I graduated from Le Wagon’s Mexico City coding bootcamp just as coronavirus started to ravage North America. Arriving back in Chicago in April 2020 felt like something out of an apocalyptic movie. Most of my friends in Chicago were fine and working from home. Meanwhile, most of my friends in New Orleans, the majority of whom work in tourism and hospitality, were completely screwed — myself included. When I went to register for my own pandemic unemployment assistance, this is the website I encountered.
I immediately sent this very screenshot to my classmates from Mexico, where they proceeded to collectively rip on it on our WhatsApp group chat for the next three days. I haven’t seen seen anything so ripe for redesign in at least ten years. Moreover, watching video testimonies about people not getting their unemployment because of systemic failures related to websites like this was heart-rending. As far as an unsolicited UX redesign portfolio project, it almost seemed too easy. There’s so much low-hanging fruit!
Where to start? Research-wise, you’d think the first thing one would look into would be unemployment stats in Louisiana, both pre- and post-COVID, but the first rabbit hole I went down was prompted by all of the text on the homepage. It’s tiny, there’s a ton of it, and the site isn’t responsive. 30% of web users in general — and a whopping 70% of people on welfare pre-pandemic — read at a 6th-8th grade level. Literacy and poverty go hand in hand, particularly so in Louisiana, where we have among the top three highest poverty rates (19.7%) in the nation, with some of the lowest literacy to boot.
Louisiana is also a frontrunner in a variety of awful pandemic-specific statistics. It entered the pandemic with the highest unemployment rate in the US — 6.9%, or 1,500 claims per week — though the real pre-Covid unemployment rate statewide is estimated to be more like 25%. At the same time, Louisiana has the second-lowest unemployment payout in the country — a maximum of $247 per week — and only five states overall are poorer. In New Orleans, the situation is particularly dire. The virus hit at the worst possible time, between Mardi Gras and the festival season that runs into June. Atleast 12% of the city relies on tourism for the bulk of their annual income, and 8% is caught up in the gig economy. In line with state numbers, 25% of New Orleanians are unemployed as a result of the pandemic.
To deal with this, the state hired a cruise ship’s worth of new call center operators — boosting their original staff from 25 up to 300— and employing an astounding 500 total staff members dealing with pandemic unemployment. Yet people in Louisiana are still having major problems getting through. The Louisiana Workforce Commission normally gets 4,000 calls per week; by mid-April they received 28,436 calls in a single week. On top of this, the website and phone hotline struggled with myriad crash-causing authentication issues both via phone and web, notably “online applicants who used the same system in the past and have since forgotten the answers to their security questions.”
Absent a direct line to anyone at the state capital, I started looking into other states’ websites in hopes that they’d help solve Louisiana’s puzzle. I found little in the way of solutions, but plenty of pitfalls to avoid. Most of the sites I reviewed were so perversely bad that exploring them one by one became a morbid curiosity. I ended up looking through every single state unemployment website in the country. Nationwide, at least 50% of people weren’t getting their pandemic benefits. Some sites told people to log in the first thing in the morning; others, like Louisiana’s, suggested that people log in in the middle of the night. The only websites that seemed to have paid a designer of any kind were, unsurprisingly, Massachusetts, who hired some designers three years ago, and New York. Despite being among the only proactive such entities in the nation, they too had significant problems:
After perusing every state unemployment website in the country, I personally rank Louisiana’s second worst, followed by the bronze-medal, void-esque website for Nevada. The gold medal for bottom-of-the-barrel goes to Florida. Their site truly has to be seen to be believed. It’d be darkly funny were it not so tragic. Florida has the highest covid-related unemployment in the nation. The first time I visited the Florida unemployment site, it claimed to be closed, which is something I have never encountered on a website. Florida’s broader unemployment ecosystem is its own singularly nightmarish rabbit hole, deliberately mis-designed by the powers that be to manipulate state unemployment statistics for politicians to use as campaign talking points. Florida’s ingenious interim solution to their myriad bottlenecks was to add a “virtual waiting room” to “improve user experience,” which, to the surprise of no one, hasn’t gone well. With a backlog of half a million claims, at the time of writing, they’ve reverted to accepting only mail-in applications.
