Email marketing automation starts with consent and a clean list
Before you build the first sequence, check three things: whether the contact has consent on record, whether the list is alive, and whether the domain is authenticated. Without them, automation just sends messages nobody receives faster, and drags down the deliverability of invoices going out from the same domain. Below: the order we put this in, and what to measure instead of opens.
- 3 DNS recordsSPF, DKIM and DMARC on the sending domain; without them the receiving server cannot confirm the message came from you
- Art. 7(3) GDPRconsent has to be as easy to withdraw as it was to give: one click in the footer, no login, no email to support
- 2 to 6 weekshow long it takes us to get from workshop to working sequences, list cleanup and domain setup included
- EUR 3,500 netone week of the process workshop, which ends with a consent map, a segment list and a plan for cleaning the database
Short answer
Where do you start with email marketing automation?
Start with three basics, not with a sequence. First, consent: a contact with no consent on record does not enter a send, even if it has been sitting in the CRM since 2019. Second, the list: hard bounces, addresses of people who left their company, and contacts from years ago go out before the first campaign, not after it. Third, the domain: SPF, DKIM and DMARC go in before anything ships.
Segments and sequences come after that. Segment on data you actually hold in the CRM and the order system: deal stage, date of last purchase, sign-up source, industry. A trait you cannot read out of a system without asking a person is not a segment, it is a wish.
Measure clicks, replies and the revenue attributed to a send. Leave opens as a secondary indicator: privacy-protecting mail clients fetch the tracking pixel themselves, so an open is recorded even when nobody read the message.
Why does consent decide who enters a send at all?
Marketing consent is not a checkbox on a form. It is a record in the database: who, when, from which source, and for what kind of communication. GDPR lists consent among the lawful bases for processing (Art. 6(1)(a)) and requires that it be as easy to withdraw as it was to give (Art. 7(3)). Separately, the ePrivacy Directive 2002/58/EC governs unsolicited commercial communication: those rules cover the contact channel, not the data.
The practical effect is simple. A contact with no consent on record enters no marketing sequence, and automation does not fix that, it just distributes the problem faster. So the first job in the workshop is walking through every source of contacts and checking which ones you can use.
- Date and time of sign-up together with the source: form, webinar, downloaded price list, trade fair
- The wording of the consent as it stood on the day of sign-up, not as it reads today
- A separate field for marketing consent and a separate one for sales contact, because those are two different things
- Date and method of withdrawal, back in the CRM the minute somebody clicks unsubscribe
- Imported and purchased contacts flagged separately, so they never slip into a send by accident
List hygiene: what do you remove before the first campaign?
A list nobody has cleaned for three years is the biggest threat to the whole company's deliverability. Mail servers score the reputation of a domain, not of a single campaign. If a marketing send collects bounces and complaints, the same drop hits the invoices and quotes leaving the same domain.
So the cleanup happens before the start. Below are the contact types we usually find in a database.
| Contact type | What we do with it | Why |
|---|---|---|
| Hard bounce, the address does not exist | remove after the first bounce | a second attempt tells the filter the sender is mailing blind |
| Soft bounce repeated 3 times | quarantine, then unsubscribe | full mailbox or a server block; after three tries it is not temporary |
| Address of someone who left the client | update from the CRM or delete | the address becomes a known bad one, and nobody is there to reply |
| Contact with no consent on record | excluded from marketing sends | no lawful basis, and nothing to defend with if a complaint lands |
| No reaction for 12 months | re-engagement campaign, then unsubscribe | dead addresses drag down engagement across the whole list |
| Shared addresses like office@ and contact@ | separate list, outside personalised sequences | several people read them, and a first-name greeting reads badly there |
Remove bounces and contacts without consent first, then count the size of the list. A list one third smaller but alive is a better starting point than a full database from five years ago.
SPF, DKIM and DMARC: three records the rest depends on
Before a receiving server delivers a message, it checks whether the sender is allowed to send on behalf of your domain. It does that from three DNS records. Missing even one means part of your mail lands in spam or disappears with no notice, and all you see is a weak result with no idea where it came from.
This is hours of work, not weeks, and you do it once. It only takes longer when DNS access sits with an outside supplier who answers next week.
- SPF: the list of servers allowed to send for the domain; your sending platform has to be on it
- DKIM: a cryptographic signature on the message, proving the content was not altered in transit
- DMARC: the instruction for what to do with a message that passes neither SPF nor DKIM, plus reports on attempts to spoof your domain
- A separate subdomain for bulk sending, so a campaign does not spend the reputation of the address your invoices go out from
How do you segment a list from the data you actually have?
