← Writing

MBA

"And then what?" — the two words that separate good decisions from clever ones

Most decisions are made on the first-order effect: the obvious, immediate result of the move. The discipline that separates durable operators from clever ones is a single follow-up question, asked at least twice — and then what? The best moves for this quarter and the worst moves for next year are frequently the same move, and only the second-order question tells them apart.

First-level and second-level thinking

Howard Marks built an investing career on the distinction. First-level thinking says: this is a good company, so buy the stock. Second-level thinking says: this is a good company, but everyone already knows that, so it is overpriced, so sell. First-level thinking sees the immediate quality; second-level thinking sees what happens after everyone else has acted on the immediate quality. The two produce opposite instructions from the same fact, and in any competitive arena the second level is where the money and the mistakes both live.

The older statement of the same idea is Frédéric Bastiat's, from 1850: the bad economist sees only what is immediately visible, the good one accounts for what is seen and what is not seen — the delayed, indirect effects that the first move sets in motion. Complex systems have a habit of biting back: as Garrett Hardin put it, you can never do merely one thing. Every intervention propagates.

Where founders pay the second-order tax

Pricing is the cleanest example. Cut the price to close more deals and the first-order effect is exactly what you wanted: more sales this month. The second-order effects arrive later and quietly reverse it — you have trained the market to wait for the discount, anchored the perceived value low, and filled your base with the most price-sensitive, highest-churn customers you could have recruited. The move looked like growth and was actually a loan against next year, taken at a bad rate.

First-level thinking is simplistic and superficial, and just about everyone can do it. Second-level thinking is deep, complex, and convoluted.Howard Marks

The pattern repeats everywhere a payoff is delayed. A growth hack lifts the metric this week and erodes the trust that produced the metric. An incentive rewards the number and quietly corrupts the thing the number was supposed to measure. The discipline is boring and cheap: before committing to the obvious move, ask "and then what?" — and then ask it again of the answer. Two iterations catch most of the self-inflicted wounds.

A worked cascade: the referral discount

The tool is cheap to run, so run it on a move that looks obviously good. Someone proposes a referral offer: refer a friend, both of you get fifty per cent off. The first-order effect is exactly what the growth deck promises — signups climb, acquisition cost per new user looks wonderfully low, the chart points up and to the right, and the meeting is ready to ship it. First-level thinking stops here, because here is where it feels best.

Ask "and then what?" once. The users you acquire this way came for the discount, so a large share of them churn the moment it expires, which means the low acquisition cost bought you a cohort with a collapsed lifetime value — cheap to get, worthless to keep. Meanwhile your existing full-price customers see the offer, do the arithmetic, and conclude they are the suckers paying double; some ask for the deal, some quietly resent it, and the goodwill you spent years building takes a small, invisible hit.

Ask "and then what?" a second time, of that answer. Support load rises to handle the billing confusion and the resentment. Margin erodes on the discounted cohort just as they consume the most support. And the referral discount, once offered, becomes an expected baseline: the next time you try to grow without it, growth stalls, because you have taught the market that this is simply what your product costs. The move that looked like cheap growth was a structural discount you cannot easily remove, sold to you by a metric that only showed the first order.

None of that required a model of the future; it required two iterations of a single question, and it took the length of one meeting. The point is not that referral offers are always wrong — sometimes the arithmetic genuinely works — but that the obvious move and its delayed cost are frequently the same move, and only the second-order question tells you which one you are actually buying.

Where second-order thinking overreaches

Like every good lens, this one fails when it is treated as universal. Three limits.

One: the regress is infinite, and you are not. There is always a third and fourth order, and you cannot compute them all. Past a point, chasing further-order effects is not wisdom but paralysis, and bounded rationality means most decisions should be made quickly at first-order and kept reversible. Reserve the deep analysis for the moves that are expensive to undo.

Two: it can dress up contrarianism as insight. "Everyone thinks X, so the smart second-order move is the opposite" is a seductive template and often just wrong. Sometimes the crowd's first-order answer is simply correct, and the person congratulating themselves on seeing deeper has only talked themselves out of the right call.

Three: your second-order model can be as wrong as their first. Predicting the delayed effect requires a model of the system, and that model can be mistaken. Second-order thinking done overconfidently trades a visible error for an invisible one and feels smarter while doing it. Hold the predicted downstream effect as a hypothesis, not a certainty.

So ask "and then what?" before every move whose bill comes due later — twice, not seven times. Then act, keep the cheap decisions reversible, and stay humble about the fact that your second order is also a guess.

Sources

  1. primaryHoward Marks, The Most Important Thing (2011) — second-level thinking.
  2. primaryFrédéric Bastiat, "That Which Is Seen and That Which Is Not Seen" (1850).
  3. secondaryRobert K. Merton, "The Unanticipated Consequences of Purposive Social Action" (1936).
  4. secondaryGarrett Hardin, on ecological interconnection — "you can never do merely one thing."