In the absence of real users to test, I had to rely solely on indirect research to get an idea of the actual experience of applying for pandemic assistance. I quickly gleaned that Americans come to this expecting the worst. In terms of dread, applying for unemployment is like going to renew one’s driver’s license, but worse. It’s the diametric opposite of what Americans generally tend to expect from, say, the thoroughly streamlined experience of online shopping. People intuitively understand that the pandemic would bring a level of utter chaos to a process that’s painful during “normal times,” and calibrated their expectations accordingly. Nevertheless, social media was full of logistical horror stories juxtaposed with memes displaying the easy, breezy, beautiful ease of applying for pandemic assistance in other countries— especially Canada. Indeed, social media seemed to be the de facto place to go for intel in the absence of communication from unemployment agencies. People of all stripes were seeking reassurances from others further along in the process: who was getting benefits, who wasn’t, wait times, back pay, etc.
One Useful Insight
The biggest surprise of this little investigation was two-fold. On one hand, I saw surprisingly few accounts of people being stymied by lengthy, confusing forms, poorly designed websites, and other common trappings of bureaucracy. Again, it’s what’s Americans expect. On the other hand, by far, the #1 complaint I discovered was massive stress in people not knowing what their unemployment status. In most parts of the country, people are lucky to even see a vague “pending” status message on their online portal, while most receive no feedback at all. Stories abound of people in dire straits calling unemployment offices literally hundreds of times in a week, creating even more bottlenecks.
So, my first area of focus was to simply clean things up. And with the aforementioned in mind, my second area of focus would be somehow relieving the terror of the unknown. Doing so could drastically reduce call center volume and dramatically improve the experience for the user. It seemed like a simple pressure release valve would be to programmatically update a timestamp on the user’s end automatically every time an individual’s case was viewed, edited, or updated in any way by an admin on the state end. This would be relatively simple code-wise, and create zero extra work for already over-taxed state employees. Users even could opt in to email or text notifications whenever the timestamp changed, further mitigating their stress.
Starting the Redesign
My first priority in redesigning the homepage was cleaning up, simplifying, and placing the focus on Covid unemployment insurance, which is why the vast majority of people were suddenly visiting the site in the first place. Below is the basic flow I drew up for the vast majority of the site’s users. I’m focusing on the left side of this diagram, the only part of this ecosystem I had any access to.
Though Massachusetts’ site looks nice, especially compared to the competition, it still veers towards the overwhelmingly complex. Even the website of reigning state-services champion Canada has the same problem: a mind-numbing amount of text. Adding prominent videos presenting the same content would be a simple, straightforward improvement. Relatedly, given the quantity of users accessing the site who don’t own computers, Louisiana’s site design would need to be designed mobile-first.
I had just reread Don’t Make Me Think, and the one thing that really stuck with me the second time around was: “If your audience is going to act like you’re designing billboards, then design great billboards.” I wanted to start with a simple banner to highlight the covid-centric part of the site, and not including a hero photo would draw attention to the videos.
On the current website, the #1 most important link is “My HiRE Account” — it’s where you log in, file for, and manage your unemployment. (Six months later, I still don’t know what this peculiar “HiRE” acronym stands for). Yet this all-important link is buried, in microscopic 10-point type, in an inconspicuous three-column upper nav. This is what needed to be converted into a billboard-esque banner. If people could successfully manage to even begin to file online, it would take the massive stress off of the state’s 300 new call center workers. And, for the large portion of the population with low access to technology and/or literacy, a link to call in by phone still needed to be prominent.
The other inescapable inconvenience in this whole process is that everyone has to return to the site every single week to keep their claim going. The insurance gods call this the “weekly certification,” though from a UX writing perspective, “Weekly Check-In” would be better. Though this weekly element should be the second-most important button on the homepage, a link to it doesn’t exist on the homepage of the current site whatsoever.