Segmentation comes down to one question: can this trait be read out of a system without asking a person? If not, the segment stops updating within a week. Start with the fields that are filled in your CRM and your order system, then add more.
Four to six segments is enough to start. More means every one of them gets copy written in a hurry. When the list sits in Mailchimp, we build segments from data pushed in from the CRM rather than from manual tags; the scope of that connection is on the Mailchimp integration page.
- Deal stage in the CRM: lead, quote sent, customer, lapsed customer
- Date of the last order or the last invoice, straight from the sales system
- Sign-up source: form, webinar, downloaded price list, trade fair; each one carries a different sending rhythm
- Industry and company size, but only if those fields are filled in the CRM and not empty on half the records
- Behaviour in earlier sends: a click on a specific subject, a return visit to the pricing page
What do you measure instead of opens?
Opens were a convenient metric while a fetched image was what counted them. Privacy-protecting mail clients fetch that image on their own servers before the person sees the message, so an open is recorded even when nobody read anything. The metric still works for comparing two subject lines in a test, but not for judging results.
For the report, take what you can tie back to sales. Below are the metrics we set up before the first send, and where each one is read from.
| Metric | Where you read it | What you do with it |
|---|---|---|
| Replies to the message | the sender's mailbox, or the CRM if mail is logged there | a real signal of interest; the contact goes to a salesperson the same day |
| Clicks on the target link | the sending platform | you compare content against content; the click on the offer counts, the one in the footer does not |
| Unsubscribes and spam complaints | the sending platform, DMARC reports | if they climb, you cut the frequency or fix the segment |
| Hard bounces | the sending platform | you clean the list straight away, because this hits domain reputation |
| Opportunities and orders from contacts in the send | CRM or order system, by contact source | the only number that holds up at a board review |
| Opens | the sending platform | only for comparing two subject lines in a test, with the caveat that they are inflated |
We write the metrics down before the first send and read them at 30 and 90 days. A metric that cannot be read without manual work does not go on the list: two months in, nobody will calculate it.
What sending rhythm will your list stand?
You set the rhythm once and hold it. Irregular sending hurts twice over: the recipient forgets who is writing to them, and mail servers see volume spikes that look like sending from a hijacked account.
The starting point out of our workshops looks like this: a newsletter every two weeks or once a month, an event-triggered sequence spaced 3 to 5 working days apart, transactional messages immediately. Triggers and the moment a contact is handed to a salesperson are covered on the automatic email sequences page.
- Newsletter: every two weeks or once a month, same time of day and same sender
- Event-triggered sequence: 3 to 5 messages, 3 to 5 working days apart, stopping when there is no reaction
- New list or new domain: volume climbs gradually over 2 to 3 weeks
- One recipient does not get two sends on the same day; that is a rule in the platform, not goodwill
How do you roll this out, step by step?
Order matters, because each step depends on the one before it. This is how we sequence it on implementations: process first, tools second. The wider scope, including campaign measurement and handing a lead to sales, is on the marketing automation page.
Consent audit
We walk through every source of contacts and check where consent is recorded, with its date and wording. Contacts without consent go to a separate list, outside marketing sends.
List cleanup
We remove hard bounces and dead addresses, and flag contacts with no reaction for 12 months. We record the count before and after: that is the reference point for the metrics later.
Domain and DNS records
SPF, DKIM and DMARC on the sending domain, a separate subdomain for bulk sending, warm-up over 2 to 3 weeks. Tested on real Gmail and Outlook mailboxes, not in the platform preview.
Segments and the first sequences
Four to six segments built from fields that are actually filled in the CRM. Two sequences to start, each with a trigger, its intervals, and the point where the contact goes back to a salesperson.
Measurement at 30 and 90 days
We read the metrics written down before the start: clicks, replies, unsubscribes, revenue from contacts in the send. At 30 days we fix what is out of line, at 90 we compare with the reference point.
Common mistakes
Six mistakes that burn a domain and a campaign
Importing an old database into a new platform without checking consent
The first campaign collects hard bounces and complaints. Domain reputation drops, and with it the deliverability of invoices and quotes from the same address.
Sequences built before SPF, DKIM and DMARC
In the dashboard the automation runs perfectly and the messages do not arrive. The result looks like a lack of interest when it is a DNS configuration problem.
