A list is permission, not a database
The number of addresses on a list is the least useful thing to know about it. What matters is how each address got there, what the person was told at the time, and whether what they have received since matches it. Two lists of the same size can behave entirely differently: one built from sign-up forms that promised a weekly digest, the other assembled from checkout boxes ticked by default, a conference badge scanner and a file somebody bought. The first list opens, clicks and buys. The second complains, and complaints are the signal mailbox providers weight most heavily when deciding whether a sender's mail belongs in the inbox.
This is why list building and deliverabilityWhether sent email reaches the inbox rather than spam or rejection, decided by sender reputation. are inseparable. The deliverability guide covers what happens after a message is sent; this guide covers the decisions before it that determine how that message will be received. The working principle is simple to state and hard to keep: collect addresses only from people who have shown they want them, tell them what they are signing up for, record it, and make leaving as easy as joining. Everything below is the practice of that principle.
Sign-up placements and offers
A sign-up form converts when it appears at the moment interest is highest and asks for the smallest thing it can. The placement decides the moment; the offer decides whether the moment is enough.
| Placement | When it works | Offer that fits | Watch for |
|---|---|---|---|
| Checkout or account creation | The person is already giving an address; one clear, unticked box adds marketing consent | Order updates plus an honest description of the marketing mail | Pre-ticked boxes, bundled consent, vague wording |
| Inline on content pages | A reader has finished something useful and wants more of it | The next piece: a series, a guide, a template related to the page | Forms placed before the content has proved itself |
| Exit or timed overlay | A visitor is leaving a page with clear commercial intent | A first-order incentive or a reminder of what the newsletter contains | Overlays on every page, on mobile, or within seconds of arrival |
| Footer and navigation | A persistent, low-pressure option for people who look for it | A one-line promise of frequency and content | Treating it as the only placement |
| Post-purchase and in-product | Trust is highest just after a good experience | Tips for using what they bought; early access | Asking for marketing consent inside a transactional message without a separate choice |
| Events and offline | Face-to-face conversations with a real reason to follow up | A specific follow-up the person asked for | Scanning badges and assuming consent |
| Gated resources and tools | The person wants something that took effort to produce | The resource itself, with a separate choice about ongoing mail | Treating a download as consent for a newsletter |
Measure each placement by its sign-up rate (the CVRShare of visits or clicks that convert. of visitors to subscribers) and, later, by the behaviour of the subscribers it produced; a placement that converts well but yields people who never open is a cost, not a win. Sign-ups taken face to face at a till, counter or reception desk, or on the phone, are set up step by step in the tutorial on collecting email sign-ups in person. The conversion rate calculator turns impressions and sign-ups into the rate to compare. Where a form asks for more than an address, its form abandonment rateThe share of people who start filling in a form, by entering a first field, but leave without submitting it., the share of people who start it and leave without submitting, shows whether the extra questions are costing subscribers. The offer should be the thing the newsletter actually delivers, so that the welcome message confirms a promise rather than making a new one; the welcome series tutorial picks up from the moment the form is submitted. Where the list exists to produce sales enquiries, the guide to email for lead generation explains why the sign-up offer should be one only the target buyer wants.
Where an exit overlay is under consideration, the exit-intent popup test weighs the extra sign-ups against interrupted purchases by judging the popup on revenue per visitor.
Consent, lawful basis and records
Consent is the normal basis for sending marketing email to individuals in many jurisdictions, and the ones that define it tend to define it strictly. The EU's General Data Protection Regulation, which applies alongside separate electronic marketing rules in member states, describes consent as a freely given, specific, informed and unambiguous indication of the person's wishes, given by a statement or a clear affirmative action. It requires that the organisation be able to demonstrate that consent was given, that withdrawing consent be as easy as giving it, and it gives people the right to object to direct marketing at any time. The UK's Information Commissioner's Office publishes direct marketing guidance that explains the UK's rules on electronic mail marketing, including when consent is needed, the conditions under which existing customers can be contacted about similar products without fresh consent (often called the soft opt-in), and the requirement to offer a way to opt out in every message. Other countries have their own regimes, some consent-based and some opt-out-based, and the obligations often differ for consumers and for business contacts.
