The guide to email deliverability

What is email deliverability

This guide on email deliverability is an adaptation of the emailing course co-written by Marion Duchatelet and Jonathan Loriaux for OpenClassRoom, which itself is an offshoot of the email deliverability training that we have developed in recent years.

Before we start, let's define! What is deliverability? Deliverability is the science, or art, of delivering an email to the inbox and not to the spam box.

As we'll see in a moment in this guide, this is still a significant challenge, and you have all the keys. Unfortunately, deliverability is a victim of many myths which sometimes have existed for many years... when they have not been necessary for a long time.

As we will see in this guide, it is not your sending platform that makes good email deliverability. It is your practices and the respect you have for the recipients of your messages.

Enjoy reading, and don't hesitate to let us know if you have any questions or need a hand!

Why are your emails going to spam?

Identify what is spam

There are several approaches to understanding what spam is.

The law

The first approach is legal. In the eyes of the law, spam can be considered as a message sent without the recipient's consent. But in the end, if consent is a crucial element in email marketing, it is not the most important element in deliverability. We'll see about that right away.

The perception of the recipient

The perception of the recipient is the most important approach to deliverability. You, me, and all the recipients of e-mails, we have all been irritated by an e-mail at one time or another. Because it did not arrive at the right time, because the message was out of place, because we thought it was aggressive, because we receive too many...

This notion of irritation is absolutely central. What spam filters are trying to do is to understand and guess whether or not an advertiser's messages are considered irritating by the recipients.

Fight against spam

If you want to understand how anti-spam works (and also why it concerns you, even if you think you have good practices)It is important to understand why it has become extremely important to fight against spam.

In recent years, it is estimated that the proportion of spam (in relation to the total number of e-mails exchanged) is about 45 % (see study below).

Source : Satista 2019
Source : Satista 2019

This with very important fluctuations from one month to another... and we can notice that we come back from very far. In the golden age of spam, between 2007 and 2010, we were at more than 80 % of spam.

Storing this spam is expensive, fighting this spam so that it does not invade the e-mail boxes of individuals and companies is very expensive.

Assimilate the challenge of messaging and the concept of "graymail

If we were to define a "scale" of e-mail respectability, we would have on one side the interpersonal e-mails, those you and I exchange with our family, friends, colleagues, administration. On the other side of the scale, we would find criminal messages: phishing, extortion, scams, illegal products.

Commercial emails are at a crossroads.

The purpose of messaging is twofold:

  1. Protecting yourself to avoid being blamed: criminal emails!
  2. Make their messaging systems free of irritants for their users!

For this reason, filters go much further than the simple legal definition of spam, and work on the "irritation" dimension. Some of them, like Vade Secure, which we will talk about later, have invented the notion of graymail.

Source : Vade Secure
Source : Vade Secure

Graymail is all commercial and advertising messages, which are not really spam, but which are not always desired by the recipients.

These messages, depending on their reputation (a concept we will see in the next chapter) will sometimes end up in spam, sometimes in the inbox, or increasingly in a box dedicated to commercial and lower priority messages.

This particular treatment of commercial emails is a great challenge for companies, since it is estimated that about 20 % of the so-called legitimate end up in spam, or are "lost".

Source: Global deliverability benchmark 2019
In purple, the messages arriving in the mailbox. In yellow, those arriving in the spam box and in red those that have disappeared. Source: Global deliverability benchmark 2019

The concept of "reputation" in email deliverability

As we have seen in the previous chapter, messaging systems aim to eliminate all irritants for the recipients of the messages. They will therefore use all available clues to understand which messages are dangerous and/or irritating. These indices will allow them to build reputation scores.

These reputation scores are built either directly by the messaging companies or by companies specialized in the fight against spam. It is important to keep in mind that each actor has its own recipe for building your sender reputation.

As in real life, this reputation "sticks to you", i.e. it is historicized; your past actions, good or bad, will continue to follow you. And like in real life, it is easier to lose your reputation... than to build it.

