Take the money last — ProblemShared self-pay

01 Context

Buying healthcare is not shopping.

ProblemShared is an NHS-partnered, CQC-registered provider of neurodevelopmental assessment. Most people reaching the self-pay route had already waited a long time for an NHS answer. They weren't browsing — they were trying to stop waiting.

The person

Anxious, informed, out of patience

Adults who suspect ADHD or autism, and parents seeking an answer for a child. Often years into a wait, often after a difficult GP conversation. High motivation, low tolerance for another dead end.

The purchase

Four figures, no guarantee

A clinical assessment is a significant, non-refundable-feeling commitment. Anything ambiguous about price, eligibility or what happens next reads as risk.

The constraint

Clinically and legally bound

Not everyone is suitable for a remote assessment. Clinicians need specific information; regulation dictates what must be captured and recorded. The flow could be reordered, not shortened.

02 Research

I went looking for where people gave up.

Four sources, chosen because they were fast and already in the building. The most useful evidence wasn't in a research repository — it was in refund requests.

M1

Refund & support ticket analysis

Every refund reason and support conversation from the self-pay route, coded by cause. The single richest source: people telling us, in their own words, what they had misunderstood and when.

M2

Funnel & session review

Step-by-step drop-off across sign-up, service selection, intake forms and checkout — then watching sessions at the steps that leaked hardest.

M3

Interviews with people mid-journey

Adults and parents who had started and stalled, plus people who had completed. What they were afraid of, and what they were checking for before they trusted us with money.

M4

Clinical & ops working sessions

With assessors and operations: which data is genuinely needed to run a safe assessment, which is needed for compliance, and — critically — when each is actually needed.

01

People paid before they knew they were eligible

Suitability was checked late, and in some paths after payment. Refunds clustered around one sentence: I didn't know you might not be able to see me.

So what

Every pound taken before a yes was a refund waiting to happen — and a trust cost we couldn't buy back.

02

Sensitive questions arrived before trust did

Sex at birth, GP contact details and an emergency contact were requested early, with no explanation of why or who would see them. Several people stopped at exactly that field.

So what

The questions weren't the problem. Asking them before we'd earned them was.

03

The progress checklist resorted itself

Completed steps jumped position, so the list read 1, 3, 2, 4. Returning users lost their place and re-read the whole thing to work out what was left.

So what

A progress indicator that moves isn't progress. Order had to be fixed; only state could change.

03 The change

Same steps. Different order. Different product.

Nothing was removed — regulation and clinical safety wouldn't allow it. What changed was the sequence, and what we explained at each point.

Move 01

Suitability first, always free

A two-minute check moved to the front of the journey, before any payment or clinical intake. A clear yes, a clear not-for-us with an onward route, or a "let's talk first".

Nobody pays for an assessment we can't deliver

Move 02

Clinical intake after the yes

GP details, emergency contact and background moved behind eligibility and payment — where they're framed as preparing your clinician, not as a barrier to entry.

Sensitive data asked once, in context, with a reason

Move 03

A checklist that holds still

Steps keep a fixed order and only change state: done, in progress, or locked with the reason it's locked. Progress became readable at a glance from any device.

Returning users resume instead of re-reading

Product Getting started

One list, fixed order, honest states.

The first screen after sign-up answers three questions immediately: how far along am I, what's next, and what happens after I pay.

problemshared.net / get-started

ProblemShared

NBNaomi Bright

Hi Naomi — here's where you are

Four steps. You'll know if we can help before you pay anything.

Step 2 of 4 ~6 min left

✓ Check we're right for youFree · 2 minutes · no card needed Suitable

2 Choose your assessmentPrices and what's included Continue

3 Details for your clinicianUnlocks once you've chosen an assessment Locked

4 Book your first appointmentYour dashboard opens here Locked

What happens after this

So you know what you're buying.

Day 0You choose an assessment and pay

Day 0Short form so your clinician can prepare

Day 1–3We match you to a qualified assessor

Week 2–4Your assessment appointment

+10 daysWritten report and next steps

NHS partnerCQC registered650+ reviews

9:41

ProblemShared

Hi Naomi

2 of 4

✓ Check we're right for youFree · no card needed Suitable

2 Choose your assessmentPrices and what's included

3 Details for your clinicianUnlocks after you choose

4 Book your appointmentYour dashboard opens here

Continue — choose your assessment

01

Fixed order, changing state

Steps never move. Only their state changes, so the list is the same shape every time you come back.

02

Locks explain themselves

A locked step says what unlocks it. A greyed button with no reason reads as a bug, not a rule.

03

The purchase is made visible

Assessment, report and timings are shown before payment, because that's what people were asking support about.

Product Choosing & disclosing

Say the price. Say why you're asking.

Two patterns did most of the work: making the commercial offer legible, and explaining every sensitive question at the point it's asked.

problemshared.net / choose-your-assessment

Choose your assessment

You've been assessed as suitable. Prices include your report.