None of that is legal advice, and the right course is to read the regulator's guidance for each market a business mails and to take advice where the stakes justify it. What a marketer can do regardless of jurisdiction is build the record. For every address, store the source (which form, page or integration), the timestamp, the wording the person saw, the choices they made and, where double opt-inA sign-up process in which the subscriber confirms through a link sent to the address given. is used, the confirmation timestamp. A list that can answer "how did this person get here and what were they told?" for every row is defensible, re-permissionable and far easier to segment. A list that cannot is a liability in proportion to its size. The email sign-up source register template records the matching facts once per source: the placement, the wording shown, the opt-in method and the system that holds each record. The privacy and consent guide covers how consent is handled on the analytics side, and the two records should agree. Where several tools act on the same addresses, a consent map across the marketing stack records which system holds each permission and how an unsubscribe or objection reaches every tool that reads it. Recruitment is a common case where answering an enquiry and sending ongoing marketing need separate records, which the education industry guide covers for course providers and schools.
Single vs double opt-in
Single opt-in adds the address the moment the form is submitted. Double opt-in sends a confirmation message and adds the address only when the person clicks it. The trade is straightforward: double opt-in loses the people who never confirm, and gains a list in which every address is real, reachable and wanted by someone who took two deliberate actions.
Double opt-in is worth the friction when the cost of a bad address is high: when the list feeds expensive automation, when a market's rules make demonstrable consent valuable, when sign-up forms are exposed to bots or typos, or when the sending domain is new and has reputation to build. It is less compelling for low-stakes, high-volume capture where speed matters more than precision, provided the single opt-in form is clear and the first message gives an obvious way out. Many teams use both: double opt-in for forms on the open web, single opt-in at checkout where the address has just been used for an order. The comparison of single and double opt-in sets the two methods side by side and works through the choice source by source, from checkout to pop-ups, giveaways and sign-ups keyed in after events.
Whichever is chosen, the confirmation step should be designed, not left as a default. Send it immediately, make the subject line say what it is for, keep the message to one action, and treat the confirmation page as the first moment of the welcome experience rather than a dead end. Unconfirmed addresses should expire after a stated period rather than linger in the database as half-subscribers nobody can mail. For a form whose risk is unclear, the double opt-in test is a design that splits its submissions between the two processes and reads email revenue per submission.
Preference centres and expectations
Expectations are set in the sentence next to the sign-up box, and most churn traces back to a gap between that sentence and what arrived. "Get our newsletter" promises nothing in particular; "a short email every other Thursday with new arrivals and one recipe" is a promise that can be kept or broken. Write the promise, then build the programme to match it.
A preference centre is where that promise is renegotiated. At minimum it lets a subscriber choose which kinds of mail they receive and, where the programme has more than one cadence, how often. It should be reachable from every message, require no password, and apply changes immediately. Three rules keep it useful. Offer only preferences the team will act on; a choice that changes nothing breeds distrust. Keep the options few; a page of toggles is abandoned. And treat "fewer emails" as a real option that routes a person into a lower-frequency segment rather than as a softer unsubscribe that is quietly ignored. The segmentation guide covers how preferences become segments and how to keep the number of segments manageable. An app that sends push notifications needs the same renegotiation inside the app, and the tutorial on notification preferences by topic builds that settings screen, with app topics kept separate from email consent.
Imports, purchased lists and risk
Every list eventually faces an import: a spreadsheet from sales, a file from an acquisition, an export from an old platform, a list someone proposes to buy. The question for each is the same: did these people agree to receive this kind of mail from this organisation, and can that be shown?
Purchased and rented lists almost never pass that test. The people on them did not agree to hear from the buyer, regulators in consent-based jurisdictions treat bought-in lists as a risk area and say so in their guidance, and mailbox providers see the result as unsolicited mail with the complaint ratesShare of delivered emails that recipients marked as spam. to match. The damage lands on the sending domain, which means it reaches every legitimate subscriber too. Exports from an old platform are usually fine if the consent recordsThe stored evidence of a marketing permission: the channel, the wording the person saw, the date and the source of each opt-in, kept with the contact record. travel with them; files from sales or events depend entirely on what the people were told. Where consent is unclear, a re-permission campaignA short email sequence asking contacts with doubtful consent or long inactivity to confirm they still want email; non-responders are mailed no further., one message that asks for a positive confirmation and suppresses everyone who does not give it, converts an unknown list into a small known one. It feels like throwing addresses away. It is throwing away the addresses that would have cost more than they returned.
