The proverb goes: walk before you run.
My latest contribution to Harvard Business Review (link) makes the point that many websites can improve their user experience by focusing on simple personalization measures, like showing me my shirt size. Recommendation engines based on machine-learning algorithms still have ways to go.
I ran across a number of obstacles in my recent travel, which again highlights the value of getting the basics down. The following examples are not failed personalized marketing but poor execution of basic communications with customers.
The first problem was trying to check-in online from Delta's website. Here's what I encountered:
The only choices were using a confirmation number, a ticket number, SkyMiles number or credit card number. I am not a frequent traveler on Delta but on other airline sites, I usually offer my last name, and flight number, or destination, and that would be sufficient. As a frequent traveler, I understand that the requested confirmation number is the one issued by the airline, not the one by Travelocity, which was the website from which I purchased the ticket.
Here is Travelocity's confirmation email for the purchase:
The entirety of this email does not contain Delta's confirmation number, which is probably the single most important piece of data a travel agent needs to provide its customer!
I'm pretty sure that prior designs of Travelocity's confirmation email did include the airline's confirmation number. So this is a case of going backwards.
The second problem related to the ticket I purchased from JetBlue to go to Chicago for my upcoming keynote speech at Predictive Analytics World (register here). I needed to let the organizers know how much I spent on the airfare. So I opened the email confirmation from Jetblue.
You guessed it, this email does not report the price paid. I thought maybe they sent me a separate receipt but there was no other email in my inbox. Then I went online, using the Jetblue confirmation number, to look up my itinerary, hoping to find a "receipt" button on which I can click. Finally, I gave up and called my credit card company instead.
This post is about data. It's not about data models or data analysis. Sometimes, users just need selected raw data. Even a simple email should be designed with a view to its function. A key function of the confirmation of purchase email is data display.