Judging a campaign on opens
Privacy protection in the mail client fetches the tracking pixel by itself, so opens are inflated. A decision to rewrite the copy gets made on a number that measures nothing.
Segments built from fields nobody fills in
Half the contacts fall outside every segment and get the generic send, or nothing at all. The work on personalisation turns into cost with no effect.
Unsubscribes handled by hand, outside the system
Somebody unsubscribes from the newsletter and gets another message a week later, because the information never made it back to the CRM. That is no longer a slip, it breaks Art. 7(3) GDPR.
Months of silence and then a sudden send to the whole database
The volume spike looks to mail servers like sending from a hijacked account. Part of the mail does not get past the filter, and complaints go up.
Checklist
Checklist before the first automated send
If any line is a no, the sequences can wait another week.
- Every contact in the send has consent on record: date, source, and the wording from the day of sign-up
- Hard bounces removed automatically after the first failure
- SPF, DKIM and DMARC set up and verified on a real Gmail and Outlook mailbox
- Bulk sending goes from a subdomain, not from the address your invoices leave from
- Unsubscribes come back to the CRM automatically, in the same minute
- Segments built from fields filled in the system, not from assumptions about the reader
- Metrics written down before the first send, with reading dates at 30 and 90 days
- Sending rhythm agreed in writing: what, to whom, and how often
Sources
Sources
- Regulation 2016/679 (GDPR): Art. 6(1)(a) on consent as a lawful basis and Art. 7(3) on withdrawing it
- Directive 2002/58/EC on privacy and electronic communications: rules on unsolicited commercial communications
- DMARC.org: documentation of the DMARC standard and how it relates to SPF and DKIM
- Mailchimp help centre: list hygiene, deliverability and the effect of privacy protection on the open rate
- JustAutomate, automatic email sequences: triggers, intervals and the moment a contact goes to a salesperson
Let us start with consent and the list
The workshop ends with a consent map and a cleanup plan
On a free 30-minute consultation we go through your sources of contacts and say plainly what you can send today and what needs cleaning up first. The process workshop costs EUR 3,500 net per week and ends with a consent map, a segment list and a plan for cleaning the database. Implementation is quoted after the workshop. We run at most two implementations at a time, and you write directly to Paweł, the founder.
Book a free consultationFrequently asked questions
Can I send marketing email to business addresses scraped from the web?
Not on the same footing as a list built from sign-ups. An address collected without consent has no basis for a marketing send, and the contact channel itself falls under the rules on unsolicited commercial communication in Directive 2002/58/EC. Cold outreach runs from a separate domain, with its own lawful basis agreed first.
How often should a mailing list be cleaned?
Remove hard bounces automatically after the first failure, which is a setting in the platform. Do a full review once a quarter: contacts with no reaction for 12 months go into a re-engagement campaign, and off the list after it. Following a longer break, add address verification on top.
Does email marketing automation mean changing platform?
Usually not. Mailchimp, ActiveCampaign, GetResponse and HubSpot all have automation on their higher plans, so the first thing we check is whether your plan covers it. Make or n8n come in when the trigger sits in a system the platform cannot see: in orders, in the ERP, or in a sheet of contract dates.
Why are email open rates inflated today?
An open is counted by the fetch of a small image embedded in the message. Privacy-protecting mail clients fetch that image on their own servers before the recipient sees anything. So the open is recorded even when nobody read a word of it, which is why it cannot carry a result on its own.
What happens to deliverability if we send to a bad list?
Mail servers score the reputation of a domain, not of a single campaign. A run of hard bounces and spam complaints lowers that reputation for several weeks, and it applies to every message from that domain: invoices, quotes and replies from your sales team. That is why bulk sending runs from a separate subdomain.
How long does it take and what does it cost?
The process workshop costs EUR 3,500 net per week, and for mailing one week is usually enough. Implementation is quoted after the workshop, because the scope depends on the state of the list, the number of integrations and whether a new sending domain has to be set up. From workshop to working sequences is usually 2 to 6 weeks.
How many segments should we start with?
Four to six, built from fields that are genuinely filled in the CRM: deal stage, date of last order, sign-up source, industry. More segments at the start means each one gets copy written in a hurry. You add the next ones once the first have results of their own.
Does an AI agent help with email marketing?
Yes, though not for writing the messages. BOND, our company AI agent, connects to the CRM and the sending platform and answers in plain sentences to questions like which segments had the most unsubscribes this quarter. The answer comes back in 5 to 10 seconds, with no need to line up two reports.