tier in 2023-24 saw immediate revenue lift, then watched their top-of-funnel collapse over the next 18 months as developer adoption dried up.
- Investing: Howard Marks's memo 'Second-Level Thinking' is the best short read on this.
- Everyday life: The decision to send a sharp Slack message at 11pm is almost always a first-order decision (you're annoyed, you want it off your chest).
Howard Marks talks about this a lot. First-order thinking stops at the immediate effect of a decision. Second-order thinking carries it forward: "if I do X, what happens next, and then what happens after that?"
The textbook example. Government raises minimum wage by 20%. First-order thinking: workers earning minimum wage are now better off. Second-order: employers respond by cutting hours or hiring fewer people. Third-order: the workers who keep their jobs are better off; the ones who lose hours or jobs are worse off. None of these orders is automatically right. The honest answer requires thinking past the first move.
The same thing happens in business decisions all the time. Three patterns I see often:
Price increases. First-order: revenue goes up. Second-order: some customers churn. Third-order: the customers who stay are higher-quality, so customer service costs drop, but acquisition costs go up because the price is now less attractive. The full picture depends on which effect is bigger.
Layoffs. First-order: cost base drops, margins improve. Second-order: remaining staff get overloaded, productivity per person drops. Third-order: the best people leave because they have options. Fourth-order: hiring costs spike when you eventually need to rebuild. The 30% headcount cut sometimes ends up costing more than it saved.
Subsidising acquisition. First-order: customer count goes up. Second-order: the customers acquired via subsidy churn faster than organic ones, because they were optimising for the discount. Third-order: cohort LTV drops, but the cohort still looks great on the dashboard because the timing of churn is delayed.
Most bad strategies look fine on the first order and obviously stupid on the third. The discipline of asking "and then what?" twice in a row catches a lot of mistakes before they're committed to.
The hardest part of second-order thinking is that it slows you down. First-order answers feel decisive. Second-order answers feel uncertain. (They are uncertain.) The compromise I've landed on: don't second-order-think every decision. Reserve it for the big or irreversible ones. The small reversible ones can survive a first-order miss.
Examples in the wild
Many SaaS companies that aggressively cut their free tier in 2023-24 saw immediate revenue lift, then watched their top-of-funnel collapse over the next 18 months as developer adoption dried up. The free tier was acquisition; the revenue lift wasn't worth the funnel damage.
Howard Marks's memo 'Second-Level Thinking' is the best short read on this. His point: if you're investing on the same logic everyone else is, you'll get the same returns as everyone else, which means after fees you'll underperform. Second-order thinking is what produces non-consensus calls.
The decision to send a sharp Slack message at 11pm is almost always a first-order decision (you're annoyed, you want it off your chest). The second-order effects (the recipient reads it tomorrow morning, sets the tone for the day, talks about you at lunch) are usually worse than just sleeping on it.
Second-order thinking is one of the mental models we apply through real cases inside the Pareto MBA — a part-time program for professionals who want to think clearly about business.