Kinds
What makes a claim empirical
Not whether it sounds scientific, and not whether anybody has measured it. One test, and the three ways a claim fails it.
Claims this part is about, and what each was answered with
Teams that sit together ship faster
Empirical→A studyWork should be something you go to
Value≠A studyA statement about what work is, answered with a productivity paper
The lease runs to 2031 and the rent review falls next year
Interest→A statement about who paysSaid once, in a finance update
three claims · one empirical, one value and one interest · one was answered with the wrong kind of answer
One claim of each kind, from the twenty-four counted here.
The useful definition is narrower than the everyday one and it has nothing to do with how the claim is phrased.
A claim is empirical if there is some observation that would make it false.
That is the whole test. Not whether it is about numbers, not whether anybody has done the study, not whether it sounds rigorous. Only whether the world could, in principle, contradict it.
Three ways a claim fails the test
It has no falsifying observation at all. "Culture is built in person." What would show this to be false? Pressed, people usually answer with something about what culture is — which means the claim was a definition, not a prediction, and belongs in the second kind.
It is true by construction. "Something is lost when people are not together." Almost certainly, and also unfalsifiable as stated: any observation is consistent with something having been lost. The claim becomes empirical only when the something is named.
The falsifying observation is impossible in practice. "We would have grown faster with everyone in the building." This has a clear falsifier — run the same company again the other way — and no access to it. This one is worth its own category in your head: empirical in form, unavailable in fact, and it is the shape most likely to be argued about for years.
The test applied
| Claim | Falsifying observation | Empirical? |
|---|---|---|
| New starters take longer to become productive remotely | Time to first independent delivery, by cohort | Yes |
| Remote workers put in longer hours | Logged hours by working pattern | Yes |
| Collaboration tools have replaced what the office did | An agreed list of what the office did, checked off | Yes, once the list exists |
| Innovation happens in unplanned conversations | Ideas traced to their origin | In form; nothing records this |
| We would have grown faster with everyone in | Rerun the company | In form; impossible |
| Work should be something you go to | — | No |
Rows three and four are the interesting ones. Both are routinely treated as settled facts by the people who make them, and both become testable only after somebody does a piece of work — writing the list, building the record — that nobody has an incentive to do.
Why people over-claim the empirical
Because in most workplaces an empirical claim outranks a value claim socially, even though nothing justifies that ordering.
"I think a company should be a place" sounds like a preference. "Teams in offices are more productive" sounds like a finding. The second carries more weight in a meeting, so a person who holds the first will often reach for the second — not dishonestly, but because the stronger claim is near to hand and feels like a better version of what they mean.
The result is a discussion full of empirical-sounding claims that nobody has evidence for and nobody would abandon if the evidence went the other way. That last part is the tell, and it is a fair question to ask oneself: if the study came back against me, would I change my mind? If the answer is no, the claim was not doing empirical work.
Why this matters more than it sounds
Because an empirical claim comes with an obligation attached. If you say the world is a certain way, you have accepted that the world can answer back, and you have to say in advance what would count.
On the record counted here, nine claims were empirical, and for four of them no evidence exists anywhere, either way. Nobody had noticed, because an empirical claim nobody can check looks exactly like one nobody has got round to checking yet.
The question to ask of your own claims
One sentence, asked privately before saying something:
If this were measured and it came out against me, would I change my position?
A yes means the claim was doing empirical work and you should say what the measurement is. A no means you hold it for some other reason, and the honest move is to say the other reason instead — which is almost always a value, and almost always lands better than the empirical-sounding version.
This is not a demand for intellectual purity. It is practical: an empirical claim you would not abandon is a claim the other side can refute without changing anything, which wastes both their time and your credibility.
The one thing an empirical claim buys you
It is worth being clear about why anybody reaches for this kind in the first place, because the incentive is real.
An empirical claim, if it holds, settles the matter for everybody including people who do not share your values. That is a genuinely valuable property and nothing else in the argument has it. A value claim persuades only those already inclined; a cost claim constrains rather than persuades. Evidence, when it exists and bears on the question, is the only move that can end a disagreement rather than trade it.
Which is exactly why it is worth protecting the category from claims that cannot cash the cheque. On this record, nine claims were made in the empirical register and two of them could actually have been settled with evidence that existed. The other seven borrowed the authority of the first two without being able to deliver what makes that authority worth having.