Taxfix wanted to go beyond DIY

Taxfix is a free, DIY tax-filing app used across Germany — filling the gap between doing your own taxes and paying someone else to. In 2021 the business set out to add a tier in between: DIY (do it yourself, Taxfix's core free product), DIWM (do it with me — a real expert reviews your return with you), and DIFM (do it for me — full-service filing, fully handled by someone else). The new line of services was called Assisted Tax Filing, with Expert Review as its first block. I joined at initial product conception as the sole product designer, in a team of one PM, five developers, plus legal and business leadership.

This wasn't only a UX problem

Two constraints shaped the project from day one:

1. No license to advise — Taxfix isn't licensed to give tax advice, so the review itself had to run through a partner tax-experts firm, not Taxfix.
2. Can't take the payment — Taxfix wasn't legally allowed to receive money for this specific service, so the partner firm had to be the visible party at checkout, not us.

A third, non-legal constraint sat alongside those: finding the balance between a tailored, human experience and something that could actually scale.

Who it was for

Expert Review was aimed at people whose life just got more complicated — moving countries, a change in marital status, a new baby — exactly the moments a self-serve flow starts to feel risky to get wrong.

The team, at the start and as it grew

I started as the sole product designer, alongside one PM, five developers, and legal and business leadership. As the project matured, the team grew to include two content designers (EN/DE), a service designer, and a qualitative researcher — I went from a team of seven to owning the user flow within a full design chapter: happy and unhappy paths, interaction design between screens, content-first pieces like the wizard carousels, gathering feedback from legal, business, and engineering, test protocols, proposing A/B tests, a clean handoff to engineering, and — post-launch — setting research objectives and prioritizing what came next.

Aiming at an MVP

The North Star was feedback delivered and actioned fully in-app — including a platform for reviewers to mark feedback natively in the declaration. For the MVP, I helped make a deliberate call to ship something rougher instead:

1. User submits their return, plus any questions
2. An expert reviews it offline
3. They return a PDF report, severity-tiered: Change (a mistake to fix), Check again (confirm it's right), Optimize (optional, not required)
4. User applies the changes back in the app
5. Resubmit to tax offices

We were aware the PDF-plus-manual-changes flow wasn't optimal — but we chose it deliberately. It let us prove the process with the partner and gather the most real input from clients, with the in-app version as a deliberate next step, not a compromise we didn't notice. I even designed and shipped a small beta website — built in Webflow — so experts could generate that PDF report through a simple web form.

Getting the offer in front of the right users

I mapped every point in the filing journey where Expert Review could realistically be offered, and shipped three, each with skip-logic so declining never blocked the core free product:

1. A card on the result page — "Get your return checked by a certified tax advisor before you file for just €49.99 extra."
2. A loading-screen offer, if they continued without it
3. The option to add Expert Review later, from an already-submitted declaration

Process

Discovery ran from a proof-of-concept workshop with the PM and business team through personas, concept tests, and a first flow, into mid-fi review with product and business leadership, iteration, an engineering walkthrough, high-fidelity flow, usability testing with severity-scored issues, and a clean Figma handover. Testing ran with a native German speaker ahead of launch, in the product's home market.

Result

What shipped: a working MVP end-to-end — the opt-in flow, skip-logic across the entry points, the severity-tiered PDF mechanism, the beta expert platform, and a payment flow that correctly foregrounded our legally-required partner.

What I did after launch:  I set post-launch research objectives, gathered MVP feedback, helped define our KPIs, and prioritized fixes for the next version.