Identify the elements influencing your reputation

As we have just said, the reputation is built differently by each mailbox and by each spam filter. In fact, for each mailbox, you will often have several reputations:

  • on your sending IP addresses : these are the "physical" addresses of your email server;
  • on your domain names The domain name of the email address used to send your messages, or the domain names contained in your emails (hosting images, links...) ;
  • the footprint of your emails Spam filters now have the means to identify the footprint of an email campaign. That is, to recognize a similar message when it is sent from several different environments, and to track its reputation.

The universal signals of reputation

As we have just seen, messengers use the clues that may be presented to them in order to define whether a message is wanted or not. We can classify these cues into two main categories: universal indicesused by almost all anti-spam and email systems, and behavioral indicators (which we will see in a moment).

Spam complaints

Probably the worst signal!

Report an email with Gmail
Report an email with Gmail

When a recipient decides to classify a message as spam, the consequences are significant for the deliverability of the sender of the message. If many recipients generate a spam complaint, the consequences are disastrous.

Incorrect addresses

Sending your emails to many wrong addresses proves that you have poor data hygiene, or that you are exploiting old email lists. This will result in high bounce rates, and a significant impact on your sender reputation.

Spamtraps

A spamtrap is an email address that is used by email service providers or the anti-spam industry to trap spammers. Spamtrap addresses may have been created from scratch and scattered across the Internet, or they may be old addresses used by individuals and abandoned by them. After several months of inactivity, email companies can recycle these abandoned addresses into spamtraps.

Behavioral signals

Unlike the 3 universal signals we have just seen, behavioral signals are not used by all messengersThis is because they are not very sophisticated, or because they do not have the possibility to record these signals, or, as in France, because the legislation does not allow to watch what the user does in his mailbox.

On the French market, it will be mainly the American webmails that will be affected by these practices, and especially Gmail and Microsoft... which often represent at least 50 % of the databases.

When we say that operators use all possible cues, with behavioral cues it makes total sense. Among the behavioral cues, we will find:

  • opening
  • deletion
  • click
  • ranking/tag
  • transfer
  • answer
  • unsubscribe

To give a concrete example, if a majority of recipients delete your e-mails without even having taken the time to open them, this is a strong signal that they are of no interest... to anyone. Your reputation will therefore be greatly affected.

https://user.oc-static.com/upload/2020/08/20/15979412220154_image40-2.png
In red, negative signals, in green positive signals, in top universal signals, in bottom behavioral signals.

Calculate your reputation

As we have seen, there is no such thing as "one" reputation, but several scores. So there is no "one" universal reputation that would allow you to know where you stand. While most email companies and spam filters are very secretive about the reputation of senders, it is still possible to get information about your reputation level from some:

Work on your reputation email

You will have access to a detailed article on the subject of deliverability best practices a little further on. But keep in mind that optimizing deliverability is mainly about decreasing negative signals and... increasing positive ones.

In fact, if you look closely at the summary chart above, we are largely facing a marketing rather than a technical challenge. Deliverability is more about recipient satisfaction and engagement than on the quality of the sending platform, or the technical configuration.

How to measure the deliverability of your emails?

Before attempting to measure your deliverability, it may be interesting to understand what you are trying to discover behind these measurement points. What we want to avoid is to have an "incident".

Deliverability incidents can be classified into 4 categories:

  • poor categorization of emails In most mailboxes, some emails are classified as important or priority, while others will be stored in one or more secondary boxes (for example the Promotion box in Gmail). In some cases, your emails may be delivered to the wrong box. This will be the case for example for purchase confirmations, which should normally arrive in the priority box or the promo box;
  • Spamming This is probably the most common incident. Instead of your emails being delivered to your inbox, they are categorized as spam by the email system;
  • blocking at the entrance Spamming: a more serious situation than spamming, when your emails are simply rejected. Not even accepted by the destination server. In this case, you should normally receive a "block bounce" error;
  • black list If your IP address or your domain name is referenced by a "blacklist", this can potentially lead to a complete blocking of your emails, regardless of the email service. We'll come back to blacklists in chapter 5, and tell you which ones are of critical importance.

Understanding the concept of "destination

Before presenting the main deliverability performance indicators, it is important to understand the concept of destination in deliverability.

