# Building a product from A to Z
Youtrust · 2022 · Lead Product Designer · Case study
Source : https://albancarmet.com/case-study/youtrust-atoz
Launching an automation product that generates documents to sign, from discovery to delivery.
In one line : Shipped in 6 months, 22 hypotheses tested
Team : PM, EM, 3 engineers and me · Duration : 6 months · AI : None

## At a glance

- Problem : Before a document gets signed, its content takes rounds of back-and-forth and error-prone data collection. Youtrust only covered the signature itself.
- What I did : Designed Workflows end to end with the founders of Canyon, a company we had acquired: flows, prototypes, then two rounds of usability tests with 11 users.
- Result : Shipped in six months, on time, with all 22 test hypotheses validated after round 2. The commercial target came five months late.
- Limit : The tests missed what blocked adoption a year later: preparing the document outside Youtrust.

## The company

Youtrust (formerly Yousign) is a European digital trust platform: electronic signature, identity verification and electronic seals. More than 30,000 companies across Europe use it, from Allianz and Fnac Darty to Payfit and Qonto.
- Sector : B2B SaaS · e-signature
- Customers : 30,000+ companies in Europe
- My time there : 2021 – 2023 · Lead Product Designer

## The product : Workflows

Workflows, later renamed Forms, is the document automation product I designed at Youtrust. It is built for legal and HR teams who send the same document over and over, such as a non-disclosure agreement. It was Youtrust’s first product beyond e-signature, upstream of the signature itself.
- Prepare the template : In Word or Google Docs, the creator puts what changes each time in brackets: [Company name], [Address].
- Get a form : Once the document is uploaded, Youtrust turns each bracketed variable into a question.
- Fill in, then sign : Each answer to the form produces a completed copy of the document, sent straight to e-signature.
Example: legal prepares the NDA template once. For each new partner, a sales rep fills in the form with the partner’s details, and the NDA goes out for signature, with no copy-pasting in Word.

## Process

Each step fed the next. None of them used AI: this was 2022.
- Understand : Interviews with 9 customers, most run before I joined; I took notes on some. (without AI)
- Scope : A first release that had to earn every feature, focused on legal teams. (without AI)
- Design : Flows and prototypes, with the founders of Canyon, the company Youtrust had acquired. (without AI)
- Test : Two rounds of usability tests: 6 sessions, a pause to fix what kept coming back, then 5 more. (without AI)
- Ship : Launched in six months, on the planned date. (without AI)

## The problem — Before the signature, documents still went through email and copy-paste

Youtrust only covered the act of signing. Before that, data was collected by email, copied into Word by hand, and approved over rounds of back-and-forth.
The goal: more value before the signature, and a new revenue line. Youtrust had acquired Canyon, a Belgian company with expertise on this exact problem, but its product couldn’t be transposed as is.
Of everything I have led, the hardest project: an intricate subject that pulls templating, data collection, validation and signature into one flow, real technical limits, unusually many stakeholders for one feature, and a small team.

## The approach — Build on how people already work, and test until the issues stop repeating

Everyone we interviewed already wrote in Word, and handled the rest by hand. So the product had to fit around that habit rather than replace it, and one round of tests turned out not to be enough.

## The trade-offs — Three decisions, and what they cost a year later

The real test of these choices came a year after launch, when adoption stalled: two of them were reopened.

## What shipped — One flow, from the document to the signature request

Three parts of the first release, in detail.

## Constraints

- Six months, discovery to launch : The date was fixed, which forced a minimum marketable product: every feature had to earn its place in the first release.
- A small team, newly merged : Canyon’s former co-founders, one of them acting as both EM and PM, a few consultant developers — and a switch to autonomous product teams in the middle of the project.
- Many stakeholders, real technical limits : Product, engineering, finance and marketing all had a say, and several interactions had to be redrawn around what the existing product could support.

