Why We Track
Statistics are critical to any email campaign, be it bulk or transactional. In order to run efficient and effective campaigns, an organization must have in place some way of gaining insight into how their messages are received and acted upon by their customers and subscribers.
Data such as delivery, open and click rates and number of abuse complaints (spam reports), opt-outs (unsubscribes) and opt-ins, allow a company to analyze performance and act accordingly to optimize their content and other aspects of their emails (e.g., subject lines, headers, from: address, etc). Through calculated optimization of email-specific key performance indicators, a business can maximize its return on investment. After all, email delivery, whether executed in-house or by a third party vendor, can represent a significant operational and/or marketing expense.
Beyond the Basics
The points above should seem pretty obvious to a seasoned team in the email game. So, once you have this easy stuff taken care of, it's time to work toward a greater level of granularity. One way to do this is to segment your email traffic into "categories".
Segmentation can be applied on several levels. One very useful, high-level distinction: transactional email vs. marketing email.
The end recipient of an email understandably treats a transactional message (e.g., new account confirmation, password reminder or invoice) differently than a marketing message (e.g., monthly newsletter or promotional announcement). By isolating the two categories from a statistical standpoint, you avoid potential confusion around their relative performance and can make educated decisions regarding improvement efforts. To achieve such categorization, it is often best to send your transactional and marketing traffic from separate, dedicated IP addresses. This not only allows simple data segmentation, but also creates a situation where the [typically lower] reputation earned by marketing emails does not affect the deliverability of transactional emails.
Once you have in place this high-level segmentation, you can dig even deeper into each of these two main categories and define "sub-categories".
Within the transactional email category, one might want to view statistics on new account confirmations separately from new friend/follower notifications. These sub-categories have different levels of importance in your users' eyes and are thus treated differently when found in the inbox. Likewise, they represent different values to your business – if the activation link within a new user's account confirmation is not clicked, this will likely be more costly than if the link to the profile of a user's new friend/follower is not clicked.
Similarly, within a marketing email campaign, it could be beneficial to view metrics on your September newsletter side-by-side with your October newsletter. By comparing the opens, clicks, unsubscribes, and spam reports, your team can effectively perform A/B testing on newsletter attributes such as content, tone and design.
In this context, sub-categorization does not require separate IPs, but rather some sort of "tagging" system that defines various sub-cateogires. Typically, a "tag" can be embedded in the email header, so that your tracking system can identify each sub-category and report distinct statistics as you have defined them. This type of tracking infrastructure surpasses the basics of email delivery, especially if you are handling it all in-house. However, there are many email service providers that make tagging and sub-categorization as described here a much more trivial effort.
In the end, the practice of data segmentation, in the context of email categories and sub-categories, goes beyond "best practice" and into the land of "'best-est' practice" It certainly requires more resources and expertise, but it can also pay off big, by turning a costly email campaign into a profitable investment.