A hard bounce is an email that has failed to deliver for permanent reasons, meaning that the recipient's address is invalid (either because the domain name is incorrect, isn't real, or the recipient is unknown.) When a hard bounce occurs, the sender receives an error notification.

These error notifications may have one of these SMTP codes: 451, 550, 554, 571 (depending on the recipient’s mail server configuration). We distinguish between hard and soft bounces. Soft bounces are emails that fail to deliver for temporary reasons.

It means that the email address was valid and the email message reached the recipient’s mail server. However, common reasons it bounced back include:

• The mailbox was full (the user is over their quota)
• The server was down
• The message was too large for the recipient’s inbox
• The sender is categorized as “spam” by the recipient’s spam filter

When testing deliverability of email addresses, we must send a lot of messages. These messages cannot originate from the same mailbox. If we did so, the mailbox would be soon categorized as “spam sender”, due to the high number of hard bounces.

Hard bounces damage the sender’s reputation. Once a mailbox is categorized as “spam sender”, it stops receiving all the hard bounces. More and more hard bounce notifications are missed. Therefore, it is necessary to use only “virgin” mailboxes with a fresh clean reputation.

We send 50 messages at once, and then we create a new mailbox to send the next 50 messages. Etc. Even doing so, there are a few hard bounce notifications that we miss. They are quite rare, but still. Some recipient’s spam filters reject the messages originating from newly created mailboxes.

To catch those notifications, we should warm up the mailboxes first, to strengthen our sender reputation. We have decided not to do it since it requires a lot of effort for a very small increase of accuracy (<1%).

Our experiment was conducted over a 2-week timeframe. We tried to minimize this timeframe as much as possible. As time goes by, more and more people move to a new company. Therefore, it would be unfair to compare the number of emails found by different solutions if there was a significant time duration in between.

We tried not to add too many biases. Obviously, there are still a few. First of all, the test file is built using Linkedin, while some users might look for prospects outside Linkedin sometimes. Second, the selected geographical areas (France, USA, Germany, UAE) do not include Asia, South America and Africa.

No. If you reproduce the experiment, you won't get the exact same results because time goes by. Each month that separates your experiment reproduction from ours, more and more people move to a different company. By doing so, they might join a company whose mail server is configured in a manner that makes it harder to find email addresses. This will cause slight differences in the results. However, the overall ranking and the orders of magnitude should not change significantly.

Each year, we create a new test file of 300 prospects. Each test file is unique. The net email discovery rates are specific to each year. They are not an absolute value of performance per se. However, the ranking of all solutions can be compared between years. This ranking is not supposed to change, unless a solution's performance increases or decreases over time.