Led the UX/UI team through 14 months of research, design, and delivery of a B2B e-commerce platform for a major agricultural company - from initial workshops and user interviews through two rounds of usability testing to MVP launch.
A major company in the agricultural industry needed a completely new B2B e-commerce platform to replace the previous store and serve both existing and new customers. Until then, all purchases required direct contact with sales representatives - no independent online ordering existed. The goal was to build a store that would let customers browse, compare, and buy products on their own, while integrating with the existing eCare portal (order history, invoices, loans).
I joined as Design Lead, managing a UX designer and a UI designer. I had product decision-making authority and influenced the roadmap, though the client had a strong vision - which meant frequent workshops and the need to prepare solid arguments to steer direction. I delegated tasks based on our backlog and sprint capacity, ran daily standups, resolved cross-team conflicts (mostly between design and dev), and ensured issues were addressed immediately before they could escalate. Over 14 months we held 897 meetings - standups, weekly workshops, design reviews, and retrospectives.
We started with workshops to define business goals with the client. Together we developed milestones grouped into three areas: business profitability (increasing sales, reducing sales costs), market position (service quality, setting industry standards), and uniqueness (expertise in the agricultural sector). We identified three target user groups: family farms (RGR), new online store users, and B2B clients.
The user path was divided into four stages: product discovery (home page, listing, details), checkout (3-4 steps), payment, and order confirmation. For MVP we scoped it to the first stage - three core pages. I prepared detailed time estimates for each design phase: research, component design, variants, corrections, internal reviews, RWD versions, client approval, and final UI. This estimation framework proved essential for planning deliveries to the development team.
We ran usability tests on the eCommerce 0.5 prototype with 16 participants (5 internal pre-tests, then 11 real users: 6 family farms, 3 commercial farms, 2 B2B clients). The concept was received very positively - all participants declared willingness to use the platform. The information structure was rated very well and the prototype met their needs for browsing and finding products.
Key findings: product availability was critical information users needed upfront (to avoid calling sales reps). Seasonal product displays were highly valued - farmers specialize in specific crops, so filtering by season shortened their path significantly. Personalization emerged as a major opportunity. Product tiles underwent the biggest redesign - we made them smaller, reduced photo sizes, and highlighted key product specifications instead. We also learned that farmers are extremely precise with math and unit conversions, so clear product sizing became essential for practical purchasing decisions.
After several months I noticed coordination issues between teams. I developed a recovery plan with business analysts - a structured schedule giving the client specific time for business requirements, the design team time for design work, and dedicated slots for developer and client sign-off. This became the standard and the collaboration ran smoothly from that point.
Research showed farmers search by what they grow, not by product category. We introduced crop-based navigation (corn, rapeseed, beet, orchards), seasonal sections promoting products relevant to the current growing period, and personalized cross-sell recommendations suggesting complementary products for a farmer's specific crops - based on their profile and purchase history.
Rather than designing the full store at once, we focused MVP on three core pages (home, listing, product detail). This let us validate the concept quickly through research, iterate based on real feedback, and give developers working code while we designed the next phase.
I created a granular estimation framework breaking each component into research, design, variants, corrections, reviews, RWD, client approval, and final UI. Different team members participated at different stages, so each person's workload was calculated separately. This made sprint planning predictable and deliveries reliable.
After redesigning based on round 1 findings and extending the store to include registration, checkout, and order confirmation, we ran a second round with 7 participants. The overall purchasing journey was rated as easy and intuitive. Users confirmed that the information was clear and sufficient for making informed purchasing decisions.
The main issue was the Discount Program - users appreciated it but found the rules unclear (why do some products earn points while others get price reductions?). Registration was seen as too long and marketing consents felt excessive. Based on these findings, we simplified the discount program logic and its in-store communication, streamlined the registration flow by removing non-essential steps, reduced the number of consent checkboxes, and adjusted product tile heights after client feedback to fit more products on the listing page.
Understanding how farmers actually buy was the steepest learning curve. Their purchasing logic is nothing like typical e-commerce - they buy seasonally, by crop type, in very specific unit sizes, and they're incredibly precise with math and conversions. It took many conversations and workshops to internalize this well enough to design for it. Every assumption we brought from standard e-commerce had to be questioned.
The client had a strong product vision, which made collaboration focused and rigorous. Every design recommendation needed to be grounded in research - which ultimately made the outcomes stronger. It wasn't adversarial, but it demanded discipline: every recommendation had to be grounded in data. Cross-team conflicts between design and development also needed ongoing attention - the key was resolving them immediately, before they could compound.