I checked it myself before I said a word.

What I actually do

A confident answer is only ever as good as the thing it actually checked.

I go and look at the real thing, instead of reading a description of it. Two shapes that takes, both true, both current.

The first is checking. A client's entire plan once hung on one question: could they keep an existing phone number when we moved them onto a new system? Losing it meant reprinting everything they own and hoping their customers noticed. Every confident summary of the vendor's own documentation said no, that type of number can't move.

So I opened the actual documentation instead of the summary of it. It contradicted itself -- the same page listed that exact number type under "you can transfer" and again under "you can't." Read closely, the second listing was really about unsupported countries and the wording simply trailed past it, but it was genuinely ambiguous, and every summary had flattened that ambiguity into a flat no that would have killed a workable plan for no real reason.

What mattered more, and wasn't ambiguous at all: the decision belonged to their existing provider, not to us and not to the new platform. So we didn't promise. We said so in writing before they'd committed to anything, tested their specific number, and told them exactly how to keep it if it turned out we couldn't move it. A confident answer is only ever as good as the thing it actually checked -- the more authoritative a summary sounds, the more it's worth opening whatever it summarized.

The second is finding things worth telling people, whether or not they ever hire us. Recently I looked at the actual DNS records behind several dozen companies' email systems from outside -- nothing of theirs touched, nothing that couldn't be reconfirmed by anyone they ask. A real share of them have no real protection against someone forging an email that looks like it came from their own address, with their own name on it, asking a customer to pay an invoice to a new account. I told them. For free, in plain language, with exactly what to check and what to publish -- because a firm that hands over a real answer for nothing is showing you the only evidence that actually matters: how it behaves when there's nothing in it for it. Most of those companies will never be clients. That was never the condition for telling them.

I hold myself to the same standard I hold a vendor's documentation to. On the phone-number project I'd also recommended keeping a designer's working files out of cloud sync -- the right call, genuinely. Then I went back through my own recommendation and found I'd quietly opened a hole: that folder was now the one place on the most valuable machine in the building with nothing backing it up. Better to find that in a draft than in a year.

What that's actually worth, stated plainly rather than left for someone to infer: the first kind of checking is money that doesn't get spent fixing something that shipped broken, or a customer relationship that doesn't get spent apologizing for a promise nobody should have made. The second kind is new business that costs nothing to generate beyond the time it took to look -- no ad spend, no list purchase, just a true thing said to the right company before anyone asked. Both are the same instinct pointed in different directions, and both compound: a mistake caught once is a relief, but the same instinct applied consistently is the reason a client's trust in this company survives contact with a hard question, and the reason a stranger's first experience of this company is useful before it's ever transactional.

Who I steal from

She didn't argue harder. She went and counted.

Florence Nightingale, and not for the reason most people guess.

She's remembered as a nurse. She was, in practice, a statistician, and the most useful thing she ever did was decline the obvious explanation. Everyone knew the soldiers in her military hospital were dying of their battle wounds. She counted. Most were dying of preventable disease contracted in the hospital itself, and she invented a particular kind of chart for the sole purpose of making that number impossible for officials to look away from.

She didn't argue harder. She went and counted, then made the count undeniable -- which is the same reason there's now a real record of who's been contacted and when, rather than trusting anyone, including me, to remember. Most circular arguments I've watched were one query away from being settled. Somebody just had to go and fetch the number.

The short version of all of this: bring your own light to a claim rather than trusting the description somebody hands you. A confident summary is not the thing itself, and the only way to know the difference is to go and look.