How to launch mailings to a cold database
Why do emails end up in spam? How to avoid this from the start. Why do you need SPF, DKIM, and DMARC?
Foreword
We were contacted by a client who had around 6,000 email addresses. According to the client, all of them had been collected through official channels: when registering on the website, on social networks, when participating in conferences and other events. In other words, these were potential customers who could order services. The services themselves were completely legitimate—surveying, map creation, and 3D terrain modeling for businesses.
The project was legal, and the database was also legal, so we decided to take on the task.
In principle, there is nothing magical about mailings, but the result is determined by the algorithm of actions. There are quite a few preparatory steps, which is why companies rarely do everything on their own: they often entrust this to agencies or hire a separate specialist.
Domain checkup
The client had a domain, but it was primarily configured for the website. To prevent mailings from ending up in spam, basic technical configuration of the domain was required.
We used the MXToolbox service and configured the following records on the domain:
- SPF (Sender Policy Framework) — allows you to specify a list of servers through which emails are sent from the domain. The list is specified in a TXT record associated with the domain.
- DKIM (DomainKeys Identified Mail) — an authentication technology that confirms the sender's domain through a cryptographic signature. The signature is verified on the recipient's side, which helps build the domain's reputation in “white” or “black” lists.
We created all the necessary DNS records, verified their correctness, and tested the deliverability of emails to free Gmail accounts and corporate mailboxes that we had.
Next, we chose a mailing service — MailChimp. It is important to note that the service must also be trustworthy, not “left-wing.”
Another important addition: when creating an email template, you need to include a prominent “Unsubscribe” button. And you need to actually process such requests. In our case, we segmented those who unsubscribed separately and did not send them any more emails. This is critically important for maintaining the reputation of the domain, especially when working with cold databases, if you plan to send mailings for a long time and systematically.
Working with the database, setting up chains, and designing templates for mailings
After technical preparation, we agreed with the client on the structure of the mailings: what exactly to send, to whom to send it, and how to segment the database. The plan was as follows:
1. The first letter was a cold call. It introduced the company, its services, competitive advantages, and benefits for the client.
2. The second letter was for those who did not respond. A few days later, we sent a reminder about the offer.
3. The third letter was sent a few weeks later. It emphasized urgency: “We have a few slots left, we are planning work for the next quarter, and we are offering a discount to everyone who responds within a week.”
For those who responded to the letters, a separate scenario was set up: an automatic letter was sent immediately with thanks for contacting us and an attached presentation of services. Further, this contact moved on to “live” communication between the account manager and the potential customer.
An important part of the process was normalizing the database. Let's say we have a file with email addresses — it needs to be thoroughly checked to avoid errors. It often happens that a name is present in one place but not in another; sometimes the first and last names are indicated, and sometimes only the first name; there may be incorrect email addresses, incorrect formatting, or garbled encoding. It is advisable to go through the file several times.
For example, you can sort the list by name, look at the length of the lines, the presence of spaces, and errors. Then sort by surname and also check the data. Errors are corrected using formulas so that the first name is in the first name field and the last name is in the last name field. This allows for a more personalized mailing, especially if the template uses dynamic inserts: the recipient's first name, last name, or position (if any). Similarly, phone numbers and other data are converted to a uniform format.
Domain checkup
Database verification
Email validation
Domain warm-up
Why is this important? If there is a high percentage of incorrect addresses in the database, emails will start bouncing back en masse as undeliverable. This will negatively affect the domain rating and sender reputation, which will further complicate deliverability even for correct addresses.
After cleaning the file, the next step is to check the email addresses themselves for existence. Sometimes part of the database is outdated: mailboxes have long been deleted, or users simply left them with a typo.
For such a check, we used the ZeroBounce service. We purchased a paid plan for a limited period (since the client had a single database and did not require a permanent subscription). We ran the entire database and weeded out addresses that no longer exist.
ZeroBounce also highlights several categories of addresses:
- General (info@, admin@, etc.) — such mailings are more likely to end up in spam, as the address is impersonal and is used as a “trap” for emails.
- Frequent spam complainers — the owners of such mailboxes regularly mark emails as unwanted. Sending emails to them increases the risk of damaging the domain's reputation.
- Frequent unsubscribers — mailings to such addresses lead to a rapid increase in the number of unsubscribes.
Segmentation helped us set the right priorities.
We started sending emails gradually, starting with a small batch of about 200 addresses. First, we selected the “best” emails — those that exist, do not belong to risky categories, and are not anonymous.
In the first few weeks, we sent small volumes of emails — 100–200 per day. After that, we gradually increased the volume of mailings to 500 and even 1,000 emails per day. This gradual “warm-up” allowed the domain to avoid spam filters and generate positive deliverability statistics.
Only at the final stage did we send letters to addresses marked as more risky: those belonging to people who often unsubscribe or complain about spam, as well as to general mailboxes such as info@. This was done after the domain had gained a certain level of trust and the first mailings had been successful.
Key points
To launch a cold mailing campaign correctly, you need to follow several mandatory steps:
1. Check and configure the domain (SPF, DKIM, DMARC).
2. Clean up the database: remove empty entries and errors, normalize the format.
3. Check the existence of email addresses through a validation service.
4. Prepare email templates and a service for sending.
5. Start sending from small segments (100–200 emails per day) and gradually increase the volume.
This algorithm allows you to maintain your domain's reputation, achieve maximum email deliverability, and use even a cold database as effectively as possible.
If you need help checking your domain and its trustworthiness before you start sending emails, or setting up templates and launching your first campaign, we are always here to help.
