Working out whether a claim is actually supported, including the ones you'd like to be true.
Critical thinking is the habit of asking what would have to be true for a claim to hold, and then checking. Not cynicism, and not doubting everything equally — that's just as lazy as believing everything, and slower. The work is mostly unglamorous: noticing where a number came from, separating what was observed from what was inferred, and being hardest on the conclusions you find most congenial, because those are the ones nobody else will bother to challenge in front of you. Most bad workplace reasoning isn't a logical error anyone could name. It's a plausible story that nobody stopped to check the base of.
Researches diligently, at length, and entirely in one direction. Arrives with fourteen supporting sources for the position held before the research began, having read past a great deal of contrary material without registering it as relevant. The effort is genuine and the search is real; only the query was rigged. Would be willing to change their mind if the evidence pointed the other way, and has constructed a process that reliably guarantees it never appears to.
Cites a well-known figure with confidence. The figure comes from a slide deck, which took it from a blog post, which took it from an article summarizing a study that measured something adjacent and concluded rather less. Each hop was made in good faith and stripped one qualifier, so that a tentative finding about forty people has arrived, four transfers later, as an industry benchmark that everyone repeats and nobody can trace.
Evaluates claims by how right they sound. Accepts the well-phrased, internally consistent account delivered by someone confident, and rejects the awkward one full of caveats, on grounds that feel like judgment but are closer to acoustics. Has an excellent instinct for whether an argument hangs together and no procedure at all for whether it corresponds to anything, which makes them reliably persuadable by exactly the people best at constructing arguments.
Settles arguments by producing a number. The number exists, is accurate, and was collected carefully — it simply measures something other than the thing under discussion, a distinction that dissolves the moment it appears on a slide with a decimal point. Treats "we have data on that" as equivalent to "that question is resolved," and has never once been asked, in a meeting, what the metric's denominator actually contains.
A vendor's benchmark shows their product is four times faster. What would you do — accept it as directionally right, or find out what the comparison was run against, on what hardware, and who chose the workload?
You're building the case for a decision you already believe in. What would you do — assemble the strongest supporting evidence, or spend the first hour looking specifically for the best argument against it?
Someone quotes a well-known industry statistic to close down a debate. What would you do — accept it as common knowledge, or ask where it comes from, given that "everyone knows it" is a claim about circulation rather than evidence?