What is sludge?
Maria has been on Medicaid for three years. Each year, the state sends a renewal packet, she fills it out, and her coverage continues without incident. This year is different. When she moves to a new apartment, she logs into her state’s benefits portal to update her address, because she knows how important it is to keep her information current. The portal has more than one place where an address can be entered, one tied to her core profile, and another tied to her active case. Maria updates the one that looks most like “my account settings.” It seems like the obvious choice. It is also the wrong one. The renewal notice goes out to her old address, and she doesn’t know she’s supposed to respond. She doesn’t know a deadline has passed, and she finds out she’s lost coverage the way a lot of people do: at the pharmacy counter, when her prescription is suddenly $140 instead of $10.
Maria hasn’t become ineligible for Medicaid. Nothing about her income, her household, or her circumstances has changed. What’s changed is that a form didn’t reach her. She spends the next several weeks navigating a call center, resubmitting paperwork she’s already submitted once, and going without medication in the meantime. Eventually, she’s re-enrolled, approved for the exact same benefit she had before, just with a gap in the middle.
Nobody decided this should happen. No policy says people should lose coverage they’re entitled to and then have to fight to get it back. But it happens constantly, and it happens because of something that rarely shows up in a policy debate or a budget hearing: bad user experience (UX).
Legal scholar Cass Sunstein and economist Richard Thaler gave a name to the experience Maria went through. They called it sludge: unnecessary friction in the form of paperwork, confusing forms, and procedural hurdles that costs people time or money, that can be “frustrating, stigmatizing, or humiliating,” and that too often ends up cutting people off from benefits, opportunities, and services they’re entitled to.
Sludge isn’t limited to government. It shows up anywhere a process is harder than it needs to be. Like an app that hides the ‘cancel subscription’ button behind too many clicks, or a form that asks for the same information twice. But government has a particular relationship with sludge, because government sludge often touches the things people can least afford to lose: health coverage, unemployment benefits, housing assistance, disability support.
And critically, sludge isn’t free. Somebody always pays for it.
The cost of sludge
Every organization that puts a confusing form, a broken login, or a dead-end web page between a person and what they need is generating a cost. What differs is whether that organization feels the cost.
In the private sector, the cost is loud.
If an online checkout is hard to use, a share of customers will abandon their cart and buy from a competitor instead. The company doesn’t need a sophisticated study to notice, because the effect shows up in conversion rates and revenue within days. That visibility is exactly why private companies invest in UX: bad experience translates almost immediately into lost business, and lost business is a language every executive speaks fluently.
This isn’t hypothetical. Poor UX has cost companies billions of dollars.
In government, the cost is quiet but it isn’t absent.
Government mostly holds a monopoly on the services it provides. If a state’s unemployment portal is confusing, people can’t take their unemployment claim to a competing state. If a benefits application is broken, there’s no alternative agency to apply through instead. So the signal that a private company gets automatically in the form of customers leaving never arrives. That doesn’t mean the cost disappears. It means the cost goes somewhere else, and stays hidden until someone goes looking for it.
Sometimes the cost is absorbed internally. A confusing online login process pushes people toward paper forms instead, which then have to be manually keyed in by a caseworker who could have been doing something else. Web content that doesn’t actually answer people’s questions doesn’t stop the questions from existing, it just moves them to a phone call, which is far more expensive to handle than a page of clear text. The organization still pays. It just pays through a call center budget or a caseworker backlog instead of a product budget, which makes the connection to the original bad experience almost invisible from where the money is actually spent.
Sometimes the cost is absorbed by the person first, then boomerangs back at a markup. This is what happened to Maria, and it happens at scale. When someone loses a benefit for procedural reasons (a missed form, a lost letter, a confusing portal) rather than an actual change in eligibility, and then has to reapply, it’s called churn . Processing that reapplication isn’t a quick reactivation; it’s a full new eligibility determination, run through the same systems and staff time as a brand-new applicant. Research estimates the administrative cost of a single churn episode (one disenrollment and one re-enrollment) at somewhere between $400 and $600. Multiply that by the number of people who churn in a given year, and you get a state absorbing millions of dollars to re-process people who were never actually ineligible in the first place. The state pays twice for a service it only needed to deliver once, and the person pays with a gap in coverage in between.
Why doesn’t this get noticed sooner? Mostly because of organizational silos. The team that owns the website usually isn’t the team that owns the call center budget, and neither of them typically owns the re-enrollment process. Each team can see its own numbers without ever being in a position to connect those numbers back to the web form that started the chain reaction. The cost is real and it’s being tracked, just never in the same spreadsheet as the thing that caused it.
The fix: teams built to see across silos
None of this is inevitable. It’s the predictable result of how the work is organized, which means it can be fixed by how the work is organized. Two things are especially important:
- Empowered leadership that can see across silos. Someone needs the mandate and the access to data to look at the website team’s numbers, the call center’s numbers, and the re-enrollment team’s numbers side by side, and ask whether they’re actually telling the same story. Without that role, each team is optimizing its own piece in isolation, and the connections between friction and downstream cost never surface.
- User experience research that goes looking for the friction, on purpose. Sunstein proposed a concept for exactly this practice: a sludge audit. Instead of waiting for cost to show up somewhere else in the organization, a sludge audit means deliberately walking through a service to find the points where people get stuck, give up, or make an error, and asking what that friction is actually costing downstream. Done well, this turns an invisible, distributed cost into a specific, named, and fixable one.
Neither of these requires a bigger IT budget. They require deciding that user experience research is part of how government delivers services, not a nice-to-have layered on top once the “real” work is done. Teams sometimes hesitate to invest in this kind of research because it feels like it will slow delivery down. The more accurate framing is that not doing it doesn’t make the cost disappear, it just shows up later, bigger, and in someone else’s budget.
Back to Maria
Instead of spending money making Maria’s renewal process clearer and ultimately saving money, the government spent far more reprocessing her application, staffing the call she made in frustration, and absorbing the cost of a system that quietly assumed no news was good news. Maria, for her part, paid in a different way: a gap in coverage, a scramble at the pharmacy counter, hours on hold.
The cost of sludge was always going to be paid by someone. The only real choice government has is whether to pay it upfront, through the comparatively cheap investment of good design and user research, or pay it later, at a markup, in the form of phone calls, paperwork, and people falling through the cracks of systems that were supposed to catch them.