4 Factors For Deciding the Number of IP Addresses For Your Email Marketing Program

Knowing how many IP addresses to send from can be difficult without proper planning. The first thing you should look at is what’s available to you, as there may be some restrictions in place. Perhaps you only have a budget for 1 IP address. If you’re using an ESP, perhaps you only get 1 for transactional emails and 1 for marketing emails. There’s nothing wrong with these approaches, as long as you’re experiencing high inbox placement rates and high Sender Scores. However, if you feel that you can increase your ROI with more IPs, then you will need to make a business case. How? The answer has always been in the data.

Below are a few items to consider when evaluating the number of IPs required:

1. Risk – Depending on the riskiness of the email, there may be a need to have separate IP addresses. Use your inbox placement rates, Sender Scores, SNDS data, complaint rates, unknown user rates, spam traps, bounce logs, Certified reports, research papers, etc. to help gauge the risk level. For example, it is not uncommon for senders to tank their IP sending reputations with high complaint and unknown user rates by not removing their inactive subscribers. Yet, they still come to us to help them with their deliverability problems, even though they are not following best practices. Another example I see is when senders tarnish their email deliverability rates by deploying their “invite my friend” type of emails from their “good” IPs, even though it is a known high-risk list acquisition technique. As noted in our 2012 Sender Score Benchmark Report, social networking senders have the highest frequency of spam traps because they use the import address book method to grow their list file. In these types of situations, I recommend separate IP addresses so that the lower risk emails are not jeopardized from reaching the inbox.

2. Sending Reputation &Volume – A sender should also consider the volume of email it is sending to determine if it needs extra IP addresses. There is no hard and fast rule based on the sending volume alone. One sender may able to send more because of their positive sending reputation, while another sender may be rating limited because of their high unknown user and spam trap issues.

3. Time sensitive emails – A sender who needs to get out a large volume of emails in a short period of time may want to consider having additional IP addresses.  An example may include senders that offer flash sale promotions that are valid for only a few hours. Here are few ways to do the calculation:

Posted Hourly Rates – Identify the ISPs most important to you. Then, check their postmaster sites to determine if they posted their hourly rate limit. For example, Comcast can send up to 86,400 emails per hour when the IP Sender Score is between 86 to 100. If you have 3 million Comcast subscribers and you need to get your email out within an hour then you’ll need 4 IP addresses.
Log Files – Not all the ISPs post their hourly rates. In those types of situations, use your log file data to help determine your actual hourly rates by ISP. For example, say your logs show that you can send 2 million emails to Windows Live Hotmail within an hour, but you need to send 10 million emails within an hour. That means you would need 5 IP addresses to accomplish this goal.

4. Other Factors- More isn’t always better. You definitely don’t want to appear as if you’re using a snowshoeing technique, or hopping from one IP to another in order to game the system. That’s one of the first reasons to get you blacklisted. In addition, you should also collaborate with your ecommerce team in regards to timing. It’s not uncommon for an ecommerce site to go down as a result of too many subscribers clicking on an exclusive promotional offer all at the same time.

Whether a sender should have more than one IP address really depends on the senders’ own specific data and emailing objectives. Hopefully, these recommendations will lead you in the right direction. As a reminder, don’t make any decision without looking at the data first.

What other factors do you look at?