You have a team that has been underperforming for a while. Not dramatically. Just consistently below what you know they can do.
You want the best out of them. Which lever do you actually pull?
Most of us reach for intrinsic motivation, and that is the right answer. Get people to care about the work itself and the rest follows. No chasing, no policing, no theatre.
I believe that. I also think we are too casual about how hard it is.
It depends on three things lining up, and you control only one.
It needs you to have the credibility and the patience.
It needs the person to be willing, because you cannot install meaning into someone who has decided this is only a job. And it needs the organisation to reward the change when it arrives.
If someone genuinely transforms and the system hands them the same increment as everyone else, you have taught them a lesson you did not intend.
When all three align, it works beautifully. It also takes quarters. Sometimes a year.
So what do you have in the meantime?
Start with what sits underneath
Before any of this, there is the base. Pay, job security, decent conditions, being treated with respect.
Get that wrong and nothing above it functions. An underpaid, insecure person does not need a motivational lever. They need the base fixed.
But once the base is fair, notice how poor it is as a day to day lever.
Salary moves once a year. The increment arrives months after the work, shaped mostly by budget and by comparison with peers. Nobody can trace it back to any particular Tuesday.
It rewards. It does not direct.
Job security is worse as a motivator, because it operates as a threat. Fear of losing the job produces the minimum safe amount of work. It never produces initiative.
So we have a floor that cannot direct, and a ceiling that takes a year to build. The interesting question is what sits in between.
Two problems that share one label
Underperformance gets used for two situations that have almost nothing in common. The difference is not the value of the work. It is what the work tells the person about how they are doing.
The first kind is work whose outcome is invisible. Nobody notices it until it fails.
The floor gets cleaned and nobody comments.
The patches get applied and nothing happens, which is the point.
The test suite catches a defect that never reaches a customer.
The compliance filing goes out correctly and no regulator calls.
The backup runs every night for four years and is never restored.
This is not a category of low skilled work. It is prevention work, and it runs from housekeeping to the most demanding engineering in the building. What these roles share is that success looks exactly like nothing happening.
The second kind has the opposite property. The outcome is highly exposed.
The mega project, with senior stakeholders, a visible date and a name attached to the result.
People underperform in both, for opposite reasons.
Invisible work drifts, because effort and slack look identical from outside, and after a while they start to feel identical from the inside too. Exposed work freezes, because the cost of being the person who broke it is too large to risk.
One is a drift problem. The other is a fear problem. A leader who treats them the same will fail at both.
This article is about the first.
Lets start with the evidence is sitting in your phone
Last month my wife booked a temporary house help through an app.
The woman who arrived was the same one we had let go a year ago for poor work.
My wife was polite about it and completely sceptical. For the next hour, she worked better than anyone we have had in our home.
Nothing about her had changed. Same person, same work, roughly the same pay. What changed was that this time, the work was being scored.
Once you notice it, it repeats.
The auto driver you argued with about the meter for fifteen years. Put him on Ola or Uber and the fare is settled before the trip, the route is not a negotiation, and the car is cleaner.
The electrician who used to arrive whenever and quote whatever. Book the same work on Urban Company and he comes in a slot, quotes upfront, and closes it in one visit.
No app teaches anyone to wire a switchboard or read a gap in traffic. The skill was always there. What the platforms built was a measurement system. The service was the wrapper.
And notice what they never did. Nobody gave the driver a sense of purpose. They took work whose quality used to be invisible and made it visible at the end of every job.
Two loops, not one
Here is where most people take the wrong lesson, and where I nearly did.
The platforms built two loops at the same time, and we tend to see them as one thing.
The first is a feedback loop. Small, fast and clear. The standard is stated in advance. The verdict arrives at the end of the job. Three or four things a person can hold in their head while working.
The second is a uncertainity loop. That score decides how much work comes next week, and there is no salary underneath it. Do badly for a fortnight and your income disappears.
The mechanism that improves the work is the first loop. The second is what makes the arrangement harsh, and it is why the gig model is honest evidence and a poor blueprint. Most of that pressure lands on the person with the least security in the deal.
The useful realisation is that these two do not have to travel together.
A company already pays a salary. The ground is already there. That means a leader can take the feedback loop and leave the unertainity loop behind.
The score can carry information without carrying survival risk.
That is not a weaker version of what the platforms did. It is arguably the better version, because clarity works on its own. The driver does not drive well because he is afraid. He drives well because for the first time he knows exactly what counts.
Which raises the obvious question. If this works for a driver, does it work for an architect?
Does this work for knowledge work
It should. It usually does not, because of what we reach for instead.
Our standard answer to invisible work is recognition. The quarterly award. The spot bonus. The shout out on the town hall call.
Hold that against the mechanism and it fails on every count.
The loop is ninety days long. By the time the award arrives, the behaviour it is meant to encourage happened three months ago and has already been repeated or abandoned.
It is scarce and comparative. One person wins. Twenty people learn nothing about where they stand, except that they did not win.
The selection is opaque. Ask anyone to describe precisely what would earn them the next award and you will get a guess. A signal that cannot be predicted cannot direct behaviour. It rewards. It does not teach.
And worst of all for this argument, awards are biased toward visible work by design. We give it to the person who fixed the outage at two in the morning, not to the person whose work meant there was no outage. We celebrate the rescue and overlook the prevention, then wonder why nobody wants the quiet job.
Awards make invisible work more invisible.
What works instead is duller and much more effective. Find the leading indicator of prevention and put it where the team can see it every week.
You cannot count the outages that did not happen.
You can count whether the restore drill was actually run.
You can count how long a critical patch sits unapplied.
You can count defects caught before release against defects that escaped.
You can count how long audit findings stay open.
None of these are glamorous. All of them turn quiet work into something a person can point at.
And in a salaried setting the consequence attached to the number should be attention, not income. That is the whole benefit of having a base. The number directs the work. It does not decide whether the person eats.
The part I will not oversell
Any measure with consequences attached gets optimised eventually, including in ways you did not intend. Whatever you count, expect people to become very good at that exact thing and nothing beside it.
And clarity buys compliance quickly and commitment slowly. That maid worked hard in our home, but she was protecting next week's income. A number gets you a floor. It never gets you a ceiling.
Where this leaves a leader
Fix the base first. Nothing here substitutes for fair pay and stable ground.
Then, for work nobody notices until it fails, do not start with purpose.
Start with visibility.
Pick two or three things a person can repeat without opening a document. Make the signal weekly. Take the feedback loop and leave the uncertainity loop where you found it.
Then keep building the intrinsic side underneath, because the number tells someone how not to fail. It does not tell them why it matters that they did not.
That part has not been automated yet, and it is still the reason the team needs a leader at all.
The other kind of underperformance, where the work is exposed and fear is doing the damage, needs a completely different answer.
That one is for the next piece.