Any import should be staged: added to a holding segment, mailed in small batches with close attention to bounces and complaints, and merged only once it behaves like the rest of the list. The deliverability guide explains how volume changes and complaint spikes affect reputation, which is why batching matters. A partner's list offered for a joint promotion fails the same test unless its people agreed to hear from you, which is why the checklist for launching a joint campaign with a partner has each side promote to its own audience instead. Loading the file into the CRM is a separate job, and the tutorial on importing contact lists into the CRM covers the suppression check, matching and test sample that keep an import from overwriting existing records. A list of your own that has gone unmailed for a long time deserves the same care, and the tutorial on restarting email to a dormant list takes one from rebuilt consent records to staged sends.
Unsubscribe done properly
Unsubscribing is part of the product, not a failure of it. A person who leaves cleanly is a neutral signal; a person who cannot find the link, or who is asked to log in, or whose request takes a fortnight to process, complains instead, and complaints are what damage the domain.
The mechanics have two layers. The visible layer is a link in every marketing message, easy to find, that
works in one or two clicks without a login and confirms the result. The header layer is the one mailbox
providers act on. RFC 8058 specifies a way for a sender to signal that a message supports one-click
unsubscribe: the message carries a List-Unsubscribe header with an HTTPS address and a
List-Unsubscribe-Post header whose value is List-Unsubscribe=One-Click; a mail client can then send a
single POST request to that address, and the sender must process it as an unsubscribe without any further
interaction from the person. The specification requires the message to be DKIM-signed with those headers
covered by the signature, so that the request cannot be forged. Google's sender guidelines document one-click
unsubscribe as a requirement for bulk senders of marketing and subscribed mail, alongside a clearly visible
unsubscribe link in the body, with requests processed within the documented window; the
deliverability guide covers that requirement in context.
Whatever the platform, confirm three things: the headers are present on every marketing send, the one-click endpoint removes the person from marketing mail immediately, and the suppression persists even if the address is imported again later. Transactional mail is a separate question, which the promotional vs transactional comparison covers, but a person who has unsubscribed from marketing should never receive marketing content dressed as a receipt. Opt-outs that arrive outside the link, in a reply, a call or a social message, follow the marketing opt-out request SOP so they reach every tool that holds the address. When someone who unsubscribed is still receiving mail, the diagnosis for unsubscribed contacts still receiving email traces which system, rule or file let the message through.
A homeware retailer rebuilds its sign-up points
A homeware retailer with a list of 32,000 reviews its sign-up points over one quarter. The only form is a footer box converting 0.3% of visitors. The team adds an unticked marketing consent box at checkout, with the wording "A short email every other week with new pieces and care guides"; 58% of customers tick it. An inline form at the end of care-guide articles offers the full guide series and converts 4.1% of readers. An exit overlay on product pages, shown once per visitor and never on mobile, converts 1.9%. Double opt-in is switched on for the two web forms; 71% of sign-ups confirm within 24 hours and unconfirmed addresses expire after 14 days. After 90 days the list has grown by 5,400 confirmed addresses, the complaint rate on newsletters has fallen from 0.21% to 0.06%, and the welcome series converts 6% of new subscribers to a first order.
When an import or a form starts returning refusals, diagnosing a high email bounce rate sorts the bounces by type and source to show which list decision let them in.
When one send or one source suddenly loses subscribers, the unsubscribe spike diagnosis locates the exits by send, segment and sign-up source before any copy is rewritten.
A sunset policy for people who have stopped engaging can be tested before it is adopted with the inactive subscriber sunset experiment, which weighs the revenue given up against the complaints avoided.
When sign-ups slow or the list stops growing, the diagnosis for an email list that is not growing splits new arrivals by form and source from losses by type before any form is redesigned.
Common mistakes
- Counting addresses instead of permissions. A list's size says nothing about its value; its sources and consent records do.
- Pre-ticked boxes and bundled consent. A choice the person did not actively make is not a choice regulators or subscribers respect.
- Promising "updates" and sending promotions. The gap between the sign-up promise and the mail that arrives is the main driver of unsubscribes and complaints.
- Treating a download as a subscription. A resource request and ongoing marketing mail are two separate choices.
- Importing without a holding segment. Merged straight into the main list, one bad file damages placement for everyone.
- Hiding the unsubscribe link. Every person who cannot find it becomes a complaint instead of a clean exit.
- No record of source and wording. Without it the list cannot be defended, re-permissioned or segmented by origin.
- Confusing lower frequency with leaving. A "fewer emails" option that changes nothing teaches subscribers that preferences are decorative.
The fundamentals hub sets this guide alongside deliverability and segmentation. The list's growth rate is a KPIA metric tied directly to an objective. worth tracking, but only next to the engagement and click-through (CTRShare of impressions that result in a click.) of the subscribers it produces.
