In this issue, I’ll show you how to …
- Find your influencers,
- Passively capture email addresses on your website, and
- Link in-store and online behavior
A customer data platform is the best tool for this, and I’ll be discussing these use cases in that context, but the insights and concepts are valuable whether or not you have a CDP. (And if you’re curious about CDPs, please contact me.)
Find Your Influencers
By “influencers,” I mean the people who effectively share your content with others, and by “effectively share,” I mean they share your content with other people who then visit your website.
To illustrate, imagine that Fred has a million followers on Twitter and posts links to your content, but nobody ever clicks on Fred’s links, while Ethel has ten followers on Twitter, and all ten of them click on every link she shares. Ethel is the more effective influencer for your brand despite her small Twitter following.
You can find your Ethels by appending a customer ID to the links in the emails you send. That ID then gets shared, and you collect it on your website.
Your email service provider very likely has a record id for each of the email addresses in your database. Let’s say Ethel’s ID is 12345. When you send an email to Ethel, append that ID to the link, so it looks like this.
When someone clicks on that link and visits your website with that ID in the URL, you know they came because of the link you sent to Ethel. What you don’t know (more on this later) is whether it was Ethel who clicked on that link, because Ethel forwards her emails to her friends and posts your content on social media, and all those shares include her ID.
Here’s where the CDP comes in. Every time someone visits your site, tell the CDP to look for that ID parameter and write it to the visitor’s profile. Then, find out how many profiles have that ID in common. In this case, what you’re finding are Ethel’s friends and followers — the people who clicked on the links she sent or shared with them.
When you see that 12345 is in more profiles than any other ID, you know Ethel is your most effective influencer. (Reach out to Ethel and see how you can make your site even more valuable to her.)
If you’re following my reasoning here, you probably noticed a hole in this model for finding influencers — that is, it relies on content you sent by email, where you could append the ID parameter. What if someone simply visits your site and shares that link, which doesn’t have the ID in it? Aren’t you missing the influence of people who share those links?
Yes, you are. You can’t plug this hole entirely, but you can make it smaller by appending the same ID to the URLs on your website. When Ethel comes to your site, and your CDP knows it’s Ethel, use some javascript to append her ID to every link on every page she visits. That way her ID follows her around the site. When she shares that content, it will also have the ID in the URL.
I know this is a little complicated. If you’d rather hear me talk this through, I created a simple video on the subject here. And if you still need clarification, give me a call.
Passively Capture Email Addresses
As should be obvious from the preceding, you can’t safely assume that because a person’s profile has 12345 in it, that person is Ethel. But … you should be able to do something else with that data, right?
To dig into this, we need to discuss deterministic vs. probabilistic matching.
The main use case of a CDP is to merge all of a customer’s records to create a single customer view. One person might have multiple records in your CDP, both from data you’ve imported from other systems, and from the fact that people visit sites on multiple devices.
CDPs differ in their strategies for merging those records. Some of them use complicated AI routines to make a probabilistic match. Others only merge profiles on a deterministic match, like an email address. Some do both.
Let me tell you a secret. “Deterministic” is a bit of an exaggeration. All of the ways we collect email addresses are subject to some error and leakage. For example, let’s say you’re going to trust the email address a person uses in your cart when they make a purchase. That seems reasonable, but my sister did a lot of Christmas shopping for my mother, using her credit card and login.
In the real world, nothing is absolutely deterministic, but things are often deterministic enough for some use cases. You have to get comfortable with the idea that you can have a different level of certainty for different situations.
That brings us back to Ethel’s ID. I can’t say “this visitor came to this site with Ethel’s ID, therefore this visitor is Ethel,” but I can say that if that visitor comes to the site often, right after we send an email to Ethel, then it’s pretty likely that visitor is Ethel.
You'll have to work with your data scientist to come up with appropriate rules for this. Once you’re reasonably comfortable you’ve covered all the bases, and believe you can infer that a particular email address belongs in a given profile, you should then store the email address that you inferred from the ID in your “not confirmed but probably right” email address field.
(Look for a future Krehbiel Report where I discuss different types of emails and their uses.)
To summarize, the method to passively capture email addresses looks like this.
- Append an ID to the links you send in your emails, and possibly also on every link on your site.
- Collect those IDs in your CDP and write them to each user’s profile.
- Decide on a rule that’s certain enough for your use case.
- Enrich user profiles with the emails that correspond to the IDs you’ve collected.
(Clarification, in case this isn’t clear: This use case is not for capturing new email addresses, but for connecting your online profiles with the email records you already have.)
If you have any questions about how to implement this, give me a call.
Link In-store and Online Behavior
Last week I bought a shirt that very helpfully included a little plastic bag with a replacement button. I immediately went to my button master file, logged this button and stored it in its own special drawer with an ID so I could find it when the shirt loses a button two years from now.
I did no such thing, of course. Sending me an extra button in a bag seems like a thoughtful gesture, but it's fairly useless, practically speaking. I’m simply not going to collect a million buttons and then have to search through them all later.
Along those same lines, I recently bought a pair of light brown men’s shoes, and last week I went to the store to find the right shoe polish. None of the polishes matched my shoes.
Both these examples are opportunities for retailers and/or brands to link in-store and online behavior.
Imagine that my shoe included a model number so I could go to the manufacturer’s website and lookup and purchase the right polish and replacement laces for that particular shoe. That would be an incredible service to me, but it would also allow the retailer or manufacturer to link my in-store purchase with an online account. That kind of data is gold.
The way things typically go right now, the retailer gets the in-store sale, and whatever data they capture with that transaction, but Amazon sells me the (almost, but not quite right) polish and laces.
This is not smart. With a little effort, retailers can capture better data on their customers and create a more lasting relationship.
Finally …
I hope you found this issue of The Krehbiel Report interesting and helpful. Please feel free to comment (the link to comment is near the top of this post on the left), and please pass this along to other people who may find it useful.
And by all means give me a call if you have any questions about this email, or about anything in the marketing / technology / publishing space.
