Alaska Airlines’ Lean Startup Test in Airport Lounges

By Scott Kirsner |  March 4, 2015

Seattle-based Alaska Airlines can already boast about plenty of things, including on-time performance (#1 major airline, according to FlightStats) and customer satisfaction (tops among traditional carriers, according to JD Power). But the internal Customer Research & Development team is always hunting for ways to upgrade the experience of flying with Alaska, which served nearly 30 million passengers last year.

Sandy Stelling, Managing Director of Customer R&D, and Jerry Tolzman, Customer Innovation R&D Manager, walked us through a recent effort to develop more personalized services, and shared some slides. Both will be leading a session at our Seattle Field Study later this month.

Goal: Get More Personal

For this project, we focused in on one area of our mandate: personalizing the travel experience for our customers. We wanted to explore what areas within personalization are ripe for our type of work. That started in June of 2014. One way we brainstormed was by asking what it would be like if we were the worst airline for personalization, and so we came up with funny ideas around that. We’d make customers tell us who they are over and over again. You’d have a prison jumpsuit with a number on it. You’d constantly show your boarding pass. Some of it wasn’t very far from reality, actually. We then brainstormed on ways we could make identification simpler, less of a hassle.

We boiled it down to three different approaches. Two were biometric identification (face and fingerprint.) The third was using a smartphone. Then we started using lean startup tools. We have a captive audience of customers in our airports. So we’d go into the airports as people were waiting to board and ran a fake door test.

We said, “We are now offering a service where you can drop your bags, get through security, board your flight, pay for things on the plane — all without pulling out a boarding pass or ID. We have three options – facial recognition, fingerprint, or phone – which do you want to sign up for?” We then pretended to collect the necessary information – a picture, a fingerprint or phone information – to confirm that they actually would enroll before we gently let them down by telling them it wasn’t a real product, yet. It was very close between fingerprint and phone. But almost nobody would let us take their picture.

We then explored fingerprint and phone. But we felt like the phone wasn’t going to get people through the security checkpoint, whereas a fingerprint potentially could. We then looked for our next biggest risk or assumption. We knew we could get hardware and were fairly certain we could connect it to our systems, so we decided our biggest assumption was, Will people sign up for it?

Finding the Easiest Way to Experiment

From there, we looked for the easiest way to experiment. That wound up being in our Board Rooms, or airport lounges. Those are some of our most loyal customers, and since they fly frequently, we’d get their repeat business more rapidly. It’s also our own contained area. We own the space and can do whatever we want within it.

We started in our Seattle Board Room. We stood at the end of the counter. When members came in, we asked if they’d be willing to sign up so that they could use a fingerprint scanner to gain access. We were hoping for 50 to 60 percent participation. To our surprise and delight, more than 80 percent of customers signed up. We used a fingerprint scanner that connected to a tablet via USB. After they registered, we wanted to know if they liked the experience.

We checked back every time they used the tool. We hypothesized at least 80 percent of customers would like the experience, and found that about 92 percent who used it for check-in again said it was good or very good, compared to the normal process. In terms of the chronology, we did our passenger interviews and our first testing in July 2014. By August, we had the first Board Room test up and running. We had a device and we were signing up people. In September, we rolled it out to all four of our lounges. (See timeline below.)

We had a really interesting debate as a team about whose responsibility it should be to roll the technology out to the other three lounges, after we’d gotten Seattle up and running. Is it our job, as the R&D team? Is it the product team’s job — the people who run the Board Rooms? We wrangled on that for a couple weeks. The purist approach might have been to hand it over to the product team, but we decided the R&D team was capable of doing our four locations. It felt like the right thing to do. (Pictured above is the Los Angeles Board Room.)

We’re still working on improving the scan accuracy. We’re looking at other options of plug-in fingerprint scanners to use. We want a higher percentage of instances where you put your finger down, it recognizes you, and you’re in. Then, the question is, how can we take this out of the Board Room and into the rest of the day-of-travel process… bag drop, boarding, and eventually the security checkpoint. That has historically been the #1 pain point, but it’s also probably the most challenging area to go into.


Before this, we’d been looking at what you might call “bright shiny objects” — solutions and technologies in search of a problem. But here, we started with the problem and looked to match it up with solutions. We had much better success. We probably missed bringing in a key stakeholder early enough — the IT services organization that supports hardware and installations. We did a pretty good job bringing along the software development team.

But we picked one tablet to use as our platform, and the hardware group said they didn’t know how to support it. So we’re now switching over to a different tablet. That has been a four month process. By design, innovation teams move quickly. A big question for all of us is, how do you make sure that the rest of the organization can keep pace with what you’re doing?

Timeline for the Project

(all dates are 2014)

  • 7/15 – Personalization Brainstorming
  • 7/18 – Identified Fingerprint Recognition
  • 7/28 – Verified Customers Willingness to Enroll
  • 8/1 – Approached Board Room for Testing
  • 8/5 – Application Designed
  • 8/11 – Minimum Viable Product Developed with IT Support
  • 8/21 – Rolled out in Seattle Board Room
  • 9/22 – Rolled out to all 4 Board Rooms


These are the team’s list of “unknowns.” The first shows them before the test was conducted, and the second after.

With regard to the yellow “Below Hypothesis” cell above, Tolzman explains, “Our hypothesis was that we could get 97 percent of users eventually checked in with the fingerprint and 80 percent on their first attempt. We started at 75 percent eventually and 65 percent first try. We are now closer to 90 percent eventually and 80 percent first try. We are still working on improving the accuracy, and have tried a number of ways to do so with some success.”