Petit Duc
B2B SaaS · Marketing Website Redesign
Shipped
A sustainability SaaS wanted to move upmarket with an unfinished product and mostly regional proof. I helped redesign its marketing site around two questions buyers kept asking: Will this help me? And can I trust it?
Quick facts
Company
Petit Duc
an event-sustainability SaaS platform by Merisia (Sherbrooke, Québec)
My role
Research and discovery lead Client relationship lead
Design system owner
Page designer (home + price)
Team
3-person design team
in partnership with Merisia Founder & CEO, Benjamin Rancourt
Timeframe
5-week sprint
Shipped; launches live site within three weeks
METHODS
Heuristic audit · competitive audit · expert interview · survey (n=7) · secondary research · proto-personas · moderated usability testing (n=3)
01 · The problem
Context and problem
The product worked. But the website wasn’t earning the trust needed to sell it to bigger buyers.
Petit Duc helps event organizers measure and reduce their events’ carbon footprint, with a focus on participant travel. It had about 20 pilot clients, almost all regional, and the company wanted to move upmarket to larger events. The existing site was confusing and cluttered, offered two vague CTAs (“Join the adventure” / “Join the waitlist”), and showed no product. A first-time visitor couldn’t answer three questions: What is this? Can I trust it? Is a demo worth my time?
What made this hard was not the visual refresh. It was that the two things buyers trust most, product proof and customer proof, were the two things Petit Duc didn’t have yet. The platform was still in development, so there were few screens to show, and there were no case studies or testimonials. The real problem: how do you earn a skeptical B2B buyer’s trust before the proof exists?
Constraints we designed inside
• Unfinished product: no existing screenshots or visual proof of concept
• Thin, regional proof: ~20 pilots (Québec-only), no case studies
• Content didn’t exist yet: much of the copy had to be drafted as placeholder
• 5 weeks, 3 designers splitting pages, max 3 research participants per the brief
02 · Discovery
Discovery
Buyers weren’t shopping for sustainability expertise; they were shopping to outsource a lower priority.
What we set out to learn
I wrote the research plan as eight assumptions to test rather than questions to answer, because the brief came with a lot of built-in beliefs (that visitors understand sustainability, that carbon measurement is the core need, that one visit is enough to book a demo). We ran a heuristic audit, a competitive audit of two direct and four adjacent competitors, one in-depth expert interview with an enterprise event leader, a 7-person survey of event planners, and a secondary scan of industry and regulatory sources (EIC, MPI, GDS-Index, CSRD). With so few participants, I treated survey numbers as directional and looked for places where the sources agreed.
Carbon measurement is the core need
Partly. Buyers care more about reporting to stakeholders, staying on budget, and not doing extra work
Visitors are ready to book after one visit
No. They want a low-commitment contact option, then a human
Case studies are the strongest trust builder
Partly. Seeing the actual product mattered as much, and Petit Duc had neither yet
We compete with other carbon platforms
Mostly with spreadsheets and doing nothing
Findings that changed the design
01
Sustainability ranks last, so don’t lead with it. Budget, guest experience and venue were each top priorities for 5 of 7 planners; sustainability was a priority for only 1. Every respondent named “not prioritized” as a barrier. Sell relief and ROI, not virtue.
02
The real competitor is the spreadsheet. 4 of 7 track with spreadsheets, 1 uses software, and 5 of 7 were “not confident at all” measuring impact.
03
The buyer isn’t always the user. The site had to equip an internal champion to sell it upward, not just persuade one person.
04
Ready to talk isn’t ready to book. That shaped a two-tier CTA strategy rather than one hard “Book a Demo” push.
“I need to see what they’re selling. You eat with your eyes first.”
— Event leader, expert interview
03 · Users
Users
We designed for one hero buyer and gave everyone else a door, not a homepage.
Personas
The brief listed six audiences, and we could reach only a few people in five weeks. So we built four proto-personas from secondary research (EIC, MPI, PCMA, GDS-Movement, isla), used AI to help cluster recurring segments, and checked them against our interview and survey. I labeled them hypotheses, not findings, and then made the call that mattered more than the personas themselves: optimize 60–70% of the site for one hero.
Where user and business goals meet
Understand it fast
A clear value proposition
Outcome-first hero with the product beside it
Trust it before investing time
Credibility as an emerging brand
Proof placed where doubt shows up, not on one page
Justify it to her boss and sponsors
Support a multi-stakeholder sale
Pricing tiers by team size; plan details she can forward
Take a low-risk next step
Qualified demo requests
One primary CTA plus a lighter “talk to us” path
That overlap became the page logic for every key page: understand → trust → justify → act.
04 · Approach
UX approach
I turned research into decisions the client could sign off on, then built the system that kept three designers consistent.
Before wireframing, I turned the research into a one-page direction brief with five calls the client had to make, each tied to what we heard: tone, density, outcome-first messaging, showing the product, and the contact path. Framing them as decisions, not deliverables, got us alignment in one meeting. One call was really an operations question: if we promise a personal email reply instead of a cold call, can his team keep that promise? We designed the contact flow around his answer.