This notion tries to group e-mail addresses according to the e-mail operator. To make it simple, let's go through a short example. Let's imagine a list of 10 email addresses (all these addresses are fictitious for the exercise) :

  1. j.dupont@live.fr
  2. marievv@orange.fr
  3. lukaszbugi@outlook.com
  4. omaris@wanadoo.fr
  5. gwendu@yahoo.fr
  6. leilarepo@neuf.fr
  7. cocotteminute@msn.com
  8. justintime@sfr.fr
  9. gigilamo@hotmail.com
  10. cocochcha@ymail.com

On these 10 e-mail addresses, we see that we have 10 different domains. On the other hand, some areas belong to the same companies, and therefore use the same spam filters and reputation scores:

  • Microsoft: live.fr, outlook.fr, msn.com, hotmail.com
  • Orange: orange.fr, wanadoo.fr
  • Yahoo! : yahoo.fr, ymail.com
  • SFR: neuf.fr, sfr.fr

We will see in a moment what all this can be used for!

Analyze delivery performance

Bounce rate

Bounces, along with opens, are one of the most valuable indicators of deliverability.

They may indicate a blockage due to a bad reputation or a too high sending volume, or a poor basic quality (targeting of inactive addresses or poorly qualified/non-opt-in addresses).

Bounce rate calculation = number of bounces/number of emails sent

Having an overview of the bounce rates per destination can help you to know very quickly when and on which destination you have an "entry block".

https://user.oc-static.com/upload/2020/08/09/15969780246653_5.3.1.png

The opening rate by destination

If there are any other indicators that may tell you of an incident (the bounce rate, for example)the opening rate is the most important and reliable.

If overnight your open rate to Microsoft (Hotmail, Live, MSN, Outlook) drop by half, while the other destinations remain stable, you have a delivery problem at Microsoft.

Opening rate over a week
Opening rate over a week

Need help?

Reading a guide can not make everything perfect. Maybe the best thing to do is to ask us. No ?

Adopt the best practices of deliverability

Deliverability is a clever mix between good technical configuration and quality of interactions with your recipients.

We have chosen to focus on 5 good practices that will allow you to have a good basis, when, later, you will be confronted with a need to optimize your deliverability.

The first point is resolutely technical, the next two cross technical and marketing practice, the last two are clearly marketing oriented... everything is always intertwined in deliverability!

Authenticate your emails correctly

Email as we know it today was born in 1982 (well, in fact, it is the SMTP protocol who was born in 1982, but we won't go into details at this point). 1982 was an eternity ago in terms of technology. Just think, it's 8 years before the invention of the World Wide Web. At the time, Internet users were still only a handful, and no one was thinking at the time of the different uses to which this technology could be put.

So originally, anyone could take the identity of anyone to send an e-mail... without being in violation of the SMTP protocol rules.

Almost 40 years later, we immediately realize that if anyone can send an e-mail in the name of Paypal... it can do a lot of damage! That's why throughout the history of e-mail, several successive technologies have been used to reinforce the identity of the sender in order to avoid abuse. This is what we call authentication.

Normally, modern routers will not let you send a single email from their platform without the 3 main authentication techniques being properly configured by you. These 3 techniques require adding information to your DNS server (the directory that links a domain name to the physical address of a server).

SPF authentication

Sender Policy FrameworkSPF is an authentication standard to reduce the risk of identity theft. With SPF, you declare the IP addresses and servers that are allowed to send email for a given domain.

For example: v=spf1 a mx include:spf.mailjet.com include:sendgrid.net include:spf.protection.outlook.com ip4:204.128.192.36 ~all

The domain name with this SPF registration would authorize Mailjet, Sengrid, Outlook.com and another IP address to send e-mails on its behalf.

Schematic presentation of SPF authentication
Schematic presentation of SPF authentication

DKIM authentication

DomainKeys Identified Mail is an authentication standard that reduces the risk of identity theft.

DKIM uses a signature system to ensure that the e-mail sent has not been altered during its transport to the recipient.

Schematic presentation of DKIM authentication
Schematic presentation of DKIM authentication

DMARC authentication

DMARC (Domain-based Message Authentication, Reporting and Conformance) is a technology for monitoring problems related to email authentication and domain name abuse (phishing).

