How to start an email list from nothing

1. Write the promise in one sentence, before anything else

Open a blank file and write what somebody gets for their address, in the words they would use, with a number in it. Not a newsletter about productivity, but five emails that get a first paying client, one a day, about ten minutes each. If you cannot finish the sentence, you do not yet have a list to start; you have a topic.

Test the sentence on one person who is not you. Ask them what they would expect to arrive and when. If their answer differs from yours, the sentence is not specific enough yet, and every later step will inherit the vagueness. This is the highest-leverage fifteen minutes in the whole process, and it is the step people skip in favour of choosing software.

Keep the sentence. It becomes the heading on the page, the line in the form, the first sentence of the first message, and the standard you measure the list against later. A promise that appears in all four places in the same words is what makes a new subscriber recognise your first email instead of deleting it.

2. Set up a sending identity that authenticates

Mail from a personal address does not survive being sent in bulk, and mail from a domain with no published authentication records is increasingly not delivered at all. The large mailbox providers ask bulk senders for SPF and DKIM on the sending domain, a DMARC record, valid forward and reverse DNS, and a complaint rate kept below the threshold they publish.

Publish the records before the first send, not after the first problem, because a reputation is built from the first message onward and a bad start takes weeks to undo. Choose the sending domain deliberately: mail to strangers and mail that matters to you personally should not share a reputation, which is the argument for sending course or newsletter mail from a domain of its own.

Then warm up rather than launching. Send to a handful of addresses, then dozens, then hundreds, over days. A domain with no history that suddenly sends a thousand messages looks exactly like a compromised one, and the mailbox providers respond to it as such.

3. Build one page whose only job is the offer

Make a page that contains the promise, what arrives on each day or in each part, who you are in one line, and a form with a single field. Nothing else: no navigation to your other work, no three other calls to action, no carousel. A page like this commonly converts 15 to 30 per cent of the people who reach it, in the published landing-page benchmarks, where the same form in a sidebar typically converts 1 to 3 per cent.

Ask for the address only. A first name field is commonly reported to cost around a tenth of sign-ups, and it buys nothing when what follows is identical for everyone. Say beside the button what will happen next, including that a confirmation message is coming and what name it will arrive from — that one sentence is the cheapest recovery of confirmation losses available.

Write the page for somebody who has never heard of you and will decide in nine seconds. The most common defect is a page that describes the subject rather than the thing being sent: a reader can evaluate five emails, one a day, each ten minutes. They cannot evaluate insights on growth.

4. Make every sign-up confirm before anything else is sent

Set the list up so a submitted address receives exactly one message — a confirmation — and nothing further until the link in it is clicked. This is confirmed opt-in, which most people call double opt-in, and starting with it is much easier than adding it to a list that already has ten thousand unconfirmed addresses on it.

It costs sign-ups, and the cost is published: a fifth to a third of sign-ups never confirm, which is the band commonly cited in double opt-in comparisons. What it buys is that every address on the list is provably a person who acted twice, that mistyped addresses never get a second message, and that a bot filling in your form cannot put you in front of a spam trap.

Keep the evidence while you are at it. Record the wording the person agreed to, the timestamp, the page and the campaign, and the moment of confirmation. The GDPR requires you to be able to demonstrate consent, and a record that says website with no date is not a demonstration of anything.

5. Write the first message before you collect one address

The message that arrives after confirmation is the only one you can be sure of being read, so it should deliver the promised thing rather than announce it. Put the substance in the body where it can be read without a click, name yourself, and give one small action a reader can complete in a few minutes.

Having it written before launch is the point of doing this now. A list whose first message does not exist yet collects addresses into silence, and silence in the first day is what turns a new subscriber into somebody who does not recognise your name a week later.

Then decide what the next few days look like while the attention lasts. A finite sequence that visibly ends is read more completely than an open-ended one, because the reader can see the end and has a reason to stay to it. Five days is a common shape for exactly that reason.

6. Ask the people you already know, one at a time

The first fifty addresses do not come from search or advertising. They come from individual messages to people who already know you: past clients, people who replied to something you wrote, the contacts in your own sent folder. Send each of them the promise sentence and a link to the page.

Ask; do not add. Typing somebody into a list because you have their card is not consent, and in the European Union it is not lawful — the narrow exemption for existing customers covers similar products and still requires an opt-out offered when the address was collected. A personal ask converts better than anything you can buy anyway, so the lawful route is also the effective one.

These first readers are worth more than their number. They will tell you whether the promise was worth making, which sentence in the first message confused them, and whether the thing arrived at all. Nothing later in the life of the list gives you feedback this cheaply.

7. Measure the four stages, then repeat the one that is weakest

Once mail is moving, count four things in order: visitors to the page, submissions as a share of those visitors, confirmations as a share of submissions, and clicks on the first message. Each one only means something if the one above it is healthy, and the weakest of the four is the only work worth doing next.

Do not use open rates. Apple Mail Privacy Protection loads remote images through a proxy whether or not anyone looked, so the open rate on a consumer list rises when nobody reads anything. Measure a click on a link that a reader who actually did the thing had a reason to press.

Then widen the traffic, one channel at a time, and watch the complaint and bounce rates as you do. Those two are conditions on your mail being delivered at all, which makes them the only numbers on a new list that can take the channel away entirely.

The weakest stage names the work, and it is rarely the work people choose:

• Few visitors: the page exists and nothing links to it. Put the link in everything that mentions the subject.
• Visitors but few submissions: the promise is not specific, or the traffic came for something else.
• Submissions but few confirmations: the confirmation is landing badly, was not expected, or arrives too late to be recognised.
• Confirmations but no clicks on the first message: the first message explains instead of delivering.