Letting people compare Medicare plans before they ever pick up the phone

A concept I pitched, not a product I shipped

This is an exploratory concept. I want to be direct about that up front: it was not built or released while I was at GoHealth. I designed it because I kept seeing the same gap in the work I was actually shipping, and I wanted to show what closing it could look like.

I'm including it because the problem it addresses is real and the direction held up. After my time there, the market — and GoHealth itself — moved toward exactly this kind of self-service plan comparison, which reinforced that the problem was worth solving.

Every road led to a phone call

I spent my time at GoHealth designing the front of the funnel: Medicare lead forms in English and Spanish, blog entry points, eligibility messaging. Those forms did their job — they collected enough information to hand someone to a licensed agent.

But that was the only ending available. Someone could read three articles, understand the difference between HMO and PPO, know exactly which doctor and which prescriptions mattered to them, and still have no way to see a single plan. The next step was always "talk to an agent."

That's a mismatch with how people shop now. Plenty of people 65 and up want to look first and talk second — and the ones who do call arrive cold, with no shortlist and no vocabulary for what they want, which makes the call longer for everyone.

• No self-serve pathComparison only existed inside an agent conversation.
• Trust gapAsking for a phone number before showing any plan lowered trust — it felt like a sales gate, especially for people new to Medicare.
• Real constraints ignoredDoctors, prescriptions, and pharmacy are what actually decide a plan — none were inputs.

Show plans first, make the call better second

The premise wasn't to replace agents. It was to let people arrive at the call already informed — and to let the people who never would have called at all get somewhere useful on their own.

So the tool had to answer the three questions people actually ask, in their own words: Can I keep my doctor? Are my prescriptions covered? What will this really cost me for the year?

Comparing plans shouldn't require a phone call. The call should be for the part a person can't do alone.

Two people I designed for

These personas came directly out of the Medicare research study I ran with 40 people — Katherine, who re-shops every year and dreads it, and Rosie, who had just turned 65 and researched online before calling.

They earned their place by settling arguments. Every filter in the tool exists because one of them needed it, and anything neither of them needed stayed out.

A second destination for the content funnel

GoHealth already had strong top-of-funnel content — the problem was that all of it emptied into the same lead form. So the tool sits one level below home as its own branch, linked from articles and from the form itself, rather than living inside the shop flow where it would read as a step in a sales sequence.

The path, plus the states that usually get skipped

The happy path is six steps and stays on one page — filters, results, compare, next step. What mattered more to me was the unhappy paths: an empty result set, a drug that isn't in the formulary, a county with one plan. Those are where a 68-year-old decides the tool is broken and picks up the phone frustrated, so I designed them alongside the main flow instead of after it.

The proposed homepage: ZIP code, nothing else

This was the proposed entry point. The homepage asks for one thing — a ZIP code — and the View Plans button takes the person straight into the comparison tool.

No name, no email, no phone number at this stage. Asking for contact details before someone has seen a single plan is the friction that sends people away: most are still researching, and a form that reads as a sales gate makes them close the tab or brace for a call they didn't ask for. A ZIP code is the one input the tool genuinely needs, because plans are priced county by county.

Contact information still has a place — just later, once someone has a shortlist and an actual reason to talk to a licensed agent. Then it's an ask they say yes to instead of a toll they pay upfront.

The 'Find your plan' entry point was also a small trust signal. In the research, 30 of 40 participants said they felt anxiety about choosing the wrong plan, and several said they wanted to see options before giving contact info. Letting people look first — especially people new to Medicare — would have made them more confident and more likely to trust the next step.

Low-fidelity mockup

The hero drops to a single line, "Find your plan," with nothing under it, because the tool itself is the explanation.

Below that: one comparison block holding every input, then a stacked list of filtered plans, then a help block. One column, one decision at a time, no side rails competing for attention.

The branded concept

The designed version uses GoHealth's blue, its type hierarchy, and its existing header and footer patterns so it feels native to the site rather than bolted on.

