Specific vs Automated SegmentationBy Brennan Dunn
I'm a big proponent of building in the open.
The very fact that you're getting this email means that you're a potential beneficiary of what we're doing with RightMessage, and we want to make sure what we're doing aligns with what you actually need.
By creating a dialogue around what we're doing with RM 1) drives us forward faster and 2) helps us make sure that we're staying on track.
Additionally, I realize that you're likely not doing (much) personalization yet. There are few tools out there, and they're targeting more enterprise ecommerce businesses. From the looks of the domains I've seen sign up so far, that's few of you. So part of my job is to provide you with examples and education on how you COULD increase sales through personalization.
Today I'm going to share some thoughts on segmentation, which is the root of any effort to personalize. If you'd rather not get these sort of emails, but want to be notified when we launch, just reply and I'll set that up for you.
Personalization just means tailoring the experience someone has with you. Amazon does it, Facebook does it, every web app on the planet is personalized by showing just the content that's related to your account.
Few marketing websites or blogs personalize, and it's probably because most people think of these things as static. (Reading a blog or feature's overview page doesn't usually logging in.)
However, if we know who's looking at our website, there's a good chance we know something about them. And everytime someone opts-in to something on your site or clicks a link in an email of yours, an authentication event is happening.
If [email protected] gets an email from you and I click a link in that email, the website receiving that click should now know that [email protected] is looking. And if it knows who I am and has information about what I've bought, what lead magnets I've acted on, and so on, it should be able to make some decisions by treating the email service provider that sent the email as a database of user information.
RightMessage is going to make that relationship foolproof and automatic for you. Most impressively, it's going to allow you to do really interesting things — like customize the call-to-actions on your site, swap in testimonials from people like the person viewing, etc. These things, done right, can massively increase the results you're getting on your website.
But to be effective, you need to segment.
You're mostly responsible for segmenting. You should be applying tags or custom field data about lead magnets downloaded, products purchased, survey answers chosen, and so on. My typical subscriber who's been on my list for a few months has more than 100+ tags applied to them because I track literally everything. And you should do the same.
I call the above specific segmentation. "People with the tag 'customer'. People who have the 'business_type' custom field set to 'designer' as a result of completing a survey we send them."
There's also automated segmentation, which is a very interesting rabbit hole to explore. This is information like, 'What content consumption trends can we pick up on?', 'Where are they located?', or 'What key pages have they visited on our website?'
RightMessage allows you to combine both forms of segmentation data, specific and automated, in a pretty interesting way.
When you setup a change in RightMessage (like, "Swap out this call-to-action with X" or "Replace the hero image with Y"), you're going to tie that change to a particular segment.
And within RightMessage, you'll be able to create segments using the following criteria:
Browser - Information we get from the user's device
- URL Parameters (like UTM variables)
- Browser (e.g. mobile Safari)
- Device (e.g. iPhone 7)
- Cookie (any cookies that are set for the user)
- Language (browser's language)
Behavior - Information we get about how the user has engaged so far
- Original referrer
- Whether they're a first-time visitor
- Time since last visit
- Viewed X number of pages within Y timeframe
- Viewed a particular page, and whether they viewed it within the last X page views
- Original landing page
- Viewed X% of a particular page
- Content scoring - e.g., of the groupings of content of your website, like blog categories/tags, what trends can we pick up on?
- Key events being triggered (like the sort you send to Google Analytics, Mixpanel, etc.)
- Goals being triggered (like purchase events)
Contact - What internal data we store about the user
- Whether the viewer is known (true/false)
- Tag data
- Custom field data (name, email address, etc.)
Through the combination of the above, we could say something like:
- If the user hasn't yet opted in and has primarily read material on marketing, change the end-of-page CTA across all pages to be something marketing related, and on the homepage change the hero to be marketing-y.
- If the user hasn't yet opted in but showed up on our homepage from a blog focused on marketers, change all the CTAs across the site and the core language to assume the viewer is a marketer.
- If the user is known but hasn't yet bought anything, change the CTA and the highlighted option in the navigation to point people toward the entry-level product.
- If the user is known to be a designer and has clicked at trigger link signifying that she needs help closing deals, when she's viewing our course's sales page change the headline and copy to speak directly to designers, show testimonials from other designers, and when presenting the offer really focus on how the product will help her close more deals.
The above is pretty common sense. I don't think anyone needs to A/B test whether any of those bullet points, when implemented, would move the needle.
I've been doing variations of the above for almost 2 years now, and line-for-line it's been the most profitable code I've ever written.