Here’s a beautiful rough sketch of the mobile design with this and Krug’s “design a billboard” mantra in mind:
And the corresponding desktop version, a drawing somehow even more strikingly beautiful than the last:
After doing these and other low-fi paper mockups, I was so deeply in my early honeymoon phase with Figma that I made the rookie mistake of going straight for hi-fi mockups. As my just punishment, I got completely lost in variations for the mobile version of the homepage:
The clear superiority of large buttons for the homepage didn’t become clear until I went in to design the dashboard, where an accordion-style menu would be more appropriate for navigating the dense site architecture of actually filing for unemployment. Here’s what I came up with for the mobile homepage (static version here):
And here’s the desktop/laptop homepage (static version here):
Redesigning the Dashboard
Now, on to the dashboard. Once you log in from the homepage, you’re confronted with an even more preposterous assemblage of irrelevant widgetry. Behold!
As a person who was myself filing for pandemic assistance, I expected a multi-hour nightmare in my immediate future when first encountering the dashboard above. I was surprised and heartened when it wasn’t all that painful after all. The form pages mostly looked like this, and took under an hour to fill out:
Sure, these aren’t mobile-optimized, and there’s lots of insurance jargon throughout, but at least they added a progress indicator at the top. I had a few moments of confusion filling out the form because being a tour guide was cash-based gig work, but nothing crazy. A section at the end of the form where you could attach notes and questions to the representatives handling your case would be helpful. As is, users have to send questions and notes in a totally separate, disembodied-feeling, explanation-requiring message in the “communication” center, which complicates regular email and inspires dread. But, in sum, and especially when compared to the totally bonkers homepage and dashboard, my quibbles with this core part of the process — the form — are blessedly minor.
On the dashboard, the three most important things you need to see once logged in — filing a weekly certification, messages, and your application’s status — are buried deep in the inconspicuous menus at left. Instead, screen real estate is placed on a needlessly huge photo carousel, job listings that have nothing whatsoever to do with the user’s skills and experience, a mysterious calendar tool, and a bevy of links that are already accessible from the main site.
So, without further ado, here is my dashboard redesign, which sweeps 97% of the bloated foolishness of the current dashboard to the sidelines where it, at least during a global pandemic, belongs, to focus only on the essentials:
This is all 80% of people need: access to messages, a clear path to mandatory weekly re-filing, and most importantly, status updates, with a timestamp of when a given case was last reviewed. If people see that their case was reviewed in the last 24 hours, they’re going to be far less likely to frantically call the unemployment hotline fifty times a day. Everything else is relegated to the menu — at least until we get on the other side of the pandemic.
One other small improvement: the weekly certification makes you go through eight screens to answer the same eight questions each and every week when doing the “weekly certification.” It’s so annoying that I couldn’t help myself: I put them all on a single page, reducing unnecessary page loads on the state’s server, caching last week’s answers, and dramatically speeding up the process for the user:
Finally, the myriad other links from the existing dashboard can fall under the main hamburger menu at top, with simple, concise language, like so:
Coda: The Final Piece
These changes alone would iron out the a significant share of the problems faced online by the 25% or one-million-plus Louisianans unemployed during the pandemic.
But one huge, dreaded puzzle piece remains: login and sign-up. What’s often discussed as a relatively straightforward, consult-the-best-practices kind of thing gets a little more complicated when you’re dealing with such an objectively broad slice of the populace. I linked to the authentication problems Louisiana dealt with in the spring above. But better yet, quick: think of the username and password management of 97% of people you know above the age of sixty. Each of my parents spend at least an hour dealing with simple login issues every week. Even after I explained having a secure password schema and put LastPass on my parents’ computers, they’re still consulting a crazy, scratched-out list of passwords in back of a faux leather address book my family’s been using since 1993. Multiply that by the sensitivity of a process that requires 100% of low-literacy users to submit their social security numbers, and you have a nightmare that stewed in my unconscious literally for months.
So: should those seeking unemployment create unique usernames? Register by email address? Or strictly by SSN? What about phone number? Could the site accept ID numbers? Should we add a sign in with Google/Facebook/Twitter? It goes on and on! After abandoning the problem as worthy of its own entirely separate case study for months, suddenly it occurred to me: what would Canada do? It appeared that they used a combination of SSN and mother’s maiden name. When I told this to my friends who are already professional, senior UX designers, their response was basically: too dangerous! Too easy to hack! Maybe the SSN + mother’s maiden name was just a thing for registration, which then prompted you to create a new, secure password. So I asked my Canadian friends. I’m still waiting to hear back.