INT/CTG-06
INT/CTG-06
Cartiga
Cartiga
LegalTech
LegalTech
Litigation Funding Intake
Litigation funding runs on qualification. Before anyone gets money, somebody has to establish whether the case is fundable, and that means a long list of questions that exist for legal reasons rather than product ones. Here is the part that matters, and it took me a while to build the flow around it. The person filling this out is usually having the worst month of their life. They are hurt, or they are in a lawsuit, or they need money by Friday. Sometimes all three. The old intake asked everything, of everyone, up front. People stopped partway through. Attorneys chased the rest by phone, which is the most expensive way to collect a form field ever invented.
Litigation funding runs on qualification. Before anyone gets money, somebody has to establish whether the case is fundable, and that means a long list of questions that exist for legal reasons rather than product ones. Here is the part that matters, and it took me a while to build the flow around it. The person filling this out is usually having the worst month of their life. They are hurt, or they are in a lawsuit, or they need money by Friday. Sometimes all three. The old intake asked everything, of everyone, up front. People stopped partway through. Attorneys chased the rest by phone, which is the most expensive way to collect a form field ever invented.
You can’t delete a question the law requires. You can decide when to ask it, and who has to face it.
You can’t delete a question the law requires. You can decide when to ask it, and who has to face it.
Progressive disclosure
Progressive disclosure
A/B testing
A/B testing
Underwriting
Underwriting
INT/CTG-06
INT/CTG-06
Role
Role
Senior Product Designer, Creative Director
Senior Product Designer, Creative Director
Deployed
Deployed
2022 to 2023
2022 to 2023
Domains
Domains
Intake / Attorney CRM / Document handling
Intake / Attorney CRM / Document handling
Team
Team
Two direct reports
Two direct reports
Constraint
Constraint
Compliance questions that could not be removed
Compliance questions that could not be removed
Tools
Figma, Claude Code, Cursor
Measured
FullStory, Google Analytics, marketing reports
20%
20%
20%
20%
20%
more qualified leads
more qualified leads
35%
35%
35%
35%
35%
less drop-off
less drop-off
01
01
The artifact
Four gates, five checks each. Nothing gets called done until all four sign the same version.
Every intake feature ran past four reviews before anyone called it finished: UAT, PM, stakeholder, compliance. Each gate checked the same five things. Usability. Responsive logic. Whether the integration actually worked front to back, not just in the mockup. Compliance. ADA.
That order isn’t decorative. Compliance last means the legal review is looking at something real people have already used. Compliance first means every usability decision downstream gets made inside a box somebody drew before anyone tested anything.
ADA sitting in the same checklist as compliance is deliberate too. Both get treated as legal requirements, because in this product both are, and neither survives being the thing you get to later.
It stuck. The CRM dashboard went through it, and so did paralegal onboarding, which is where it earned its keep. Internal tools are exactly where a checklist like this gets skipped, on the reasoning that the users work here and can be trained.
02
02
The work
Three surfaces, one pipeline, two very different kinds of user.
Intake became a dynamic form with conditional logic. You answer what applies to you and nothing else. Questions that don’t apply never render, so the form gets shorter as the system learns more about the case instead of longer.
The attorney CRM sits on the other end. Lead status, case tracking, funding requests, one view instead of scattered across email and spreadsheets. Document upload feeds the same pipeline, so a funding request and its supporting files arrive together rather than needing to be matched up by hand later.
The other end of the pipeline. Attorneys see lead status, case tracking, and funding requests in one place instead of four inboxes.
Then paralegal onboarding, which was the surface I expected to be easy and wasn’t. Paralegals are the people actually moving cases through this thing all day. They’re trained, they’re fast, and they have almost no patience for a system that makes them think. Designing for somebody in distress and designing for somebody doing this four hundred times a week pull in opposite directions, and both flows feed the same records.
The hardest part was cognitive load, not compliance. A required question is a fixed cost. What you control is how many of them somebody has to hold in their head at once, and whether the next one arrives before they’ve recovered from the last.
So the flow got built for somebody with about forty percent of their usual attention. One decision per screen where the decision is heavy. No question that requires reading a previous answer to make sense. Nothing that punishes you for stopping and coming back an hour later.
Five steps, and the second one is the hard one.
The worst screen was choosing an attorney.
We couldn’t pick one for you. Legally we’re not allowed to, and that isn’t a rule anybody was going to bend. But the alternative, as originally scoped, was handing somebody in the middle of a bad month a list and asking them to rank lawyers. That is a lot to ask of a person who came here because they need money, not because they wanted to evaluate legal representation.
What shipped was the top three near you, sortable. Not a ranking task, a short list with the sort under your control if you want it. Three is small enough to hold. Nearby is a reason you can understand without explanation. And sortable means the people who do want to compare can, without making everyone else do it.
That screen is where the constraint and the emotional state collide most directly, and it’s the one I’d still like another round on.
03
03
How the work actually moved
The work started with analytics. Audit the funnel, find where people leave, then go ask why. Interviews on both sides: applicants who dropped out, and the attorneys chasing them.
When somebody said a question was mandatory, I checked it. Every time, against the objective, the end goal, and the actual compliance requirement.
Not because I doubted anyone. Because “we need this” and “the law requires this” get said in the same tone of voice, and only one of them is immovable. Once you know which is which, you know what you’re allowed to sequence, defer, or drop.
That verification is what bought the room to design for the emotional state. You cannot lower cognitive load on a flow where every question is untouchable. You can lower it a lot on one where six of them turn out to be preferences.
Workshops with attorneys, engineering and support ran weekly throughout.
04
04
What I didn’t get to do
I never got to test with a broad enough range of people.
That’s a real problem on this product specifically. The whole design rests on an assumption about who is filling the form out and what state they’re in. Somebody stressed, distracted, possibly in pain, working on a phone, maybe not in their first language. I designed for that person. I validated with a narrower group than that person represents.
So the cognitive-load decisions are reasoned rather than proven. I still think they’re right. I’d rather know.
The checklist helps here in a way I didn’t plan for. UAT is one of the four gates, which means every feature after this one gets in front of somebody before compliance signs it. That doesn’t fix the range problem. It does mean nobody ships this flow again without a real person touching it first.
I designed for somebody I never got enough of in a room.
05
05
Outcome
Qualified leads up 20%. User drop-off down 35%.
Fewer people abandon the form, and more of the ones who finish are worth funding. Those two moving together is the whole point. A shorter form producing worse leads would have been a failure dressed up as a win.
Attorneys stopped chasing incomplete applications by phone. Support stopped fielding “which document goes where.”
What I’d do differently: trace every requirement to its source before designing against it, not partway through. Most of the room I found came from asking where a rule originated, and I was doing that reactively. Start there and the flow gets built once instead of twice.
06
What I wanted to show you
Regulated products are where most design advice stops working. You can’t delete the question, you can’t reword it into something friendlier, and the person insisting on it is usually right.
What’s left is sequencing, and who has to carry the weight of it. On a product where the user is already having a bad month, that turns out to be most of the job.
ケビン・チャップマン
No. 26 / Lot 01 / SS26