Aaaaaahhhhh, the killer question (literally if we're talking about your sender reputation). It's funny, because I don't think I've ever covered this topic here, even though it's a question that comes up regularly.
If I take the time to write an article on the subject, it's because I've been asked this question very regularly lately and just this morning in the context of a deliverability audit.
So the question was, "But why do you want us to lower the number of IPs we shoot with, when we've always been told that the more IPs we have, the more effective it is?"
Well, first of all, it is true that there was a time when shooting from many IPs allowed to be less fooled. This was true for example because the spam complaint thresholds before blocking at some ISPs were absolute volumes. Today, because of this practice, everyone has switched to %age thresholds. So it doesn't work anymore.
Then there may be some good reasons to have many IP addresses. In general, this reason is related to the speed of sending. When you are in the private sale sector for example, it is necessary that all the emails arrive in a very short time window (1/4 of an hour, half an hour, ...) and we all know that sending emails is not done "in real time".
The main reason why you should not exaggerate the number of IP addresses, is also related to the sampling that operators do and that will allow you to increase your reputation according to the behavior of your recipients. This problem can appear on two different occasions:
- Migration to a new router: In this case, if you have brand new IPs, you will have to perform a WarmUP of them. In this case, webmails like Hotmail will deliver all your emails as spam and let only a few copies through in order to measure the reaction of your recipients. If the reactions are positive, your emails will progressively go to the inbox until the end of the WarmUP. If you ever send from too many IP addresses, the volume sent by each of them will be very low and the number of samples going to the inbox may be close to zero. Your Warmup may take a long time.
- Getting back on track after deliverability problems: It's a bit the same scenario as for the previous point. One of the most effective methods when you have decided to change your deliverability practices is to start a warmup from scratch, with very low volumes but hyper responsive recipients. This allows you to rebuild your reputation, but also to know at what level of activity the recipients have a negative impact on your deliverability. The concern here is exactly the same. A too low volume will arrive entirely in spam, and the samples delivered to the inbox will not be sufficient to get back to a satisfactory performance.
Do you know Snowshoe spamming? It is an attack carried out by spammers and consists of sending a very large number of emails in a very short time to a very very large number of different IP addresses. Obviously you wouldn't want to be compared to such a practice 🙂
Need help?
Reading content isn't everything. The best way is to talk to us.
Yes, but then, how many IPs do I need?
First of all, this is a method that I have been using for some time but which is not universally used. I don't claim to have the universal truth on the subject, so don't hesitate to give me your opinion on it.
The rule of thumb is to consider that an IP can swallow between 100,000 and 300,000 emails per hour (this is a range that can be wider, depending on your reputation and the number of destinations you address). If for you the speed of sending (I push the button and x minutes later my recipient should have received the email) is not paramount, I generally consider 250,000 as a good compromise.
So if you have 500,000 emails to send each day, with standard constraints, two IP addresses should be enough.
Where it gets a bit tricky is when your constraints are not standard, irregular frequencies, need to manage large peaks, ... there, you can try to work on your average emails sent each day versus the maximum volumes of your peaks. But if your averages are multiplied by more than 3 during your peaks, you'll have to ask yourself other questions, such as going back to shared IP, there, it's a case by case basis.
Don't forget to isolate your most risky/important emailing programs
Another essential point to consider is the separation of your different programs. I won't deal with this point in depth here, but when you separate a reactivation program, transactional emails and a marketing program, you obviously have to do this work separately, according to the parameters of these different programs.
Need to move forward on email deliverability issues?
10 réponses
There is no maximum per domain name?
Thomas > Can you repeat the question 🙂
To me, giving a global maximum per IP doesn't really make sense.
Be careful with the maximum number of emails sent by ISP
I'm surprised by your figures, I've always heard that you shouldn't exceed 50k emails per IP per day to stay within the limits, you can however go up to 80-90k if you have a very good reputation.
Hello
May I object?
Putting aside acquisition or affiliation mailings, I recommend reactivation via the same IPs and senders as loyalty programs.
If best practices are followed, the reactivation volume should not exceed 5 to 10% of the retention volume.
For example, for every 100,000 retention messages sent on D-Day (to recent actives), 5,000 reactivation messages (reactivation attempts) can be sent in their wake.
Separating the streams means having to maintain different reputations on very different programs.
But to build and maintain a reputation, you need volume, and consistency in volume (at least not too much inconsistency). It's difficult to have consistency on reactivation (except when you have a daily automatic program that runs continuously of course).
I think you have to dilute a little bit of "dangerous" (reactivation) with a lot of "over" (retention).
Your comments are welcome friends ...
Thibault
Florian > I don't have quite the same experience 🙂 But maybe we are dealing with very different types of traffic, between acquisition and retention there is necessarily a significant difference. To maintain reputation, you need traffic, if possible regular (and quality). The clients I work with who have the best reputation shoot between 2 and 4 million per day on only 4 to 5 IPs... and after a little work to stabilize, it works like a charm.
Then, as Thomas said, it depends on the distribution of the destinations. If you have an MX server that swallows more than 50% of your traffic, the numbers are bound to be very different.
Thibault > Actually we agree 🙂 What deserves to be really separated, what we can afford, is only the hyperqualitative and hypersensitive sendings (if the emails fall into spam, the SAV explodes 🙂 ).
Reactivation programs should be put with classic newsletter flows ... except at the beginning of their deployment if they are set up to save a situation close to disaster (unfortunately few advertisers set up reactivation/deactivation when everything is going well).
Hello,
thank you for the article.
You say that it is better to do a WarmUP of a new IP address than to rehabilitate the reputation of an old IP address once it has been tainted by bad practices.
Isn't it longer?
I hesitate between the 2 : go back up the hill or create a new IP...
Thank you for your advice.
Hi Caro. This article is from 2016... there are probably some elements that are not quite current anymore. Feel free to contact us if you want to tell us your specific case.
Hello,
Relatively "precise" question:
How many emails per hour can be sent with garbage IPs in B2C with ISPs (with French tendency)?
Let me explain:
With RGPD, the prospects of more than 3 years must be deleted (CQFD: inactive of more than 36 months having never bought products or services)
A service email will then be sent to them to inform them of the deletion of their account
=> this can be considered as reactivation BUT
1- they have never been active
2- they have not been solicited by email for at least 30 months (only the active ones of less than 6 months being shot)
The risk of spam trap being then extremely high compared to the usual, it is better to use its usual subdomain with the usual IPs (excellent reputation)
Or go for the classic "use of unheated IPs" and new subdomain... And then, how many emails can you send on unheated IPs to stay "under the radar" of the reputation system?