Tag Archives: sentient databases

The Importance of Targeting

I’m super behind on more “Lessons in Product” posts, but this one begged to be written when I got yet another badly-targeted email from Weebly, a site building tool I used for my portfolio domain (www.tiffanyshih.com).

I created an account with Weebly on May 11 and built my entire site that day. Weebly makes it very easy, and I haven’t had to touch my site since.

In the 19 days since I signed up, I’ve received 11 emails from them. The first three were confirmation-type emails (transactional emails) sent on May 11:

  1. Welcome to Weebly (signup confirmation)
  2. Congrats on your Weebly Starter Site (upgrade purchase confirmation)
  3. Tiffany, your website was born at… (site publication confirmation)

Totally normal, totally standard, totally acceptable. Then it went downhill.

  1. May 11 – Here’s $150 from Weebly to try Google AdWords
    This one represents the peak. It was good at pushing a product feature that I wasn’t aware of. The promotional aspect didn’t bother me as much because I’d just purchased an upgrade and it was pushing something that seemed to be bundled. Bonus points: it called out my recent purchase by package name. (1 of 1 features useful)
  2. May 12 – Getting the most out of your new Weebly Starter Plan!
    Good intentions, but a day late. This email pushed four features of the Starter Plan, two of which I’d already used the day before. (2 of 4 tips useful)
  3. May 12 – First Steps to Creating Your Website
    The first truly useless email. First steps? Aren’t I done? Didn’t you send me a confirmation that I’d published my site and it was live in the world? This one actually had the gall to say, “The race is on! Is your website going to cross the finish line?” (0 of 1 calls to action useful)
  4. May 14 – You’re ahead of the pack!
    This one at least referenced the fact that I’d published a first version, but then it proceeded to suggest three things I’d already done. (0 of 3 tips useful)
  5. May 18 – Congrats! Your website traffic is skyrocketing
    Aha! A useful email with embedded, real stats on how many page views I’d gotten so far. It also suggested five features to optimize performance — including the suggestion to register a domain with Weebly… which I did when I set everything up on May 11, and they’d already suggested in email #7. (4 of 5 features useful)
  6. May 22 – Tiffany, check out your mobile website!
    “Did you know your website is mobile friendly?” Why yes, I did. I toggled between the mobile and desktop views in the build mode back on May 11th, remember? And what’s the point of this email anyway? The email just says, in essence, that Weebly did it for me already. “Hopefully you’ll rest a little easier at night knowing that!” No, I’m actually just thinking about unsubscribing from your email list, but I’m so fascinated by how bad you are at targeting useful content at me that I’m still hanging around. (0 of 1 calls to action useful — but was there even a call to action here? 0 of 0?)
  7. May 26 – Photo Galleries | Contact Forms | Audio & Video | Maps
    “You may want to consider adding some or all of these to your site!” I have considered! In fact, I have a photo gallery on my site, and I plopped in a contact form before deciding to take it out. (2-3 of 4 features useful)
  8. May 29 – 5 Tips for Choosing a Great Domain Name
    Oh man you’re right, I should really choose a great domain name. How about [firstname][lastname].com?? That’s pretty good, right? Oh wait, I registered that with you guys already. (0 of 1 calls to action useful, 0 of 5 tips useful)

So what’s the lesson to be learned here? Send your users relevant content. We live in an age where behavioral tracking on websites is normal — and even if people are going to get all upset about privacy, most of the useless tips revolved around aspects of the service I’d used and would assume Weebly would know already. 15 of 25 things Weebly tried to tell me were completely useless to me; that’s 60% useless content being sent to me. If I weren’t a newly minted PM with an interest in content personalization, I would’ve unsubscribed around email #7.

How could Weebly have been less shitty at sending me emails? Targeting. All that really means is asking a question (or, for the coders out there, writing an if statement) before sending the email — has this user done anything that means the content is irrelevant?

Here are some questions Weebly should have asked:

  • Has Tiffany already uploaded a favicon?
  • Has Tiffany already used a custom footer?
  • Has Tiffany published her site (or, have we already sent Tiffany an email congratulating her for publishing her site)?
  • Did Tiffany already enter stuff in the “organize your site” on-boarding exercise on our site?
  • Has Tiffany used a custom theme?
  • Has Tiffany linked a registered .com domain to her account already?
  • Has Tiffany already put a photo gallery on her published site?

Targeted content is rapidly becoming the standard behavior for websites. I expect that Google knows where I am and will only return business search results in my area (take note, Apple Maps — I was not looking for the Whole Foods in Boston when I’m physically in San Francisco). I expect that Buzzfeed is going to recommend articles that align with my interests based on what else I’ve read lately. I expect that Amazon will tell me what baby toy I should buy based on what other people buy after viewing the stuff I keep looking at. All in all, targeting creates a better user experience, and it surfaces your site’s content to the people who will actually benefit from it.

So why don’t sites do targeting? Well, targeting means you walk a fine line between personalized and creepy (for more on this, you might want to listen to Vienna Teng’s The Hymn of Acxiom). Yet basic targeting based on transactions I’ve made on a site — like Weebly should have done — isn’t creepy. It’s just lame when you don’t do it. So the real reason is laziness.

Targeting requires knowing what your users want and how they behave. As targeting complexity increases, you start needing to store information about your user’s behavior that you don’t already. For example, a tip about uploading a favicon should be targeted to me if I haven’t uploaded a favicon. Pretty simple: you check the records for a user and see if you’re storing a favicon for that user — if not, by all means send me an email. So let’s up the complexity of what we’re remembering about a user: what if I went to the favicon page and didn’t end up uploading a new one? I probably didn’t because I didn’t want to, or I didn’t have a file ready, or I was in the wrong place and meant to go to favorited pages or something. In order to de-target this tip that I don’t need, you would have to implement something that either (1) remembers that I shouldn’t get the favicon tip, or (2) remembers that I visited the favicon page and left without uploading. Option 1 is sufficient to de-target the tip, but option 2 leaves open the option to send a more personalized message that says, “Yo, you didn’t upload a favicon when you were on that page. Did you forget? Here’s how you get there! Did you need help? Here are some FAQs!” Option 2 also has the potential to differentiate that came-and-left behavior from uploaded-and-removed behavior, for which you’d probably send a different personalized message. But as you store more and more behavioral data, that’s when the stalker database factor starts creeping up.

So we know that targeting is tricky as you store more and more information about a user’s behavior and analyze what that means about what they want to do on your site. The other consideration here is implementing the content presentation. How much do you want to personalize an experience by sprinkling little if-else statements all over your code? How big of a content chunk is getting wrapped in this if statement — a bullet point in an email? a whole email? a whole feature on the website? You have to ensure that you aren’t completely de-targeting a useful feature from someone who might want to use it because you’ve misestimated what type of user behavior maps to this content.

Sure, it’s tricky. Sure, you might have to do some user testing, some experimentation, and some deep thinking about user behavior. But do something. And if you’re a website website like Weebly, you might want to invest some time and resources in targeting before you completely embarrass yourself.