What is hysteresis at work?

Hysteresis at work is the property that a team's current state depends on its entire history of decisions, not just the inputs acting on it right now. Borrowed from physics—where "hysteresis is the dependence of the state of a system on its history"—it explains why reversing a policy doesn't reverse its effects. The path a team takes going into a change differs from the path coming back out, so returning the inputs to their original values leaves the team somewhere new. In short: some workplace decisions don't undo.

Why doesn't reversing a bad decision fix it?

Because by the time you reverse it, the decision has already changed the system you're trying to restore. A return-to-office mandate that drives out senior staff can be lifted, but the people are gone, hiring stays slower, and trust sits at its residual level. The reversal removes the cause, not the effect—the same way removing a magnetic field leaves a paperclip partly magnetized. Undo is not the inverse of do; it's a second, separate action that traces a lower curve.

Why is rebuilding trust harder than breaking it?

Because the hysteresis loop is asymmetric. In a magnet, simply removing the field doesn't erase the residual alignment—you have to drive a reverse field (the "coercive force") harder than the original to get back to zero. Trust works the same way: one distrust-signaling decision spends it in an afternoon, but restoring it requires sustained, opposite-direction effort over months. The break and the rebuild are different sizes, and the rebuild is far larger. That asymmetry is structural, not a matter of effort or attitude.

Is organizational hysteresis always a bad thing?

No—and that's the key insight. Hysteresis is a property, not a flaw. A thermostat deliberately uses a hysteresis "deadband" to avoid rapidly cycling on and off, and your culture uses the same effect: trust, good habits, and strong norms persist after the effort that created them fades, carrying a team through rough patches. You want some stickiness. The skill is choosing which loops to enter—building durable good state while refusing the cycles that burn it.

How do you manage irreversible decisions at work?

Sort decisions by their loop, not their budget: ask whether reversing this one would let the team snap back or leave it dented, and give the dented ones real scrutiny regardless of size. Treat trust as capital that's repaid at a premium, pre-mortem the moves that quietly spend it, and keep genuinely reversible experiments truly reversible with real off-ramps. Above all, stop assuming symmetry—identify which levers are springs and which are magnets before you pull them.

Hysteresis at Work: Why Trust Doesn't Bounce Back

Stroke a paperclip with a magnet, then take the magnet away. The paperclip is now a weak magnet itself—it stayed changed after the cause was removed. Physicists have a clean line for this: "Even when the field is removed, part of the alignment will be retained: the material has become magnetized." The cause is gone. The effect isn't.

Most leaders run their teams as if the opposite were true—as if you can apply a policy, dislike the result, lift the policy, and snap back to where you started. You can't, and there's a name for why. Hysteresis at work is the property that a team's state depends on its history, not just its current inputs. Reverse the decision and you do not reverse its effects, because the path down isn't the path back up. The bell does not un-ring; it just gets quieter.

The textbook definition is almost eerie in a management context: "Hysteresis is the dependence of the state of a system on its history." Your team in June is not a function of your June decisions. It's a function of every decision you've cycled it through. This is a deep-dive into that mechanism—what hysteresis actually is, why "just undo it" is a fantasy, why the rebuild always costs more than the break, and which loops are worth entering anyway.

What hysteresis at work actually is

Here's the part nearly everyone gets wrong. People hear "hysteresis" and translate it to "lag"—the effect shows up a little late, then catches up. That's not it. Lag is a delay on the same curve. Hysteresis is a different curve entirely depending on which direction you're moving.

Plot it and you get the giveaway shape: a loop. The same input value maps to "different values of one variable depending on the direction of change of another variable." Push a system up to a point and bring it back down, and it traces one arc going up and a lower arc coming back. The clearest statement of the whole phenomenon is this: "If an input variable X(t) cycles from X₀ to X₁ and back again, the output Y(t) may be Y₀ initially but a different value Y₂ upon return." You set the dial back to zero. The needle does not return to zero.

