Lyris and Hotmail

We have been working on new ways of improving response rates from recipients with Microsoft email domains. We have had remarkable success in recent weeks, in particular, we have in some cases been able to improve the open rate for Microsoft domains by 50% (e.g. from 14% to 21%), through a series of techniques which I would like to share with you.

Checking your open rate by domain is a very good way of identifying ways of improving results from your mailing. We have noticed that Microsoft has recently been reacting very quickly to response rates (in particular the percentage of recipients who hit the Junk or 'mark as Junk' options) by first rate limiting (slowing down the speed they accept email from you) and then delivering the content to the junk folder, rather than the inbox. We have also seen more cases recently, where Microsoft has gone beyond this and actually blocked the mailing altogether. In these cases they send back either a soft or a hard bounce in the SMTP transaction. Microsoft has filters and response management tools which are watching the metrics of your mailings and taking decisions as to how to treat your email on a hour by hour basis.
Microsoft does give legitimate senders good feedback on how they are treating mail from your IP address. You can see this by signing up to the Microsoft SNDS program

Microsoft changes the way your mailing is treated over time, whilst your mailing is being sent. So, your mailing can be delivered to the inbox in your tests and in the first couple of hours and then, due to response rates, the rest of your mailing can be delivered to the junk folder. How can you tell if this is happening?

- Firstly, you can check your general open rate against previous mailings. If you have a consumer list, Microsoft domains will make up a large percentage of your list and so you will see a lower open rate.
- Secondly, you can check your open rate for hotmail and compare with previous mailings. You can do this by creating a segment of all hotmail recipients who have opened the email and dividing the total by the number of hotmail recipients who received the email (another segment). Alternatively you can use SQL queries for this or make a custom report. If you have any questions on this, please do contact me on andrew@lyris.co.uk.
- Thirdly, download the .csv file of the segment of hotmail recipients who have opened the email and check the alphabetical distribution of the first letter of recipient addresses. Listmanager starts sending at letter 'a' and goes on through the alphabet. If you find that you have a large percentage of hotmail openers with 'a' or 'b' at the beginning of their email address but very few with 's' or 'w', there is a strong likelihood that your mailing has been delivered to the inbox at first and then after the first couple of hours, delivered to the junk folder or blocked.

How do you optimise your chances of consistent delivery to the inbox at Microsoft Domains?

The method we have employed is to split a mailing into two; send first to all the recipients who have opened an email in the last 3 months, then send the rest of the mailing to recipients who have not opened in the last 3 months. This means that the responses that Microsoft sees in the first few hours of a mailing are overwhelmingly positive. The mailing therefore doesn't get assigned to the junk folder. This means that engaged recipients don't miss out on the mailing at the start and 'inactive' recipients are more likely to respond because they will actually 'see' the email in their inbox. This method also shows you how well your content is doing at re-engaging the 'inactive' members of your list. If you get an open rate of 4% on your inactive list, for example, that is a cause for celebration. This means thousands of people have opened the email who have not done so in the last 3 months.

I was reading Ben Chestnut's interesting post on the MailChimp blog about how Sending to old lists will kill your deliverability. Our approach, as outlined in the paragraph above, goes one step further than what Ben is suggesting and gives you a much better chance of improving response rates.