We ran a moodboard workshop that sorted references by approach (photography-led, illustration-led, text-forward, bento hybrid). The client ruled out stock photography and bento layouts. Illustration didn’t scale: every new page would need an illustrator. Photography and bento layouts depended on images we didn’t have, and stock felt cliché. We asked of every image, does it bring value to the user? That became our visual rule: calm, text-forward layouts with icons, and a denser, data-forward register reserved for proof. I scripted the homepage as three acts: earn the premise, earn the trust, earn the action.
With three designers each owning different pages, consistency was the biggest craft risk. I built and owned the design system (type scale, color tokens, buttons, cards, spacing) and reviewed every page against it before client reviews.
How AI changed the work
AI let three people cover research, content and product visuals that would normally need a larger team. It also produced our clearest mistake. We asked Benjamin up front whether AI use was acceptable, which it was.
Synthesis
Mapped survey data to design implications
Kept what the data supported; sent feature ideas as product recommendations
Personas
Clustered audience segments
Labeled them proto-personas and validated against interview and survey
Content
Drafted placeholder copy and sample proof
Every placeholder flagged for replacement with permission-cleared proof
Product visuals
Generated illustrative dashboard screens
Solved “show the product” before the product existed
Test notes
Summarized usability sessions
Checked against my notes; AI had softened a participant’s criticism
05 · Iterations
Iterations and trade-offs
Testing didn’t ask for more content; it asked for the same content in a better order.
We tested the desktop prototype with three people in 30-minute moderated think-aloud sessions. Only one was a true target buyer. So I weighted findings by who said them: the buyer’s reactions decided questions of audience fit and language, and the designers’ notes informed hierarchy and craft. Trust ratings came back at 4, 5 and 5 out of 5, and the most qualified participant was the most skeptical.
01
Show the product before the product exists
Research said buyers trust what they can see; the client had no stable screens and initially wanted none shown. We chose illustrative product visuals, kept them to what the platform does today, and treated them as placeholders for real screenshots. In testing, the dashboards were the single strongest trust signal.
02
Sequence the proof instead of adding more
In V1, our strongest trust content sat on the Book a Demo page, after the decision was already made. Testing moved it to the homepage. Pricing had the opposite problem, so we moved plans to the top. We kept real regional proof, removed region-locked copy, and recommended a proof plan.
03
One word for one action
The prototype used multiple labels for overlapping things. We settled on one primary action (Book a Demo) and one lower-commitment path (Contact us), renamed “Solutions” to “Who We Serve,” and made the CTA pair a single system component.
04
Remove, don’t add
Two participants flagged the plan comparison as redundant. We cut it and kept tiers named by team size. The workflow explainer was the one place all three asked for more, so we rebuilt it as progressive disclosure that can hold screenshots and video later.
06 · Outcomes
Solution and expected outcomes
A seven-page MVP that answers buyers’ questions in the order they ask them, plus a roadmap for the proof it still needs.
We handed off seven high-fidelity pages (Home, Features, Pricing, About, Book a Demo, Contact, and FAQs) as a clickable desktop prototype with mobile layouts, a UI kit, a sitemap with MVP and future pages marked, and strategic notes in Figma. The homepage moves from problem to product to proof to action, with the product visible in the first screen and trust signals placed where doubt appears. The notes went beyond the brief: a proof plan, audience pathways post-MVP, and a product recommendation to explore budget-vs-carbon trade-off insights.
PETIT DUC homepage
The site would launch shortly after the sprint ended, so I framed each major decision as a hypothesis with a metric, and with the signal that would tell Petit Duc we were wrong.
Showing the product early speeds understanding
Scroll depth past the hero; Features page visits
Visitors bounce from the hero at the old rate
Proof placed before the ask builds confidence
Demo requests per homepage visit
Demo rate holds flat while engagement with proof rises
A lighter “Contact us” path captures not-yet-ready buyers
Contact-to-demo conversion within 30 days
Contacts rarely become demos
Team-size tiers help buyers self-qualify
Plan mentioned in demo requests; lead quality
Sales still spends first calls on plan fit
In testing, all three participants could explain what Petit Duc does in a sentence or two and found their way to Book a Demo; the booking flow itself drew no friction from those who completed it. That validates direction, not outcomes. Petit Duc shipped the live site within three weeks.
07 · Reflection
Lessons learned
Confidence comes from sequence, not volume, and from testing with the people who’ll actually buy.
01
Trust is a sequence problem. I came in assuming more proof would build more confidence. Testing showed the same proof, moved earlier or later, changed how credible the company felt.
02
Recruit for the decision, not the calendar. Only one of three testers was a real buyer, and he found the issues that mattered most. Next time I’d push for 5–6 paid target buyers and a quantitative baseline.
03
Synthetic content has a credibility cost. AI placeholders let us test full flows weeks before real content existed, but on a site whose whole job is trust, a fake placeholder isn’t as reliable in testing.
04
Craft I’m still building. Problem framing and client alignment came naturally; prototyping rigor did not. Broken buttons in one test session became a durable component-wiring habit.
Key Takeaway
User confidence comes from sequence, not volume: the right information at the right moment, matched to how people actually evaluate something unfamiliar.
© 2026 · Michael Cowen · Product Designer