How Much Segmenting Is too Much?

• Tweet
• Buffer

In a perfect world, every e-mail you get would be driven off of your actual interactions with the brand sending it. Maybe you haven't been in the store for awhile. Maybe you recently bought online from the site, or maybe you are new to subscribe to the list and most recently searched for "kids' socks" on the site.

In the real world, creating templates that support dynamic datafeeds is difficult enough. Creating database integration points that feed all of the personal and behavioral data you need to speak personally to your consumer is close to impossible. Over 65 percent of companies eliminate all personalization except first name from their programs due to issues with datafeeds and dynamic content generation.

But how much segmentation do you need? And, in reality, how much is too much? I posed this question to 25 people I know in the e-mail industry who I consider "experts" and got this set of responses and recommendations.

How much segmentation do you need? (Unedited comments from various industries.)

• Retail: Segmentation should be ignored until you are building a win-back program. Until then, standard messaging should be built to meet all audiences
• CPG: Segmentation should never be built after the fact, it should always and only be part of the registration process to ensure 100 percent accuracy. If the subscriber doesn't tell you, you shouldn't guess.
• B2B: No segmentation is the best route to go. A drive to a phone call is where the segmentation should occur.
• Auto: Segmentation should only be driven off of cookied site activity. If the customer tells you what they want, it's almost always different from what they actually look at and interact with on the site.
• Publishing: Outside of staying within the right gender offering, segmentation should not be about behaviors as much as what will drive spend on subscriptions.

Five recommendations to create good segments:

1. Preference center
2. Using search terms from your site
3. Building a netmining profile and applying it to clusters of your database
4. Using a recommendation engine from your site
5. Choosing the top five clicked-on offers or links from prior efforts

I found these results and thoughts interesting. The recommendations seem like something almost any company could apply, but the perception around the importance of segmentation varied significantly by industry. This is certainly one of those areas where the answer is "it depends" until you are able to dive deeper into the specific behaviors that your brand or industry emulate and define a strategy from there.