Email Marketing – Results From the Best Split Testing Campaigns
The best way to improve email marketing campaign results is to test and analyze campaign results for your specific audience. As an email marketing vendor, we analyzed the results of several thousand email campaigns to get a better perspective of what works, and what doesn’t work for business-to-business (“B2B”) customers. We then distilled this down into a set of split tests to demonstrate a handful of interesting best-practices for email marketing. The split rest tips are summarized in this article in a Q and A format. Most of the results may also apply to business-to-consumer campaigns; however the results here are specifically form reviewing B2B email marketing campaigns.
Q: I am running an email marketing split test. how many emails should I have for each email campaign variation in order to have meaningful data?
A: If you are comparing the results for multiple email campaign variations, then you should shoot for the greater of the following:
- A test that will result in at least 100 data points per variation. For example if you typically see a 2% click rate and are comparing click rates in your email marketing results, then each variation should be sent to at least 100 / 2% = 5,000. If you have 4 variations, then you should send to at least 20,000 total contacts for meaningful results.
- Or, if you have a large list and are following best practices and varying only one item per variation, your total sample size should be 10% – 15% of the total contacts in the list.
Q: I have an email campaign and I want to test two different types of changes. How many changes can I include in each variation?
A: For basic split testing you should limit each variation to a single change. Otherwise, if you make multiple changes between variations, you will not be able to determine *which change* caused the different in response rates. On the other hand, if time is limited and perhaps you have created 2 or 3 different email campaign variations and want to see which performs best, then the versions can have multiple differences.
Q: How do people respond to attachments – how do open rates with emails using attachments compare with emails not using attachments?
A: Most spam systems will perform a virus scan on attachments, and many spam firewalls are set by default to quarantine attachments of most types, other than.PDF documents. Email campaign split tests demonstrated for us that it is better to host the attachment on a site, and include a link to the document instead of adding the document as an attachment. If you do opt for an attachment, stay away from large attachments (more than 2MB); PDF documents seem to get through ok.
Q: When sending an email marketing campaign, is it better to use a ‘send from / reply to’ address such as ‘sales [at ] mycompany [dot] com’ or ‘myname [at] company [dot] com’?
A: In most cases, for B2B companies, you will see better results by using a specific contact person (yourself for example) vs. a general address like ‘support@’ or ‘sales@’. Keep in mind that your recipients take only a second or two to decide whether or not to open your email. If they do not recognize your company, they are likely to skip over the email. Also, general addresses are less personal, which reduces open rates.
Actual results. By using several different email marketing split tests, results indicate that using a specific personal name vs. a general email address can improve the net open rate by 15% – 35%.
Q: What are the best days to send our email marketing campaigns?
A: Based on our data we consistently find the following:
- Business to Business (B2B). Scheduling for delivery on Tuesday Wednesday and Thursday early morning are best. Peak open times are between 5 and 7am local time.
- For B2C recipients (specifically, when sending to a contact list with a high concentration of consumer addresses such has @yahoo.com, @aol.com etc), we find that peak open times are commonly Saturday morning, Saturday evening and Sunday evening.
- However, as more and more companies focus on these times, you may find that you get very good responses by sending at different times – when overall email traffic is lower. The best bet is to test your campaign timing with your audience.
Q: What is a reasonable length for an effective email marketing campaign?
The results depend on the objective of the email and the type of your email campaign. For example, a monthly update / newsletter format might be 1-2 pages.
TIP: Be sure to put a list of topics at the beginning of the newsletter, so it is easy for the recipient to ‘jump’ to specific sections that may be of interest.
A brief offer to download a whitepaper or join an event, usually performs best if the email is short, to the point and includes links to landing pages with additional information. Another common style is the ‘educational, email blog’s style. If you subscribe to ‘The Motley Fool’ for example, you’ve seen this. The writing style is personalized, 1 on 1 communication that typically covers a topic or two, and explains the topic in detail. This might be 2-4 pages long, and can include links to landing pages and external documents with additional supporting materials.
Q: How many links can I use in my email marketing campaign? Will too many links reduce my response rates?
A: Based on analysis of our B2B customer data, adding more links will almost always improve response rates for an email campaign. Key points to consider:
- Try to include a link within your initial opening paragraph (“Read More..” for example). This can increase overall click rates by 12-15% and is virtually always the top-clicked link of an email marketing campaign.
- Do you know where the second best performing link is? In the ‘P.S. or Footer section! Don’t be afraid to ‘advertise’ or promote below your signature in your email marketing campaigns.
- Work more links into your email campaigns. Our analysis shows that campaigns with more links produced higher total click response rates while *not* increasing the unsubscribe rate. 15-20 links in a campaign is not unreasonable.
Q: How often can I send email marketing campaigns to our contacts, and how does this vary from B2B vs. B2C contacts?
A: The best rule of thumb is to communicate with your database when you have something interesting, useful and relevant to say, so the frequency will depend on your audience and your business. From analyzing our customer data, here are some guidelines to consider:
- ‘Typical’ B2B customers send email marketing campaigns to their database on average, 1-3 times per month.
- B2C customers tend to send more frequently: 2 – 8 times a month is the most common email campaign send frequency.
- Based on analysis of several hundred million emails and tens of thousands of email campaigns, we have seen that – if you stick to best practices and send targeted, useful information, the net response rate or ‘engagement’ rate improves overall if you send more frequently. Customers who send less frequently (less than 1x per month for example), see higher bounce rates, and experience declining open and click rates in subsequent campaigns.
Q: We use images frequently. Can you suggest a good text to image ratio?
A: This refers to the ratio of images to text in your email marketing campaign. SPAM filters (and image blocking) are both impacted by the ratio of images to text content.
For business-to-business email marketing, a ration of 80% text to 20% images is a reasonable guide. B2C campaigns will often have a much higher image ratio but interestingly, the subject line and introductory paragraph impact response rates more than imagery and design.
Avoid using large image blocks. Use images only where necessary and use text wherever possible. In addition, you can use ALT tags for images – this will improve your text to image ratio.
Earlier SPAM filters relied heavily on specific keywords to determine how ‘spammy’ an email marketing campaign was. Spammers simply created images that included the spam words, or created an email campaign that was a single big image. This allowed them to get by SPAM filters. However most SPAM filters now look at the amount of content that is text vs. images and if you are using a large number of images, content will often be considered SPAM.