DMARC builds on the SPF and DKIM authentication systems to derive two elements:

  • the ability to tell the receiving server what to do with emails that are not SPF and DKIM compliant;
  • receive alerts in case of non-compliance or lack of authentication.
Schematic presentation of DMARC authentication
Schematic presentation of DMARC authentication

Make your sender identity consistent

There must be consistency in the policy for using domain names. Spammers try to hide, so you need to be as transparent as possible:

  • use a domain name(s) that matches your brand;
  • use this domain name (or a sub-domain) all over your email;
  • ban the use of external domain names (your router's domain whenever possible, "cryptic" domains like dr753gv.com...)..

Good practice example:

  • Sending address: yesreply@badsender.com
  • Reply address: service-client@badsender.com
  • Link tracking domain: link.badsender.com
  • Image hosting domain: images.badsender.com
  • Technical sending domain: em.badsender.com

To ensure as much transparency as possible, also make sure to:

  • Ban the use of noreply@ or nepasrepondre@ sender addresses, and use a real address in order to be able to react to the answers on your emails (there will be few, but there will be some) ;
  • say who you are in the footer of your email;
  • add a link to your legal notice.

Choose between a dedicated or shared IP

The routing architecture (which domains and which IPs for which type of email) is an important element of deliverability.

Depending on your needs, you can establish your routing strategy according to several criteria.

These criteria revolve mainly around the notion of "risk", a notion that can take on two very different meanings:

  • some of your traffic is risky Your business or your marketing practices, on a part of your traffic, make that you have a bad reputation on certain types of sending. You risk infecting the other types;
  • some of your traffic is crucial to your business For example, if you have a part of your traffic that you cannot afford to have deliverability incidents, certain types of sending are highly strategic for you. In this case, you will probably want to isolate them to protect them.

Here is a table summarizing the advantages and disadvantages according to the type of IP address:

Dedicated IPsShared IP
+Dedicated reputation: Possibility to subscribe to an accreditation service- Less constraint on volumes and frequency of sending
- All clients configured on this pool
shared work to build the
reputation
- IP "heating" process
- Daily registered mail
- Constant volume required
- A client could negatively impact the reputation of one or all IPs
- No possibility to subscribe to an accreditation program

Contrary to what many people think, having a dedicated IP is not the grail of deliverability. On the contrary, if you make this choice, you will be solely responsible for building a reputation on it.

Collecting quality addresses

There are many methods to collect e-mail addresses. And some are more qualitative than others. By collecting quality addresses, you will be able to decrease complaints, reduce spamtraps, increase opens, increase clicks, stabilize unsubscribes... all results that are good for your sender reputation.

Encourage quality collection sources

Give preference to addresses collected when your customers place orders, spontaneous subscriptions to your newsletter, content downloads... On the other hand, ban the purchase and exchange of e-mail address files.

Keep your collection sources

If you collect your addresses from different sources, keep them in your database. This will allow you to check the performance of each of them and, if necessary, to discard the less qualitative ones and give preference to those with the best results.

https://user.oc-static.com/upload/2020/08/09/15969804997276_5.4.5.png

Consider double optin

The double optin consists, after a registration, in sending a confirmation e-mail asking for a validation of this registration. If the recipient does not validate their subscription... they are simply not registered and do not receive your messages.

This type of mechanics is mandatory in some countries (in Germany, for example). It ensures that the person who validated a form is who they say they are. Double optin is also a very good way to protect yourself from certain types of spamtraps.

Target mainly active recipients

"Permission Marketing is the privilege (not the right) of delivering anticipated, personal and relevant messages to people who actually want to get them."

Seth Godin

We won't go through the list of all the signals that positively influence reputation. But one thing is clear, you have to meet their expectations. On the other hand, once you don't meet their expectations... they get bored and gradually stop responding to your messages.

Inactives are a major threat to your reputation. They can both host spamtraps, but also bring down the overall engagement metrics.

Rather than inactive, they could be called "disengaged".

Properly managing inactives has become the main objective of any reputation optimization. Mechanically, by reducing the portion of inactives in your mailings, your reputation will improve (well, obviously, it will take some time) mechanically, if the inactives increase, your reputation will start to decrease.