Not sure yet

Consultation call

£95 30 min

Talk it through with a clinician
Ask what assessment fits
Fee credited if you continue

Book a call

Most chosen

ADHD assessment

£1,195 total

Full clinical assessment
Written diagnostic report
Titration options discussed
Shared with your GP

Choose this

Waitlist

ADHD + autism

£1,795 total

Both assessments combined
One report, one clinician
Opening to new bookings soon

Join the waitlist

Not what you expected?

Your suitability check is still valid for 30 days — you can come back to this.

See what's included

Details for your clinician

Your assessor sees this before your appointment. It won't change your price or your booking.

Sex assigned at birth

Select

Clinical reason ADHD and autism present differently, and some are diagnosed far later because of it. Your assessor uses this to interpret your results — not to categorise you. Only your clinical team sees it.

Your GP practiceOptional now

Richmond Medical, Sheffield

Emergency contactSafety only

Name of someone we could contact

Save and continue

01

Price before commitment

The total, what's in it, and a lower-commitment route for people who aren't ready. Placeholder pricing was the biggest source of support contact.

02

Why we ask, inline

Sensitive fields carry a one-tap explanation of the clinical reason and who sees the answer. Written by me with the clinical team.

03

Nothing is a dead end

Not suitable, not sure, not now — each has a route onward rather than a stop. The suitability result stays valid for 30 days.

Process AI in the loop

How I worked: prototype the reorder, prove it's safe, ship it.

Reordering a regulated clinical journey isn't a Figma decision — it only counts if clinical and engineering agree it's compliant and buildable. I used Claude to prototype the new sequence fast, then spent my time on the conversations that actually de-risked it.

01 · Prototype

A clickable reorder, not a slide

I rebuilt the four-step journey as a working prototype with Claude — suitability first, payment, then intake — so the team could click through the new order instead of imagining it. The locked-step pattern and "why we ask" disclosure were far easier to argue for once people could feel them.

My judgementAI built the flow. Deciding what came first, and why, was the design.

02 · Pressure-test

Finding the failure paths early

I had Claude walk the journey as an anxious first-time user and as someone found not suitable — the path we understood least. That's where the "nothing is a dead end" rule came from: every negative outcome needed somewhere to go, and the prototype made those gaps obvious.

My judgementI decided which edge cases mattered clinically. AI made them cheap to explore.

03 · Validate

Clinical & engineering, on the real thing

The prototype went into working sessions with clinicians and engineers. Because it was runnable, we could check specifics: is it compliant to take payment before full intake? Can suitability write to the record this way? Answers came in one session, not a fortnight of tickets.

My judgementThe prototype turned "what if" into "here it is — does this hold up?"

Working session — the reorder

Me

Prototype the journey with a free suitability check first, before payment or any clinical intake. If someone's found not suitable, they must land somewhere useful, never a dead end.

Claude

Built. Suitable → assessment selection. Not suitable → a page explaining why, with NHS and alternative routes. "Not sure" → a paid consultation, fee credited if they continue. I kept later steps locked until suitability passes. Should the result expire?

Me

Yes — valid 30 days so no one has to redo it. Now show the intake form with a "why we ask" on sex at birth and GP details.

Clinical

The wording on sex at birth needs to be ours, not generic — it's clinically load-bearing. Send me the copy and I'll adjust it.

Eng

Taking payment pre-intake is fine; suitability just needs to write a flag to the record before checkout. Small change. Buildable this sprint.

Reconstructed from the working sessions. Prototyping in hours meant clinical and engineering reviewed something real — so the risky questions got answered before a line of production code was written.

04 Trade-offs

Three calls that weren't obvious.

Qualify people out, on purpose

A free suitability check placed first will turn some paying customers away before checkout.

Short-term conversion⇄Refunds & trust

Decision

Take the smaller, cleaner funnel. A refunded customer costs twice and tells people about it.

Explaining vs. shortening

Adding "why we ask" makes forms longer, and conventional wisdom says shorter forms convert.

Fewer words⇄Fewer abandonments

Decision

Explain sensitive fields, collapse the rest. Length wasn't the barrier — unexplained intrusion was.

Modals or a single page

Intake sat in a stacked modal with a seven-dot progress bar and no sense of scale.

Focus per question⇄Sense of the whole

Decision

Keep the stepped form, but show the number of steps and let people save and leave without losing their place.

05 Outcome

What changed, and what I'd do next.

Refunds moved upstream

Fewer people reaching payment before an eligibility answer, and fewer refund requests citing suitability

Intake completion improved

Fewer abandonments at sensitive fields once each one carried a reason

Support questions changed shape

From "am I eligible?" and "what does this cost?" to clinical and scheduling questions

What I'd do next

Give the suitability result a shareable summary people can take to their GP.
Test paying a deposit at selection with the balance at booking, to lower the entry commitment.
Extend the "why we ask" pattern into the clinical questionnaires themselves.
Instrument the not-suitable route properly — it's the path we understand least.

← Back to all work Next — Clinical AI co-pilot

Rahman Malik — Product designer ProblemShared · D2C self-pay