## Signals and research

Most problem interviews had been run before I joined; I sat in on some as note-taker. The design rests on them, on what Canyon’s founders already knew, and on two rounds of usability tests I ran myself.
- Problem interviews · 9 customers — Data collection came first : 5 of 9 named it their top problem. Legal and HR met the same pains, and none of them felt at ease with technology. (5/9 : named data collection their top problem, across legal and HR.)
- Same interviews · tools — Everyone wrote in Word : All 9 used Word, and 6 handled the whole process by hand: email to collect data, copy-paste into the template. Tools that built their own editor had bad reviews for breaking documents. (9/9 : wrote in Word, and 6 handled the whole process by hand.)
- Canyon’s founders — Expertise, not a product to copy : The acquired team knew the problem well. But their interfaces and dependencies were too different: we rethought the product almost from scratch inside Youtrust. (Canyon : knew the problem well, but its product had to be rethought almost from scratch.)
- Usability tests · 11 users — One round planned, two run : After 6 sessions, the same issues kept coming back. We paused, fixed them, and resumed with 5 more users on the new version. (11 : testers: after 6 sessions the same issues kept coming back, so we fixed them and tested 5 more.)

## Decisions and trade-offs

- Text editing in Word or Google Docs rather than an in-app editor. Rejected : An editor inside Youtrust — Build a text editor into the app, to create templates and variables without leaving Youtrust. Kept : Variables in brackets, in Word — Users mark variables in their own editor with square brackets; the upload page explains how, with a sample and a video. Why : Time and resources — and the interviews warned that a homemade editor breaks documents and is hard to adopt. Every interviewee already wrote in Word. What it cost : Control over templates and versioning, and editing without leaving Youtrust. In hindsight : Brackets passed both test rounds. A year later, document preparation had become the top adoption blocker, and users said they expected to edit in Youtrust.
- Form first, as stakeholders expected, rather than document first. Decided with : Key stakeholders. Rejected : Document first — Place signers on the document, then customize the form in one go — its preview can then include the signer fields. Kept : Form first — Customize the form right after upload, place signers, then come back to review the form — the order people expected. Why : In the feedback rounds, document first felt less intuitive: to key stakeholders, the expected step after upload was the form. Splitting the form in two steps also lowered the density of each page. What it cost : The flow I had designed to avoid a back-and-forth: users now leave the form, place signers, then return to it. In hindsight : In round 2, 2 of 5 testers were surprised to return to the form. A year later, the adoption redesign merged those steps into one.
- A focused first release rather than a wider scope. Rejected : A broader first release — Cover sales use cases too, and ship with self-serve onboarding. Why : Six months and a small team forced choices. We focused on legal teams, one of the two profiles found in research, with HR close behind. What it cost : Coverage of some customer profiles, and self-serve: Sales had to demo the product to sell it. In hindsight : Shipped on time, but the commercial target was reached five months late. The first feature set was probably too thin for some customers.

## What shipped

- Public or private forms (Choosing who can fill in a form, in words people understand.) : problem : In round 1, 4 of 6 testers didn’t understand the visibility options: to reach every employee, a form had to be public, since few of them had a Youtrust account. Solution : We rewrote the options as Private and Public, and redrew their icons. In round 2, all 5 testers chose correctly.
- The version that will be signed, made explicit (When a document has several versions, no doubt about which one goes out.) : problem : In round 1, 3 of 6 testers didn’t see which version would be prepared for signature — a risk of sending the wrong document. Solution : We reworked the hierarchy: a tag on the version to be signed and an accent on its title. In round 2, 4 of 5 testers got it right.
- Form answers in the notification email (The collected data, readable without opening the app.) : problem : The person preparing the document had to open Youtrust to check what respondents had entered. Solution : The notification email lists every answer. 4 of 6 testers found it really useful — for some, the moment the product clicked.

## Results

Shipped on time. The commercial target came five months late.
(Usability tests · 2 rounds, 11 users (7 customers, 4 prospects) · launch in June 2022)
- 6 months : From discovery to launch, on schedule
- 18/22 → 22/22 : Test hypotheses validated, round 1 → round 2
- 2/6 → 5/5 : Understood who can use a form (public or private)
- +5 months : Delay on the Q2 2022 commercial target, reached in November
The commercial target itself stays confidential. We traced the delay to three gaps: no self-serve, must-have features still missing, and a positioning and pricing not yet in place.
What worked:
- Launched on time, despite a reorganisation into product teams mid-project.
- Testers understood the product and its value, and stayed engaged throughout.
- Form answers inside the notification email: 4 of 6 testers found it really useful.
What didn't work:
- The commercial target, reached five months late.
- Document preparation passed every test, then became the top adoption blocker in real use.

## What I took from it

Passing tests isn’t adoption. A prototype tested in a session can’t show what happens when users leave the app to prepare a real document. Closing that gap between test conditions and real use is what the follow-up project, Driving product adoption, set out to do.