Very often, it is arbitrarily defined that a recipient who has not opened or clicked on one of your e-mails can be considered as inactive. But this is arbitrary. Regularly, you will have to limit yourself to openers/clickers of less than 3 or 4 months to raise a reputation to the point of collapse. On the other hand, in certain sectors of activity, these limits are impossible to respect. In this case, the limit will be around 12 months.

Understand the differences between messengers

As we have already seen, each email service uses its own mix of technologies to filter the messages it receives. This obviously means, unfortunately for us, that there is no single recipe for deliverability... but a multitude of recipes!

It is therefore important to know, even superficially, what the major differences are between these messaging systems, where to go for information, and where to go for help.

Identify the main blacklists

black listA blacklist, when it comes to sending e-mails, is a list that identifies senders who have practices that are similar to spam, or worse, to cybercrime. There are several types of blacklists, some of which list IP addresses, others domain names.

If there are several hundred blacklists on the Internet, only a handful of them have a real impact on your daily deliverability. The most powerful blacklist is Spamhaus. If you are "listed" on Spamhaus, it is very likely that your emails will be blocked by almost all the email systems in the world.

Major IP-based blacklists:

  • Spamhaus : Spamhaus is the largest anti-spam organization (many ISPs/webmails use its blacklists). Its interface allows you to check if your IPs and domains are on its list;
  • SpamCop : SpamCop is a free service (provided by Cisco) which allows people to submit spam (SpamCop will send the sender of the message an Abuse complaint). SpamCop has its own network of spamtraps;
  • SORBS SORBS is a blacklist owned by Proofpoint; it lists more than 12 million servers known to send spam;
  • Barracuda BarracudaCentral is the blacklist managed by the Barracuda company. It is available free of charge, provided that its users register.

Main blacklists based on domains :

  • SpamHaus See above;
  • SURBL SURBL is a list of websites that have appeared in messages considered as spam. It is therefore not a list of e-mail senders;
  • URIBL URIBL lists domain names that have appeared in the body of e-mails considered as spam;
  • Evaluated Invalidate list of domain names that have appeared in the body of e-mails considered as spam.

There are many services that allow you to check whether a domain or IP is in a blacklist or not; this is the case, for example, and free of charge, of Multirbl.

Optimize your deliverability according to your email

For each messenger, we'll give you a generic framework, tell you if the messenger shares reputation information with email senders, and offer some links.

Microsoft

When we talk about Microsoft, we are mainly talking about the public messagingThis is in contrast to Office365, which is the messaging system that Microsoft makes available to businesses. Microsoft uses behavioral data but also user panels in addition to the universal signals, in order to build up one's sreputation core. This is mainly based on the reputation at the IP level.

Gmail

When we talk about Google as a destination, we are talking about both Gmail for the general public, but also to G-Suite for businesses. Google makes extensive use of behavioral data in the construction of its reputation scores. These are mainly based on the reputation of the domain and IP addresses. Message footprints are also very important.

Yahoo!

The Yahoo! label now includes the various companies in the Verizon Mediaas AOL and Yahoo! These companies use behavioral data and place great importance on the quality of the links contained in the e-mails.

Orange

When we talk about Orange, we mean the services Orange consumer e-mailThese are mainly used by the company's customers. Orange mainly uses universal signals to build your reputation.

Free

Free groups together the webmails of the operator and of all the companies it has acquired in its history. Free's filters are mainly based on volumes of complaintss, the spamtraps and the address quality.

Laposte.net

Webmail owned by La Poste. Its filters are mainly based on the volume of complaints and on the quality of the addresses.

SFR

SFR groups together the webmails of the operator and all the companies it has acquired in its history. Its filters are mainly based on the volume of complaints and on the address quality.

  • Examples of domains: sfr.fr / neuf.fr / 9online.fr / 9business.fr / cegetel.net / club-internet.fr / cario.fr.
  • Access to reputation information: via Signal Spam or your router.
  • Documentation and help center: none, unfortunately.
  • Support: cellule-abuse@sfr.fr