BuzbyTimeRead the requirements guide

What monitoring does to people · 7.5

What the research actually shows

What the research actually shows. What decides it, what it costs, and what usually goes wrong. For another implementation reference, see this Monitask guide.

Start with what is weak about itThe honest starting point

Research on employee monitoring specifically is thinner and shakier than the confident claims made in both directions suggest.

Controlled experiments are close to impossible: you cannot randomly assign monitoring to half a workforce and observe for three years. Most of what exists is cross-sectional, self-reported, conducted in one setting, or funded by somebody selling something. For additional context, Smartsheet is a useful reference.

Which does not mean nothing is known. It means precise claims should be treated as marketing.

What is reasonably well supported

Measurement changes behaviour. Robust across domains and not controversial.

Control perceived as reducing autonomy tends to reduce satisfaction and can displace internal motivation. A large and consistent literature, mostly outside workplace monitoring specifically.

Autonomy moderates the effect of workload on strain. One of the more durable findings in job design research.

Transparency changes how monitoring is received. Consistent, and the mechanism is obvious.

What is not established

Whether employee monitoring software increases or decreases net productivity. There is no credible general answer, the effect plausibly differs by kind of work, and both sides of the argument cite studies that do not support the weight placed on them.

Anybody who tells you otherwise, including a supplier of this kind of software, is going beyond the evidence.

How to read a claim in this field

  • Who funded it, and what do they sell?
  • Was anything measured, or was everything self-reported?
  • How many organisations, and were they alike?
  • Was there any comparison group?
  • How long did it run?

Most published claims about this category fail at least two. That is not a reason to dismiss them; it is a reason to treat them as one observation rather than as a finding.

The productivity study problem

Productivity in knowledge work is hard to measure, which is the premise of this entire industry. A study claiming to have measured the productivity effect of monitoring has therefore solved a problem the industry says is unsolved, and the resolution is usually that it measured activity.

Which brings the argument back to the entry on what a record cannot say.

Your own organisation as the evidence

Weak by scientific standards and stronger for your purposes than any published study of somebody else.

Ask before and after. Look at turnover, at recruitment, at what people say when nothing is at stake. It will not be publishable and it is the only evidence about the setting you actually care about.

What we claim

That recording where hours went supports six specific claims, listed elsewhere on this site. Not that it makes anybody more productive.

Not offered

Any productivity claim about this product would be a claim the evidence does not support, which is why none appears anywhere on this site.

Why the evidence stays poor

Nobody funds a large controlled study of employee monitoring. Vendors have no interest in a null result, employers have no interest in publishing, and academic access to workplaces is limited and consented.

Which is likely to remain true, so the practical position is to reason from mechanisms and from your own observation rather than to wait for a settled answer.

Reasoning from mechanism

Where the evidence is thin, the question becomes whether a proposed effect has a plausible route. Measurement changing behaviour has one. Monitoring improving output has a less obvious one, because the mechanism would have to run through either deterrence or information, and it is worth asking a supplier which they are claiming.

Case studies

Useful for understanding what an implementation looks like and worthless as evidence of effect. A case study is a description of one organisation that agreed to be described, selected by somebody with an interest in the outcome.

Read them for the operational detail and ignore the numbers.

What to do first

Take the most confident claim anybody has made to you about this category and apply the five questions. It is a fifteen-minute exercise and it changes how the next conversation goes.

A short summary

The evidence specific to this category is thin and frequently interested. Measurement changing behaviour is well supported; net productivity effects are not. Apply five questions to any claim. Reason from mechanism where evidence is absent. Use your own organisation as the evidence. And treat any productivity claim from a vendor, including this one, as going beyond what is known.

One line to carry

Where the evidence is thin, precise claims are marketing, and that applies to claims you like as much as to claims you do not.

Where to start

Ask your supplier which mechanism their productivity claim runs through. The answer, or its absence, is more informative than any study either of you could cite.

What would change our mind

A large, controlled, independently funded study showing a durable output effect from activity monitoring, with the mechanism identified. If that arrives, this page should be rewritten.

Naming in advance what would change a position is the cheapest available test of whether it is a position or a preference.

A last observation

The absence of good evidence cuts both ways. It does not establish that monitoring is harmful either, and a fair reading is that the mechanisms for harm are better understood than the mechanisms for benefit.

Which is a reason for caution rather than for certainty, and it is the position this site takes.

Also in what monitoring does to people