Why email marketing service actually fights spam, and how it works

Image: Unsplash

On the habra, and indeed on the Internet, it is widely believed that everyone who engages in mailing lists from marketers to developers of the relevant systems is, without exception, spammers who send unsolicited emails.

But in reality this is far from the case, and a good email newsletter service is always an element of the global anti-spam system, which just protects ordinary users from receiving unwanted emails.

At DashaMail, we also devote a lot of effort to fighting spam and have developed a number of tools that can reliably filter spam mailings that someone might try to send through our service.

Automatic scoring

Let's start by describing an automated system that calculates user reliability ratings. If the sender is recognized as trustworthy, then its messages are sent immediately, and potential violators pass more stringent filters.

The algorithm is simple: each user can be assigned a rating from 0 to 10. Reliability groups are distributed as follows:

• from 0 to 2 points – beginners (newbie),
• from 3 to 5 points – advanced users (fine),
• from 6 to 8 points – reliable users (good),
• from 9 to 10 points – senders with the maximum trust (trust).

Depending on which group the user belongs to, his mailings will be sent through a specific pool of IP addresses. That is, mailings of "good" clients are sent only from addresses that are used for letters from other reliable senders – this allows you not to spoil each other's reputation in the eyes of mail services.

Conversely, “bad” customers can send their letters only from addresses used by the same potential violators.

How a reliability rating is formed

Let's talk about how you can improve or worsen your rating within the system. For each successful mailing – that is, not blocked by spam filters of mail providers – the user receives points. But they can be lost. Here are some factors that can contribute to this:

• The recipients click the Spam button.
• The percentage of “hard returns” exceeds the system average.
• A high percentage of unsubscribes that exceeds the average for DashaMail.

Consider each item in more detail.

Clicking the Spam Button

Users of any modern mail program or service today can signal that they have received spam. The corresponding buttons help to train spam filters of these mail providers, but thanks to the FBL (Feedback Loop) mechanism, we can also receive this data.

In RuNet, FBL from Mail.ru, which is provided as part of the Postmaster service, is most often used. Thanks to him, we can receive information that a certain subscriber has clicked the Spam button and unsubscribe him from the newsletter. And of course, this is a clear signal of a problem.

It is clear that there is always the possibility of erroneous clicks, sometimes people just forget that they signed up for the newsletter, in some cases they just press the button because of a bad mood. Therefore, we allow a certain percentage of such clicks, but if it is large enough – this will affect the rating of the sender.

Hard Returns

When a message sent from our IP addresses is returned back with a note like “User does not exist”, “User is blocked”, this is a hard return. A letter to this address cannot be delivered in principle.

A large number of such returns exceeding the average values ​​set by mail providers is a clear signal about the high probability of using the purchased, not self-assembled subscriber base.

Here, too, you cannot cut off your shoulders, because email databases can have different tasks. For example, sometimes addresses are collected into a single database for subsequent reactivation, that is, contacts with users who once used your products and services for a long time, or from paper questionnaires, where there is a high probability of errors and typos. In this case, we do not block the account, but we require confirmation of the legitimacy of the database and we recommend that you first clean the database using the email validator built into the service.

User complaints and unsubscribes

Of course, all the factors that take into account the system have their own weight. Of course, the most severe sanctions – up to zeroing the rating – can be imposed after direct complaints by users of spam to the data center, to us or the mail provider.

In this case, we require the client to provide evidence of the legality of the subscription, which may include server logs, screenshots of databases, mailing policies that the subscriber agreed to, the questionnaire completed by him, etc. If within 24 hours we do not receive the required information, then the sender’s account will be blocked.

The account may also be blocked based on the work of spam filters of mail providers. Upon receipt of such a complaint, you will need to provide evidence of the legitimacy of the newsletters and fill out the form of the mail provider. If an error occurs, the newsletters will be unblocked.

In addition, we receive signals from SpamCop, CloudMark and other anti-spam systems, as well as our IPS: we process all incoming signals, which allows us to quickly block unscrupulous clients.

In general, if everything goes well, the client makes high-quality mailings, which the recipients hardly complain about, then just a few sendings will be enough to get the good class. In this case, the moderation of mailings will occur automatically.

Analysis of content and address bases

It is known that spam filters of mail providers also analyze the content of the content of letters. It is also implemented on our side based on the SpamAsassin rating (which, of course, required a lot of refinement and training in connection with the use of the Cyrillic alphabet) and verification of the content of stop words. A similar analysis is used by the leading Runet spam filters, which have a simply phenomenal database of spam content patterns. Spammers constantly mimic their behavior and appearance as a reliable sender, but we try to identify the same g * vna spoon in a barrel of honey before sending the newsletter.

We also collect various “bases for sale”, “leads for sale” and any other freely and for money distributed address databases of subscribers. This allows you to identify knowingly bad address databases collected without consent.

Manual moderation

Yes, we have such a way of fighting spam. Until the user of the system has earned the status of reliable, his mailings will be moderated manually.

Thus, we find content that violates the laws, and we can send newsletters for revision so that they meet all the design requirements. Manual moderation is quite a resource-intensive process, because through DashaMail passes up to 1000 mailings per day, but this is also the most effective way to check. In addition, manual moderation allows you to teach a new user to make the right letters and prevent the occurrence of problems with mail providers and delivery.

Conclusion

As you can see, we, as the developers of the mailing list service, pay a lot of attention to the fight against spam. To do this, we have created a whole combined system of manual and automatic moderation of newsletters. This is our contribution to the protection of Internet users from spam.

To keep abreast of current trends in email marketing in Russia, to receive useful life hacks and our materials – subscribe to the DashaMail Facebook page and read our blog.