How Metrics Hide Human Friction: The Danger of "Clean" Data
*Opinions are mine, conclusions are yours. Take what resonates, leave what doesn't. Respectful disagreement always welcome unkindness never is.
We trust numbers because numbers feel neutral. A task completed is a task complete, a form submitted is a form submitted. The dashboard turns green and we move on, because the dashboard says we should. But there is something the dashboard cannot see. It cannot see the employee who finished the report on time but spent the rest of the week recovering from the environment they had to survive to do it. It cannot see the customer who completed checkout after clicking through a maze of unclear steps, frustrated and exhausted but technically converted. It cannot see the student who got the grade after fighting for weeks to receive accommodations they were already legally entitled to.
On paper, the system worked. For the people inside it, the story is a lot more complicated.
The "Ghost" in the Machine
Human friction is the energy spent fighting a system that was not built for you. It is invisible by design, because the cost does not show up in the metric. What shows up is the output; the friction is just the price of getting there, and we rarely ask who is paying it or how much it is actually costing them.
In most organizations, a completed task reads as a success regardless of what it took to complete it. The spreadsheet does not have a column for cognitive tax. It does not track the workaround loop, which is the extra effort a person quietly absorbs when the official process does not actually work the way it is supposed to. It does not measure self-advocacy burnout, that specific exhaustion that comes from spending enormous energy just to access something you were already owed.
Here is where those costs tend to hide most consistently:
The cognitive tax: The neurodivergent employee who met every deadline but had to work through a sensory environment that left them depleted for days afterward.
The workaround loop: The customer who completed the transaction but needed three times the mental effort it should have taken because the interface assumed a kind of user they are not.
The self-advocacy burnout: The student who earned the grade but spent weeks navigating bureaucratic resistance just to receive legally mandated support.
Measuring What Matters: Looking Beyond the KPI
One of the most common mistakes we make is treating the absence of complaints as evidence of ease. If people are moving through a process quickly and not pushing back, we call it efficient. But speed and silence are not the same thing as ease. When a system controls access to something necessary, whether that is healthcare, banking, government services, or a paycheck, people will push through the friction because they have no other option. That is not a satisfied user. That is a person who has decided that fighting the process is not worth the cost of what they lose by stopping.
The loyalty that looks like in a metric is often just the absence of a better alternative. When that alternative appears, the numbers will shift fast, and nothing about the previous dashboard will explain why.
Finding the friction means learning to read the silence between the data points. It means asking different questions and being willing to sit with answers that do not resolve cleanly into a percentage. A few places to start:
Look for shadow systems: If your team is maintaining unofficial spreadsheets or side channels to get work done, your official system has a friction problem they stopped reporting because reporting it stopped helping.
Measure recovery, not just completion: A task that takes ten minutes but costs two hours of cognitive recovery afterward is not efficient. Completion time and energy cost are two different things.
Take your edge cases seriously: The person who struggles most with your system is not an outlier. They are your clearest signal. Fix the friction for them and you have almost certainly fixed it for people you never would have thought to ask.
Data is not the problem. Data is useful. The problem is when we let the cleanliness of a metric convince us that the story is finished when really we have only recorded the surface of it. The most important things happening inside any system are usually the things that do not have a field to live in yet. If we are not asking the questions that surface them, we are not actually managing the system. We are just managing the appearance of one.
A system that hits every target while breaking the people inside it is not a success. It is just a well-documented failure.