PM Gym

Systems Thinking

See the whole system, not just the event in front of you

Products never live in isolation. They sit inside a web of users, teams, incentives, and metrics that push back on every change. Systems thinking is the habit of seeing those connections — so you fix root causes instead of chasing symptoms.

Step-by-step lessons

Think in Systems

Four short lessons: look beneath events, spot the loops, respect delays, and find the leverage. No jargon — every idea comes with a plain-language example.

1

Events, Patterns & Structure

What is a "system"? A system is just a set of parts that affect each other — users, teams, features, metrics, incentives. Change one part and the others react. Your product is a system, not a to-do list.

When something goes wrong, our instinct is to react to the single event we can see and move on. Systems thinking asks you to look below the waterline — because the same events keep coming back for reasons hidden underneath.

Everyday example

Think of a doctor. A patient keeps getting headaches. A weak doctor hands out a painkiller each visit (the event). A good doctor asks why they keep coming back — bad posture? Skipping meals? — and fixes that (the structure). One treats the symptom forever; the other ends it. PMs face the same choice every week.

The iceberg model — four levels to look at

1. Events — what just happened. "Checkout crashed this morning." This is the tip of the iceberg, the only part you see easily.

2. Patterns — the same event over time. "Checkout crashes every Monday at peak hours." Spotting the pattern is the first sign there's something deeper.

3. Structure — the setup that produces the pattern. "We never sized our servers for the Monday traffic spike." This is where lasting fixes live.

4. Mental models — the beliefs that built the structure. "We assumed traffic is roughly flat all week." Change the belief and you stop building broken structures.

React at the event level and you firefight forever. Change the structure and the events simply stop happening.

Quick check

Users keep filing the same complaint: they can't find the export button. The team keeps replying to each ticket individually. What's the systems-thinking move?

2

Reinforcing & Balancing Loops

Systems are driven by feedback loops — where an effect circles back to influence its own cause. In plain terms: the result of something becomes the cause of more of it. There are only two kinds, and telling them apart explains most product dynamics.

Everyday example

Reinforcing is a snowball rolling downhill — it grabs more snow, gets bigger, so it grabs even more. It speeds up on its own. Balancing is the thermostat in your home — when the room gets too hot it turns off the heat, when it gets too cold it turns it back on, always pulling back to one target. One accelerates; the other steadies.

Snowball

Reinforcing loop

Growth feeds more growth (or decline feeds more decline). More users → more content → more users. Great for virality, dangerous for death spirals.

Thermostat

Balancing loop

The system pushes back toward a target. Rising load triggers slowdowns that reduce load. It resists change and creates stability — or stubborn plateaus.

Quick check

On a social app, users who post get likes, which encourages them to post more, which attracts more users who post. What kind of loop is this?

3

Stocks, Flows & Delays

Cause and effect are often separated in time. Understanding stocks, flows, and especially delays stops you from overcorrecting — from slamming the brakes before your last action has even landed.

Everyday example

Picture a bathtub. The water in it is the stock (say, your active users). The tap pouring in and the drain leaking out are the flows (signups and churn). The level only rises if the tap runs faster than the drain. And a delay? That's a tap with a long pipe — you turn the handle now, but the water shows up seconds later. Turn it more because "nothing's happening" and you'll suddenly flood the room.

Stock — what accumulates

A level that builds up: active users, technical debt, trust, cash. It changes slowly.

Flow — what changes the stock

The rates in and out: signups vs. churn, debt added vs. paid down.

Delay — the lag between action and result

Effects arrive late. A price change may not hit churn for months — so you can't judge it next week.

Quick check

You raise prices and churn doesn't move for six weeks — then it jumps. What trap does this illustrate?

4

Leverage Points & Side Effects

In any system, a few well-chosen changes move everything, while most effort barely registers. Those high-impact spots are called leverage points. And every change ripples — the second-order effects (what happens next, after the obvious result) often matter more than the first.

Everyday example

Weak leverage: nagging people to "use less water." Strong leverage: putting a water meter on every home so the bill reflects use — behavior changes on its own. Same effort, wildly different result. The strongest leverage point of all is usually the goal or the rule you set, not the day-to-day knobs you fiddle with. As a PM, changing what a team is rewarded for beats a hundred pep talks.

When a metric fights back

A support team is measured on tickets closed per hour. Closures soar — because agents now close tickets fast without truly solving them. Customers re-open, re-contact, and overall load rises.

Optimizing one part in isolation (local optimization) can hurt the whole. Always ask: "and then what happens?"

Quick check

A riddle: "I'm the consequence you didn't plan for — the ripple that shows up a step later, often undoing the win you just celebrated." What am I?

Review the concepts

Systems Thinking Flashcards

6 cards covering the essentials. Click a card to flip it.

Foundation Card 1 of 6

Systems Thinking

Click to flip

Tip: say the answer out loud before flipping.

Explanation

In practice

1 / 6
Apply what you learned

Practice Scenarios

15 real situations where systems thinking beats firefighting. Read each one, decide the strongest move, then check the reasoning. Wrong answers explain why — that's where the learning is.

Scenario 1

Your team is rewarded purely on the number of features shipped per quarter. Shipping is up, but user satisfaction is sliding.

What's the systems-thinking read?

Scenario 2

A referral feature quietly took off: invited friends invite more friends, and signups are compounding week over week.

How should you treat this?

Scenario 3

Two teams each optimize their own metric: the growth team maximizes signups, the infra team minimizes server cost by capping capacity.

What does systems thinking predict?

Scenario 4

Your onboarding funnel drops 20% one week. You add a bigger "Continue" button on the step where people leave. Drop-off barely moves.

What did the button fix miss?

Scenario 5

A marketplace: more buyers attract more sellers, which widens selection, which attracts more buyers. Your CEO asks where one extra engineer would create the most long-term value.

Where does systems thinking point them?

Scenario 6

To hit an aggressive quarterly signup goal, the team leans hard on deep discounts. Signups spike every quarter — but each quarter needs a bigger discount to hit the same number, and margins are thinning.

Which systems pattern is this, and what's the risk?

Scenario 7

You ship infinite scroll. Average session time jumps 30% and everyone celebrates. Three months later, weekly active users are quietly falling.

What likely happened?

Scenario 8

You set a team target: "cut average support-ticket resolution time by 40%." Resolution time drops beautifully. But repeat-contact rate and refunds both climb.

What went wrong with the metric?

Scenario 9

The app slows down under load. Each time, ops adds more servers and it recovers — for a while. But it keeps happening, the bill keeps rising, and nobody has looked at the code in months.

Which archetype is this, and what's the move?

Scenario 10

Your referral loop drove explosive growth for a year. Lately, invites are flat no matter how much you optimize the invite flow. The core audience is niche and mostly on the product already.

What does systems thinking say is happening?

Scenario 11

A shared internal platform team supports every product squad. Each squad, acting sensibly, files "just one urgent request." The platform team is now permanently overloaded and slow for everyone.

Which archetype, and how do you fix it?

Scenario 12

Trust in your brand took three years of consistent quality to build. One botched data-breach response burned a big chunk of it in a week. Signups are now sluggish despite unchanged marketing spend.

What stock-and-flow idea explains this?

Scenario 13

A competitor adds a flashy feature. You rush to match it. They one-up you. You one-up back. Six months later, both products are bloated, neither of you gained share, and both teams are exhausted.

What trap are both companies caught in?

Scenario 14

A leader says: "Our activation problem is simple — the sign-up form is too long. Cut the fields and activation will jump."

What's the systems-thinking caution here?

Scenario 15

You're deciding how to lift retention. Option A: tweak the push-notification copy. Option B: change the core onboarding so new users reach their first real "win" in minute one instead of day three.

Which is the higher-leverage point, and why?

Lock it in

Guess the Term

Read the clues and name the concept. The fewer clues you need, the more points you score.

Round 1 Score 0
Keep it handy

Systems Thinking Quick Reference

The whole topic on one screen.

The Iceberg

Most causes hide below the visible event. Go deeper to fix things for good.

1

Events

What just happened. React here and you firefight forever.

2

Patterns

The trend over time. "This keeps happening every Monday."

3

Structure

What produces the pattern. Fix here for lasting change.

4

Mental models

The beliefs that built the structure.

Loops & Leverage

Reinforcing loop

Amplifies itself — growth feeds growth. Protect and fuel it.

Balancing loop

Pushes back to a target — creates stability and plateaus.

Mind the delay

Effects lag actions. Don't judge (or overcorrect) too early.

Ask "and then what?"

Trace second-order effects before you act.

Patterns That Repeat (Archetypes)

The same traps show up in every product. Name one and you're half-way to escaping it.

Fixes that fail

A quick patch relieves the symptom but the root cause grows. (More servers vs. the real bug.)

Shifting the burden

Leaning on a crutch (discounts) starves the real capability (a product worth full price).

Limits to growth

A booming reinforcing loop hits a constraint. The lever becomes the constraint, not the loop.

Tragedy of the commons

A shared resource with no rule gets overused. Add prioritization or a cost signal.

Escalation

Two rivals one-up each other into exhaustion. Re-anchor on your own strategy.

Goodhart's Law

When a measure becomes a target, it gets gamed and stops measuring what you meant.

Ask Before You Act

Six questions that turn firefighting into systems thinking.

1

Is this an event or a pattern?

If it keeps happening, don't fix the instance — fix the structure.

2

And then what happens?

Trace the second- and third-order effects before shipping.

3

Where's the delay?

Have I given the last change time to land before judging or acting again?

4

What loop am I in?

Reinforcing (feed it) or balancing (find what it's defending)?

5

Am I optimizing a part or the whole?

Could this local win hurt the overall system?

6

What's the real leverage point?

Usually the goal, rule, or incentive — not the surface knob.

Notification