A cold email system that does not burn your domain sends fewer emails than you want.

Outbound plans tend to begin with the number of meetings somebody wants by the end of the quarter and work backwards from there, to a daily send count that looks fine in a spreadsheet and alarming to anyone who has watched a domain go cold. A company growing faster than it can hire has a particular reason to like the big number. There is nobody free to do outreach carefully, so the hope is that volume will make up for the care.

The trouble is the order things happen in. Mailbox providers watch how a sending domain behaves, and one that jumps from quiet to a flood of cold mail, with bounces climbing behind it, starts landing in spam well before any buyer has had the chance to say yes. By the time the volume would have paid off, the emails are arriving in a folder people empty without reading, and there is no one at the other end to apologise to.

The outreach OCTYN runs for an independent craft retailer is set up the other way round. It sends 25 first emails a day, 15 to its home market and 10 abroad, a split that has held since 25 August. Each inbox has limits on how many new emails and follow ups it sends. Follow ups go three and five days after the first email, under a daily cap of 60. Nothing at all sends at the weekend.

Why the daily number stays small

The retailer makes handmade goods and sells them to brands, where the volume is small and the cost of one wrong email is large. Under 300 companies were contacted to land its brand deals, with 284 counted at the 11 August reading. A made up detail, or an email dropped into a deal already under way, costs a relationship, and a list of under 300 does not have many to spare.

Twenty five a day looks timid next to what most outbound software will let you send. It is also a queue one person can read in the single window a day the system asks of them. Nothing goes out in the morning, so the owner opens the day's queue, reads it and pulls anything they dislike before the sender reaches it. Approval on each email was left out on purpose, since at 25 a day it would turn into rubber stamping within a week.

The small number has a ceiling built into the rest of the system, too. Researching and drafting happen at night, in a deliberately slow working session. Its notes name a daily volume far past 25 as the one case that would call for a rebuild. Turning the number up would mean building a different engine.

Where the caps actually live

A cap written in a planning doc lasts until the first good week. The retailer's caps are values in its settings: daily volume, the split between markets, the inbox limits, the send windows and the follow up timing. Changing one is a settings change, and keeping it does not depend on anyone's willpower.

The follow up cap is the one that bites, which is easiest to see if, for the arithmetic, nobody ever replies. Each first email has two follow ups behind it, so 25 first emails a day become about 50 follow ups on an ordinary weekday once the sequence fills, and more on the days the calendar piles them up, against a cap of 60.

That leaves very little room, and the cap has been hit before. On 10 August, 115 follow ups came due across a weekend and nothing picked them up, because the check for due emails only looked at today, and Monday already held 93 against the cap of 60. The scheduler now moves overflow forward when it creates a follow up and moves weekend dates to Monday, so the cap is applied at the moment a date is chosen.

All of this is rate limiting in the ordinary sense. The limit is one you set on yourself, before the mailbox providers set a stricter one for you. Other companies run on the same engine with completely different criteria and their own numbers, and the wider system is described on the growth engine page.

The sender gets no opinions

The last piece is the sender, and it is the plainest thing in the system. A sender driven by a model was tried first, and among other things it reported a send that had not happened, which has its own post. The replacement wakes every minute and sends whatever is due, confirming each one as it goes.

This matters for the caps more than for anything else. A sender with judgement in it could decide that a quiet Tuesday has room for a few extra emails, or that a bounced address deserves another try. This one has no way to have either thought. It sends what the scheduler put in front of it, and when the caps say stop, it stops.

A bounce rate circuit breaker sits beside the caps as well, and it stopped a campaign on 11 September. Nobody noticed for three days, which is a post for another week.

Who the emails are going to

Some of the volume goes on addresses that were never going to work. When the retailer's leads were finally counted, seven in ten sat on shared inboxes, the general addresses a whole company reads and nobody in particular answers. A shared inbox cannot carry a follow up sequence. If the system were built again, contact quality would be checked at the front, before anything is drafted, and those names would come off the list before they cost a send.

One more rule takes names off. The moment anyone at a company answers, that company belongs to the conversation, and nobody else there gets a cold first email. Every rule in this post makes the daily number smaller than a volume plan would like.

Twenty five a day will never look good on a dashboard. It has been enough, so far, for a retailer that landed its brand deals from under 300 companies and would like to keep its domain.