A quick honesty note before we map this onto people: a team is not a slab of iron, and this is a model, not a law of physics. Humans have memory, grudges, and second thoughts that no metal does. But the shape of the thing—state that depends on history, a path up that differs from the path down—describes path dependence in organizations almost too well. Some of your levers are springy and snap back when you let go. Others are magnetic: pull them once and the team stays partly pulled. The expensive mistake is treating a magnetic lever like a springy one.

Why reversing a decision doesn't reverse its effects

The cleanest case study in organizational hysteresis is the return-to-office whiplash of the last three years. A company mandates five days in. Attrition spikes. Leadership softens the rule, or drops it. And the obvious expectation—morale returns to its pre-mandate level—simply doesn't happen. The input cycled back. The output landed somewhere else.

The data shows why. A University of Pittsburgh study published in December 2024, tracking more than three million tech and finance workers' histories across S&P 500 firms, found that after RTO mandates, "firms losing their best talent... brain drain as a significant cost of RTO mandates, even for the largest firms in the world." Time-to-fill vacancies rose 23%. The hire rate fell 17%. And the turnover was concentrated among women, senior staff, and the most skilled workers—the people hardest to replace.

Now reverse the policy. Does the 23% slower hiring snap back? Do the senior engineers who left return? They do not. This is the heart of return-to-office attrition as a hysteretic event: the mandate didn't just lower a dial you can raise again, it permanently removed mass from the system. As MIT Sloan Management Review put it, these mandates "damage employee engagement and increase attrition, especially among high-performing employees." The high performers are precisely the ones with options, which means they're the first out the door and the last you can lure back. Why reversing a decision doesn't work, in one sentence: by the time you reverse it, the system you're reversing it for is already a different system.

This is the same family of trap as Brooks's law—where adding people to a late project makes it later, and removing them again doesn't recover the lost time. The action and its undo are not mirror images. They're two separate dents.

The area inside the loop is paid in full

Hysteresis isn't free motion. Every time you push a system around the loop and back, you lose energy—permanently, as heat. In the elastic hysteresis of a stretched rubber band, "the area in the centre of a hysteresis loop is the energy dissipated due to material internal friction." The band doesn't give back all the energy you put in; the difference warms it up and is gone. A magnet's loop works the same way—it's why transformer cores get hot. The bigger the loop you trace, the more you burn.

Organizationally, the area inside the loop is the trust, the institutional memory, and the goodwill that the round trip consumed and never returned. You can mandate, reverse, and end up nominally "back where you started" on paper—same headcount target, same policy—while having spent a year's worth of credibility heating the air. The ledger looks balanced. The thermometer says otherwise.

And here's the asymmetry that makes rebuilding trust at work so brutal: the loop isn't symmetric around the center. In a magnet, you can't just remove the field to erase what you did—you have to drive a reverse field—the coercive force, a stronger push in the opposite direction—hard enough to undo the residual alignment. Demagnetizing costs more than magnetizing left behind. Trust behaves the same way. It is cheap to spend and expensive to coin. One surveillance-flavored mandate can erode trust in an afternoon; restoring it takes months of consistent, boring follow-through pushing in the opposite direction. The break and the rebuild are not the same size, and the rebuild is the big one. This is why rebuilding trust is harder than breaking it is not a motivational cliché—it's the geometry of the loop.

Remanence: why morale doesn't recover on schedule

Magnetism has a precise word for the part that stays behind after the field is gone: remanence, the residual magnetization. Teams have remanence too, and it's why why morale doesn't recover is usually the wrong question—it assumes morale is sitting at zero waiting for a nudge, when really it's sitting at some leftover value set by the last cycle.

The practical consequence is that your next decision never lands on a blank team. It lands on a pre-magnetized one. The second reorg in eighteen months doesn't get evaluated on its merits—it gets evaluated through the residue of the first. A "we're listening now" initiative hits a workforce still carrying the polarity of the last three times they were told that. People who have been cycled before brace for the loop. They disengage preemptively, which is rational: if the policy will reverse anyway, why pay the switching cost twice?

This is also why recovery has a floor on how fast it can go, no matter how good the new plan is. You can announce a reversal on Monday; the residual state updates on the team's clock, not yours. Pushing for instant rebound just adds another cycle to the loop—the organizational cousin of skipping recovery and going straight back to crunch, which never restores capacity the way a real safety stop after crunch does. Remanence is the reason morale curves look like staircases, not switches.

