Note from Jeanne: Happy Throwback Thursday! Smart data is the key to success in email marketing — as I laid out roughly a year ago in this column originally published by ClickZ. Enjoy!
“Big data” is often seen as the boogeyman of marketing – and in truth, I’m not a big fan of big data.
But I am a fan of smart data. Smart data is about determining in advance what information you will use for segmentation, targeting, personalization, and customization – and then using this list as your guideline for collection. Don’t collect anything that you don’t have a solid plan to use in the next six months.
Through my work with clients I’ve seen the benefits smart data provides to both marketers and the people they market to. Here’s a primer on four ways to collect and develop smart data that will improve the performance of your marketing efforts.
There are four key paths to smart data:
This is information that is provided directly to you by the customer or prospect. The most common starting point for this type of data collection is an email sign-up form or a purchase transaction.
But don’t start there. As you build your relationship with the person, they will likely be willing to provide more data to help you market to them in a more relevant manner. At each touch point (website, email, etc.) ask them to tell you a little more.
Reported data is my favorite kind of data for two reasons:
- You have a reasonable expectation of accuracy, since it was self-reported
- The person is unlikely to question how you know this about them, since they were the source of the data
Observed data is something that you see the person do. A simple example of this is which links a person clicks on in your email messages. A more advanced version of this is tracking website browse behavior at the granular level, so you can see exactly which pages an individual visited. Observed behavior provides insight into interests.
Observed behavior itself is 100 percent accurate (assuming your tracking and reporting works properly), although there is often a margin of error when the data is interpreted. For instance, just because I was browsing books appropriate for a 5-year-old doesn’t necessarily mean that I have a child who is 5 years old; I may have been shopping for a gift for my nephew.
If you’re using observed data be careful – it can be creepy. As a marketer I don’t flinch when I receive an email from a website I was browsing that invites me to come back and includes images of the products I was looking at. But I have heard from friends who aren’t marketers that this makes them uncomfortable.
There are other, less creepy, ways to use observed data – the key is in using the data in a less intrusive way. For instance, if you see that someone always choose the “tall” size type when they browse your website, feature items that come in “tall” sizes in the emails you send them – and don’t include any petite items.
Appended data involves using a third-party to overlay additional information on your customers and prospects. For instance, you might have email address, name, and ZIP code and you might want to append full street address so you can reach them via direct mail.
You can also append information on household income, interests, and other data.
I have mixed feelings on appends, for a number of reasons:
- Compared to the other methods here, appends are expensive.
- Much, if not all, of the information you’re getting from an append is something the person could have told you first-hand if you’d just asked.
- The accuracy of appended data varies widely, both by vendor and even within an individual vendor’s data set.
- There’s a high “creep” factor associated with appended data (it’s even higher when the data isn’t accurate).
Some companies will also append email addresses to your client file if you don’t have them. I don’t recommend this. There’s no opt-in permission when you append an email address (permission is not transferrable) so you are at a higher risk of spam complaints. Also, appended email addresses don’t usually open and click at any significant rate. So often it’s a very expensive proposition which will grow your database with email addresses that are not responsive. It’s a quantity play, at the sake of quality.
This is where you get creative with the data. You might see that most people who like A also like B – so if I reported (or you observed) me liking A, you might do some testing to see if I also like B.
Extrapolated data points vary widely on accuracy (which is why it’s important to test, not just assume that I will like B and move forward). If you’re subtle about it, it should not be creepy. You can also let people know that this is a “recommendation” engine – that since you like A we thought you might like B.
An example of extrapolated data are the recommendations that iTunes includes with every music receipt. The caption says, “Those who bought your selections also bought” and it’s followed by three album cover images. Years ago a non-marketing friend of mine mentioned that he loved that iTunes did this: “It’s such a great customer service! I almost always click-through and buy at least one of the albums they recommend,” he said.
Which is the Holy Grail – when your customers perceive your marketing as a customer service you have definitely arrived.
Until next time,