ZIP code comes first, since plans are county-specific. Doctors and prescriptions are optional but sit high, because they're the filters that actually change someone's decision. Premium, carrier, hospital or clinic, pharmacy, and plan type follow as plain-labeled dropdowns. One prominent Search plans button, plus a Start over that removes the fear of getting it wrong.

Results lead with what the person asked for — "Your doctor is in network," "Covers all 3 of your prescriptions" — not with a spec table. Premium and max out-of-pocket sit together, because a $0 premium plan isn't automatically the cheaper one.

The first plan is shown with See plan details opened, so you can see what the expanded state holds: primary care and specialist copays, hospital and urgent care costs, drug tiers, dental, vision and hearing, fitness, and the yearly cost estimate. It expands in place rather than opening a pop-up — people 65 and up lose their spot when a modal covers the page, and expanding inline lets someone keep two plans open and read down the page.

Built on the Medicare study, not on a hunch

Every filter, every persona, and the "Need more help?" prompt come out of the Medicare user research study I ran with 40 people. That study is a separate case study — it shipped as research; these recommendations didn't.

The strongest findings pointed the same way: people wanted to look before they talked, they judged plans on their own doctor and their own prescriptions, and they disengaged the moment a page asked for more than they were ready to give.

Most of this audience starts on a phone

The phone screens weren't a resize of the desktop file. Three inputs stay visible — ZIP, doctor, prescriptions — and the remaining five filters collapse behind a single "More filters" control, so the first screen never looks like a form to survive.

On results, the filter summary stays pinned at the top as a tappable line ("ZIP 60601 · 1 doctor · 3 drugs") so changing an answer is one tap instead of a back-and-forth. Cards lead with premium and out-of-pocket, and every tap target clears 48px with real spacing between them.

Compare what matters most

Every plan card carries a Compare plan box. After someone selects two plans, the checked boxes carry into the comparison view so it's obvious what's being compared and easy to swap one out.

I designed the comparison for both desktop and mobile. Desktop uses a familiar side-by-side table; mobile keeps the same two plans in fixed columns with shorter labels so people can compare without losing context.

The rows aren't a spec dump. They're the questions people asked in research, in that order: premium, max out-of-pocket, estimated yearly cost, doctor, prescriptions, pharmacy, extras, plan type. Katherine's version of "which one is cheaper" is the yearly cost row, not the premium row.

A "Need more help?" prompt when someone stalls

In research, people who got stuck didn't ask for help — they left. So I designed a prompt that appears when someone has been on the page a while without interacting: a small panel offering a licensed agent, with the phone number, TTY line, and a plain "No thanks, I'll keep looking" that dismisses it.

It's deliberately not a full-screen takeover and it never appears mid-typing. It's help offered at the moment of hesitation, not a sales gate — the same idea I proposed for the Medicare lead forms after the research study.

Designed for 65+

This audience is the reason the interface is calm rather than clever. Every choice below was a constraint I set before drawing anything.

• Type size16px minimum body, 18px labels and results — no fine print anywhere in the flow.
• ContrastText and controls meet WCAG AA against white and against the blue fields.
• Targets48px minimum inputs, 56px primary buttons, generous spacing so mis-taps are unlikely.
• Real labelsVisible labels above every field instead of placeholder-only inputs that vanish on typing.
• Keyboard and screen readerNative selects and inputs, visible focus rings, and a live region announcing the result count.
• Plain language"Add a doctor," not "provider network lookup." Help text under fields that need it.

Why it never shipped

This didn't get built while I was at GoHealth. The team was already focused on the lead funnel I was shipping, and a plan-comparison tool needs carrier data, accurate pricing, and compliance review — a much heavier lift than a lead form.

It still shaped how I argued for self-serve paths. It's the example I use when a team assumes the only conversion path is to capture a phone number first, and the industry's move toward comparison tools since then suggests the instinct was right.