My cell phone number went onto the front of actium.dev at 18:49. It came off at 04:11 the next morning. I wasn’t working for most of the nine hours in between, which turns out to be where the replacement came from.
There’s a good argument for publishing a number and it’s sitting in the commit that added it. Owners buying a small-business website have been sold to by agencies hiding behind contact forms, so a number you can actually dial answers the “will this person still be here in six months” question before anyone asks it. Publishing something that costs you is how you signal you aren’t planning to disappear.
That argument isn’t wrong. It just isn’t what I was reacting to when I took the number down, and I’d rather say so than pretend the decision came out of the reasoning. What bothered me was how the page read with it there. A phone number says call us, we’re waiting, which is the right thing for a plumber’s site to say and the wrong thing for mine. Actium finds its client work by me going out and finding it, and nothing about that shape is improved by a hotline on the front page.
The analysis caught up afterward and went further than the feeling had. The reader that costly-signal argument describes is someone I cold called, which means they already have my number, because I read it to them out loud during the call, and the number printed on the page could therefore only ever reach a stranger who found the page on their own without me. A studio that does its own outreach barely has those. I’d built a trust signal and pointed it at an empty room.
Which address?
So the page needed some other way to be reached and email was the obvious one. That left a question I expected to settle in about two minutes.
The instinct is dante@actium.dev, because that’s my address and I’m the whole
company. The problem is scraping: anything printed on a page written for
strangers gets harvested eventually, and harvesting is permanent in a way that
editing a web page never is. You can take an address off a site in ten seconds.
You can’t take it out of the lists it’s already been sold into.
Mail for the domain runs on iCloud, and there’s a setting in there called Allow
all incoming messages. All it does is this: when something arrives addressed to
anything at your domain, put it in your inbox. You don’t create the addresses
first, there’s nothing to configure, and there’s no mailbox waiting anywhere for
the mail to land in. sales@, hello@, xyzzy@ all work, because none of them
are real to begin with.
Which means I could publish an address I’d never made, and read the mail sent to it.
The address is a setup
The sales page now publishes websites@actium.dev. There is no such account. I
answer it from dante@actium.dev, by name.
A stranger on /websites
Reads the page, decides to write. Sees one address.
websites@actium.devdoes not exist
Published on the page. No account, no mailbox, nothing configured.
iCloud catch-all
One setting: mail for any address at the domain lands in the one inbox.
dante@actium.dev
Where it actually arrives, and the address the reply comes back from.
That sounds like a shortcut, and it’s the actual feature.
It arrived whole, in the few seconds after I stopped trying to solve the scraping problem and started picturing the person on the other end: what the address tells them before they write, who answers, and the size of the gap between those two. The cost came attached, which is the part that made me trust it. An idea that shows up without its downside already visible is usually a preference wearing a reason.
A role address tells you what to expect before you’ve typed a word. Writing to
websites@ or info@ or support@ feels like writing to a queue: an
autoresponder, a ticket number, somebody whose job is the first fifteen seconds
of your problem and not the rest of it. Nobody sits down to email info@
expecting the founder, and that low expectation gets set by the address itself,
for free, before a word of the message is written.
Then a reply arrives a few hours later from a named person who has opened your actual site, looked at it properly, and can tell you the homepage weighs eight megabytes and which photograph is most of that.
The distance between those two moments is the impression, and I didn’t have to claim any of it. Every agency site on earth promises you’ll work directly with someone senior. It’s the least verifiable sentence in the industry and everyone who has been burned once reads straight past it. This version never says it. It demonstrates it, in the first reply, before anyone has paid or signed anything.
Why I’m not trying to look bigger
A better-funded competitor can buy a nicer site, a real support desk and a team to sit behind it, but what they can’t buy is a founder with enough time to answer the mail himself, because the whole reason they hired the support desk is that theirs doesn’t. This only works for a company small enough that it’s true. It’s the rare kind of advantage that shrinks as you grow, which is the opposite of how advantages usually work and the reason to spend it now rather than save it.
Most small operators push the other way. They write “we” when they mean “I”, invent a team page, buy an office address, and generally try to resemble the thing they’re competing with. I understand the instinct and I think it’s backwards. If the reason to pick me over an agency is that you get the person who actually builds it, then every dollar spent looking bigger is spent hiding the only real difference.
Nobody invented anything here, to be clear. Catch-all addressing has existed for
decades and every small company produces this effect by accident the first time a
founder answers info@. What I did was notice the accident was better than the
configuration I was about to build, and then not build it.
What it costs
The first thing is that websites@ isn’t disposable. That was supposed to be
half the point of a separate address: if it gets scraped and buried, delete it,
publish a new one, and the junk stops. With the catch-all on, deleting it does
nothing, because every address at the domain resolves whether it exists or not. I
traded the rotation for the reply and I’d make that trade again, but it is a
trade.
The second is that it’s fragile to success, and it fails late. At ten emails a week I can answer every one personally within a few hours. At fifty I can’t, and by then I’ll have spent months teaching people to expect that I will. The thing to watch isn’t whether this stops working. It’s that it stops working after it has already made a promise.