A growing majority of customers choose to self-install their internet equipment to save time and money, do it on their own schedule, and feel accomplished of having completed the setup themselves.
During the first year of the COVID-19 pandemic, when technician installs were unavailable, helping CenturyLink customers successfully install their services became even more critical. With potential savings of $16 million per year, there’s a clear financial incentive to make sure self-install customers can easily follow the self-install process without having to call about questions or issues that could be addressed in the instructions.
The Customer Experience (CX) team requested that our team update the paper instructions included in the self-install kits. The team consisted of a UX researcher who had been researching self-install, and me.
The existing instructions had been tested with real customers, and we knew there were some important issues to address. However, we only had a week to get something to the printer, so I had to look for the biggest improvements I could make within the existing one-page format.
Most customers couldn't remember if they'd signed the Terms & Conditions, so they weren't sure which path to choose.
The diagram didn't provide enough detail to be considered "instructions."
Customers sometimes downloaded the wrong CenturyLink app.
Even little things, like having to enter the URL perfectly, suggested that this process was complicated and easy to mess up.
To understand what we'd learned from testing the existing instructions, I sat down with Linda, a UX researcher on my team. I also watched several recordings of these sessions to get a sense of how real customers were feeling during self-install.
From the customer perspective, there were five questions that often arose from the current instructions:
I had lots of questions about the existing instructions, as well as the self-install workflow, so I interviewed five different stakeholders for the project: supply chain, mobile app, web, zero-touch (plug-and-play) activation, CX and internet product.
I learned that self-install as a whole didn't have a leader. Instead, it was a collaboration between different departments, but there wasn't a person or team who was overseeing the entire self-install journey and determining the strategy for self-install. As a result, the instructions had become an amalgamation of the requirements for each team (mobile app, web, and zero-touch activation), instead of a singular, cohesive experience for the customer to follow.
The three paths represented in the existing instructions each had their strengths and weaknesses.
But to add some objective measures, I asked for data around each path, particularly activation success rate and call-in rate.
While the organization did not have clear data, especially for call-in rate by activation method, through proxies we were able to come up with decent estimates for activation success rate.
Zero-touch activation (AKA "plug and play") had the highest success rate at 78%. However, only 40% of customers qualified for zero-touch activation, and customers have no way to know if they're qualified or not. Also, if the self-install doesn't work or they still have to accept the Terms and Conditions, the customer will have to visit the web self-install app or mobile app.
The web app had a higher success rate, but the site was essentially a dead end. It was a legacy property and there were no plans to update the site to improve the user experience.
The mobile app had the lowest success rate, but it also had the most potential. There were more resources being invested in the app, and it's a tool the customer can continue using in later phases of the customer journey, like managing their services and getting support. The app could also mask some of the backend processes from the customer. For example, if the customer still needed to accept the Terms and Conditions, the self-install flow in the app could simply present that screen as if it were a part of the flow.
Based on what I’d learned from previous user research and conversations with our stakeholders, driving people to the mobile app seemed like the best strategy for matching customer expectations to download an app to complete a self-install flow and masking behind-the-curtain process issues like accepting the Terms and Conditions.
The app also would provide a consistent customer experience regardless of their individual situation:
This reframed the goal of the one-page instructions from presenting the customer with three potential paths to simply prompting the customer to download the app.
Before I joined the project, two of my colleagues were testing another version of the one-page instructions. This one-sided version led with driving customers to download mobile app to do self-install, but also had brief instructions for connecting the modem to the computer.
This version had also undergone some testing, with these results:
Based on this info and previous research, I decided to drop the section with directions on connecting the modem to a computer and focus on the app instructions, which could become even more simplified.
After sign off from our stakeholders, I delivered this new version to our supply chain manager and printer.
Knowing that customers had downloaded the wrong app in previous testing, I added a larger image of the app icon to reinforce the correct app to download.
Given the increase in QR code usage during the pandemic, I also added a QR code as a way to download the app. In the two-week pilot program for this one-pager, 80% of customers downloaded the app by scanning the QR code.
This version was a big improvement: it didn't ask customers to remember if they'd accepted the Terms and Conditions, and it was one-sided to prevent the customer from getting lost.
Not knowing where to go for support was a common complaint in previous testing, so I wanted to provide several ways to get help and troubleshoot.
Two things stuck out to me from the previous research:
These points suggested that there was an opportunity to keep iterating on the instructions to support different learning styles and tendencies, as well as better fit the customer mental model for instructions.
I researched other companies in the home technology space, as well as other big brands with diverse customer bases, including Google Home, Ring Doorbell and Kraft. I also looked at Lego and IKEA as examples of universal design—companies that communicate to a broad audience of different ages, abilities and languages with one artifact.
There were several commonalities among these instructions:
With these takeaways in mind, I began exploring a concept for a short instructional booklet.
Knowing that people equated "instructions" with the connection steps, I broke down the connection steps into separate panels with large numerals to indicate the step, as well as brief descriptions of what to do.
