602 of 1,266 properties publish one mailbox, and it is info@

What the part before the @ tells you

We keep a working list of independent hostels and small hotels, built from public sources: the properties' own websites and public map data. This morning it holds 2,609 properties. 1,321 of them publish no email address at all, which is its own problem and one I wrote about last week. 1,266 publish one we can actually use.

Sort those 1,266 by the part before the @ and it comes out like this:

• info@ — 602
• something else, mostly a person's name or a word our matcher does not know — 398
• reservas@ — 93
• contact@ — 41
• reservations@ — 40
• booking@ — 23
• contacto@ — 19
• hola@ — 18
• hello@ — 17
• bookings@ — 9
• admin@ — 6

602 of 1,266 is 47.6 per cent. Every mailbox named for reservations — reservas, reservations, booking, bookings — adds up to 165. The generic mailbox outnumbers the reservations mailbox by almost four to one.

Why a mailbox name is a revenue question

A mailbox called info@ is nobody's mailbox. It is where the supplier invoices land, and the OTA notifications, and the tax mail, and the man selling solar panels. A guest asking whether you have two beds in the mixed dorm on the 14th is standing in that queue, behind all of it.

That guest is worth more than your average booking, because of when they wrote. They found you, they compared you, and then instead of pressing the button on an OTA they opened a mail client. That is the one moment in the whole funnel where the booking is available to you at no commission.

Here is what the commission actually is, measured rather than assumed. When we calibrated the channel economics inside our own pricing engine we used one real booking of eight bed nights through Hostelworld. The promotions on it stacked multiplicatively, not additively: a length-of-stay discount, a mobile discount and an app discount, each 10 per cent, came out as 0.9 x 0.9 x 0.9 = 0.729, and the effective commission on that booking worked out to 26.7 per cent. The worst case we encode for Booking.com is 20 per cent commission against a 1.20 markup.

Put a number on the enquiry. Four nights, two people, a bed at USD 12 a night: USD 96. At 20 per cent you hand over USD 19. At the 26.7 per cent we actually measured, USD 26. That is what answering one email is worth, before you count whether the guest comes back or brings anyone. An enquiry that sits unread for two days is not a slow reply. It is either a booking you paid commission on for no reason, or one you lost.

What happens inside those mailboxes, as far as we can see it

We write to these properties, so we have a measurement most operators never get: what a hostel's published mailbox does when a stranger writes to it.

The last seven days, from our own send log: 95 messages sent, 93 delivered, 2 hard bounced. Of the 93 mailboxes that took delivery, 11 opened the message. That is 11.8 per cent. One click, from a person rather than a scanner. Zero replies.

Now the caveat, because that last figure is weaker than it looks. Our own inbound reply route has recorded zero requests since it was built. The counter stands at 0, so a reply arriving today would have nothing on our side to record it. The zero is not evidence that hostels do not answer; it is evidence we cannot yet see them answering. The opens are firmer, because they are counted at the mailbox: 82 of 93 mailboxes did not open at all in seven days. And we throw away opens that are only a mail gateway loading a tracking image within two minutes of delivery. Four of this week's fifteen raw openers were exactly that, which is why the honest figure is 11 and not 15.

The hypothesis we had to throw away

We expected the shape of an address to predict whether it would bounce. Specifically: an address on a domain that is not the property's own — a gmail, a hotmail, an agency's domain — ought to be stale more often than one on the hostel's own domain. 96 of our 1,266 usable addresses are mismatched that way, 7.6 per cent, so there was enough of it to test.

It did not hold. Domain mismatch predicted nothing about bouncing in our data, and we dropped it rather than keep a rule that merely felt right.

What looks more promising, on a sample far too small to act on: where the address came from. Both hard bounces in the last seven days came from addresses lifted off a public map tag, and none from addresses read off the property's own website. Two bounces on 47 first contacts is not a finding, it is a direction, and we have written down in advance what would kill it.

One more thing worth knowing if you are one of these properties: four addresses on the list belong to more than one property. Somewhere, two hostels' enquiries are landing in one inbox.

Three changes, in the order I would make them

One. Give reservations its own mailbox, with one name on it. Not a filter or a label inside the shared box. A separate mailbox, with one person responsible and a stated target for the first reply. The point is not tidiness. It is that a shared box has no owner, and a queue with no owner has no response time.

Two. Publish it where the guest already is. On the page they land on, near the price, not only on a contact page two clicks down. Our own site took 4 of 13 sessions from phones last week, and most hostel traffic leans further that way than ours does, so the tap target matters as much as the wording.

Three. Measure hours from enquiry to first human reply, and nothing else, for one month. If you cannot say what that number was last week, that is the first thing to instrument, and it costs nothing but writing down two timestamps. Occupancy tells you what already happened. This one tells you what is about to.

And if email is not really your channel — in Mexico and Colombia a great many properties sell on WhatsApp and always have — then say so on the site, publish the number, and let the mailbox be honest about being a mailbox.

Read the caveats before you borrow any of this

• 2,609 properties from public sources, weighted toward the cities we prospect: Madrid 227, Lisbon 118, Oaxaca 93, Chiang Mai 91, Barcelona 67. It is not a census of anything.
• 1,266 is the count we hold a usable address for. The other 1,321 publish none, so none of the above describes them.
• The prefix sort is mechanical. The 398 in "other" include owners' own names, and some of them are certainly reservations mailboxes under a name our matcher does not recognise.
• Seven days and 93 delivered messages is a small window. Every rate in it moves on single events, which is exactly why we publish the counts beside the percentages.

If this sounds like your property, see the three Nightfill tiers and start a free 14-day pilot on your own data: view pricing.