Managing the one-way doors

If some of your levers are hysteretic, the job isn't to avoid them—it's to know which ones they are before you pull. The failure mode is treating every decision as a springy, reversible A/B test when a few of them are magnets. Here's what changes once you take irreversible decisions at work seriously.

Sort levers by their loop, not their size

The instinct is to gate big decisions and rubber-stamp small ones. Wrong axis. A small, reversible pricing tweak deserves less ceremony than a "minor" trust-spending move like adding monitoring software—which is tiny on the budget and enormous on the loop. Before a decision, ask one question: if this goes badly and I reverse it, does the team snap back—or stay dented? The dented ones get the scrutiny, regardless of headline size.

Spend trust like capital, not like air

Trust feels free because spending it has no line item. But it sits inside the loop, which means every withdrawal is paid back at the coercive rate—more than you took. Decisions that quietly spend it (surveillance, mandates justified by distrust, reversals that contradict last quarter's promise) are the ones to pre-mortem hardest. The MIT Sloan finding that monitoring is "the weakest form of management" is really a statement about loop area: it buys a little control by burning a lot of remanent goodwill. This is the same insulation problem behind moral hazard at work—the person pulling the lever rarely feels the heat the loop generates.

Keep the reversible stuff reversible

The antidote to hysteresis isn't paralysis—it's making more of your decisions genuinely springy. A real experiment with a real off-ramp, scoped small, can snap back. The danger is the fake reversible: the "let's just try it for a quarter" that quietly relocates people, rewires incentives, or breaks a promise, and so can't actually be un-tried. Honesty about which is which is most of the skill.

This is also where tooling either helps or hurts. Every time context, decisions, and history scatter across a dozen disconnected apps, a little institutional memory demagnetizes—and you pay the cost of context-switching to re-magnetize it. Keeping the work, the conversation, and the record in one place is partly a hysteresis hedge: the team's shared state survives the next personnel or tooling cycle instead of resetting. It's part of why we built Coommit so the video call, the canvas, and the AI that remembers all of it live on one surface—less residue lost per cycle.

When hysteresis at work is your friend

Now the steelman, because hysteresis at work is not a villain—it's a property, and half the time you want it. A thermostat uses hysteresis on purpose: a deadband so the furnace doesn't frantically click on and off every time the temperature wobbles by a tenth of a degree. The lag is the feature. It buys stability.

Your culture is the same. Remanence cuts both ways: the trust, norms, and habits you build also persist after the effort that built them fades. A team that's been treated well stays loyal through a rough quarter; good defaults survive a bad week; a strong onboarding magnetizes people in a direction that holds. You want a culture that doesn't reset to zero every time conditions wobble. The goal was never to eliminate hysteresis—it's to choose your loops deliberately. Enter the ones that build durable good state. Refuse the ones that burn it. The leaders who get this stop asking "can we reverse this?" and start asking "which direction do we want the residue pointing?"

The dial you can't turn back

The seductive lie of every dashboard is that the dials are symmetric—that whatever you turned up, you can turn down, and the readout will retrace its steps. For the springy levers, fine. But the human layer of your organization is magnetic, and on a magnet the only way back from the residual is to overdrive a reverse field harder than the one that magnetized it. There is no setting that says "as you were."

The AI era will make this trap worse before it makes it better, because so much of the new tooling is genuinely reversible—turn the feature off, roll back the deploy—which trains leaders to expect that everything reverses. The model, the workflow, the integration: sure. The trust you spent forcing them on people: no. As more of your stack becomes a clean undo button, the few hysteretic decisions get easier to mistake for reversible ones, and more expensive when you're wrong.

So before the next big lever, find the loop. Ask whether this is a spring or a magnet—whether the undo is real or just a story you'll tell. That is the one question hysteresis at work puts to every leader. Spend the irreversible stuff—trust, attention, the patience of your best people—like the non-renewable resource it is. The team you have is the sum of every decision you've ever cycled it through. The only move you fully control is the next one.