Then I wondered if I could reduce the copy even more and use only illustrations to communicate. In theory, this would be even more universal since they could be understood regardless of primary language spoken. An illustration-based approach also could better support people's tendency to skim or not even read the instructions.
The sequence of the steps was intentional: after steps 1 and 2, while the modem booted up and connected to the network (which takes one to a few minutes), the customer would be downloading the app. Thus we could minimize their perceived waiting time.
Also, as more customers qualified for zero-touch activation and could essentially "plug and play," leading with the two connection steps would help future-proof these instructions. Like the one-page version of the instructions, downloading the app would be to simply confirm the activation was successful or allow for troubleshooting, while introducing the customer to the app as a tool to manage their services and account.
But would the illustrations be interpreted as intended? At this point, I chatted with Linda, my UX research colleague, about testing the booklet with real customers. She had been testing self-installs with customers on a semi-regular basis, so after getting our next tester's mailing address, I headed to FedEx to have a copies of the prototype booklet printed and shipped to our next two customer testers.
I liked how Ring balanced large copy for the main step, with more detailed copy and an accompanying illustration.
The insets in Lego's and IKEA's instructions are also helpful for showing details.
We were able to test the booklet prototype with two customers who were self-installing new modems.
Both quickly took inventory of what's in the box, and both were able to easily complete the two connection steps. However, the third step of downloading the My CenturyLink app caused some confusion.
“Do I go to the CenturyLink app that I [previously] downloaded? I would say that I would have to open the app.”
"Now I'm not sure where to go from here... looks like maybe setting up an app on my phone."
Although the second customer wasn't able to complete their activation due to a faulty DSL jack, both customers gave high ratings to the instructions (10/10 from the first and 8/10 from the second).
They also had some feedback on what could be improved.
"When you get to page 3 where the app store is, you have questions there. Would probably be better to say 'download this app.'"
"Just tell customer to download app and set it up. I can assume but not always good to assume."
"Not sure about bottom left [illustration]. Don't know if that's an arrow or what. Those cause some questions."
These tests also validated the strategy of driving to the mobile app. Neither customer was bothered by having to download the app—a stakeholder concern—but rather they expected to do so.
Stakeholders were also concerned about older demographics, in particular, struggling with the app. Yet one of the testers, who is in their 80s, had downloaded the app without issue in advance of the self-install session. While some customers theoretically could struggle with this step, it seemed age was not as strong of a predictor as some stakeholders thought.
While testing the booklet with real customers was the best way to assess its overall effectiveness, our pipeline of real customers was running dry, unfortunately.
But because my main questions were around the illustrations on the dos and don'ts panel and we could test for comprehension of those illustrations outside of an actual self-install context, we decided to continue testing only that panel on UserTesting.
My colleague ran an unmoderated test that asked testers to explain, in their own words, the meaning of each image. Then, when the intended meaning was revealed, testers rated how well the image communicated the intended message.
Images A, B and E all tested well, but Images C, D and F stood to be improved. The key takeaways included:
"[Image A] looks like a basic illustration of putting the modem in your home; makes sense in context of illustration B."
"I get it but it's a little confusing. Instead of all of the arrows that indicate airflow, maybe put the modem between two books so it's obviously that it's too close and maybe a fire symbol to indicate heat and put an X on it."
"I didn't get the overheat part of it, I took it as a signal perspective. Without image D to compare it to, I wouldn't have gotten it."
"I didn't get that at all. I thought it meant don't put it around clutter or objects. I thought it was a weird combination of items."
Other tester feedback I addressed:
We then ran the test again with Version 2.
This time, Image A was rated less effective than before, but Images C and D were rated better. Interestingly, Image B was rated worse even though no changes were made to that image.
The main takeaways from this round:
We then ran the test for the third time, this time testing only Images A, D, E, F and G (B and C were unchanged from the previous round and not tested).
Images A and D both received a 9.3/10 average rating, but there were some lingering issues with Images E, F and G.
Although E, F and G didn’t produce perfect comprehension, we agreed the overall results indicated these illustrations were in a good enough place for us to move forward. In these cases perhaps a line of copy could help raise comprehension, but according to the verbatims, people got the gist.
Both customers also mentioned needing "get their glasses" to complete some steps where they had to reference small copy on the modem itself. Using large copy and text benefits everyone!
I learned from a stakeholder that all new customers would be receiving CenturyLink’s signature Greenwave modem within the next month, so I drew the Greenwave modem and swapped it in for the generic box modem illustration.
I also did some design explorations for the front and back covers of the booklet and critiqued them with our design team.
Around this time, I was tapped to take on a new project and this was project was tabled.
After a two-week pilot program, our sponsors decided to continue with the one-page instructions. While it was difficult to connect the new instructions' direct impact on activation success rates and call-in rate, app users grew by 15% month over month, which could be attributed to increased self-installs in the app.
There were a few things I wanted to do next had I stayed on the project: