Redesigning B2B Platform Connecting Brands with Retailers

Two years, four user types

The platform served four meaningfully different users — retail buyers, brand vendors, investors, and large retail chains — each with distinct jobs to do inside the same product. Designing for all of them without making the product feel fragmented was the central challenge across two years.

The product’s direction shifted multiple times as the team learned what the market actually needed. The design had to move with those shifts rather than lock the product in place. Research was embedded in everything — sitting in on sales demos, watching users navigate live, and listening to what confused them and what clicked.

Two complete redesigns — and why

The full platform was redesigned twice.

The first version had a more modern, youthful aesthetic — approachable and contemporary. As the team refined its target market toward larger retail chains and institutional buyers, the feedback was consistent: the product needed to feel more trustworthy, more data-forward, more enterprise.

The UI went through a complete overhaul — shifting from the warmer visual language to a structured, data-driven interface:

• Metrics and data points moved to the top of the visual hierarchy
• Colour palette shifted to a neutral, professional register
• Typography optimised for legibility at data-dense scales
• Component patterns aligned with enterprise SaaS conventions buyers in this space already recognised

The design system problem

When the engagement began, two separate component libraries had accumulated from different contributors. Components looked different between screens. Engineering was implementing the same element multiple ways. Every new feature added inconsistency rather than resolving it.

Decision: rebuild the design system from scratch. Both libraries were replaced with a unified system — one implementation reference for engineering, one visual language across the product. Features designed in the second year were faster to produce and more consistent than anything from the first.

The four retailer flows

After multiple iterations and direct user research, four core workflows were defined for how retail buyers use the platform. Each flow has a distinct purpose and moves the buyer through a different stage of vendor discovery.

Flow 1 — Scouting and brand discovery

The entry point to the retailer experience. Buyers find brands using different approaches depending on how they’re thinking:

• Category and trend browsing — for buyers exploring a space without a specific brand in mind
• Location-based filtering — for buyers sourcing locally or with specific geographic requirements
• AI natural language search — buyers type a full sentence and get relevant results immediately, without manually configuring filters

The design challenge: search had to be genuinely flexible without being complicated. A buyer typing one word should get useful results. A buyer specifying ten criteria should also get useful results. Both needed to feel natural. When a brand looks promising, buyers save it to their private brand list — the starting point for the next flow.

Flow 2 — Vetting

The active evaluation phase. Buyers work through their private list deciding which brands are worth pursuing. Two distinct views served two fundamentally different types of evaluation:

Visual view — brand aesthetic, social presence, product imagery, visual identity. Designed for the gut-check phase: does this brand feel right for our stores? This can’t be answered from a spreadsheet.

Table/data view — performance metrics, sales velocity, distribution history, pricing tiers. Designed for due diligence: once a brand passes the visual check, do the numbers support stocking them?

Actions available during vetting:

• Favourite — mark brands for closer attention
• Archive — remove brands that don’t fit without losing the record
• Internal notes — private notes visible only to the buyer’s team
• Connect — send a direct query or invite the brand to apply to an open vendor request

Flow 3 — Monitoring

After vetting, buyers have a shortlist — typically around 20 brands they’re interested in but not ready to commit to. Monitoring is the ongoing tracking of that shortlist over time.

Buyers follow brands to see how they’re developing: new product launches, distribution growth, social momentum, pricing changes. The key design requirement: the monitoring view had to surface change signals — what’s new or different since the last check — without requiring buyers to manually review each brand individually.

Flow 4 — Presenting

Allows buyers and consultants to generate structured vendor intelligence reports from their Cohere data and share them externally. Reports can be generated from a private saved brand list, a market category, or a curated brand selection.

Reports are generated by Cohere’s AI from the platform’s private vendor database — published as branded shareable links, accessible without a Cohere login. For consultants presenting vendor landscape analysis to retail chain clients, this replaces the manual process of exporting data and building presentations from scratch.

Brand user experience

Brands on Cohere have five distinct areas of the product, built around visibility — understanding how they’re perceived, where they stand competitively, and how their applications are progressing.

Brand dashboard — the Cohere Score

The dashboard is built around a proprietary metric: the Cohere Score — a composite retail-readiness rating calculated from eight distinct dimensions, expressed on a scale of 0–100.

The Cohere Score is designed to function as an industry-standard benchmark — a way for retailers to quickly gauge whether a brand is ready for serious stockist conversations, and for brands to understand where they stand and what to improve. Designing it required making an eight-dimension calculation feel legible rather than opaque.

The dashboard also shows:

• Category ranking and leaderboard — where the brand sits within their product category
• Next rivalries — brands immediately above and below them
• Open retailer requests — how many vendor requests in their category are currently active
• Application and invite stats — requests applied to, direct invitations received, status of both

AI-powered review analysis aggregates publicly available reviews through AI to produce dimension-level scores across quality, pricing, in-store experience, packaging, and innovation — each scored individually, with explicit pros and cons and the underlying reviews available in full.

Retailer finder — pre-filtered opportunity discovery

The retailer finder shows brands a pre-filtered list of vendor requests matching their category, location, and profile. Split-view layout: left panel shows ranked applicable opportunities; right panel shows full request detail and the application form. Brands apply without leaving the page, with form fields prefilling from their saved profile data.

Vendor portal — application tracking

A table of all active applications showing current stage, when status last changed, and retailer actions. When a retailer archives or moves a brand forward, the brand receives a notification immediately — removing the black box that most application processes become after submission.

Brand detail page — managed and curated content

The brand detail page is the brand’s public-facing presence on Cohere. Brand-managed: company information, brand story, product catalog, geographic reach, distribution history — all editable by the brand. Cohere-managed: review-derived insights, Cohere Score breakdown, competitive intelligence signals — brands can see this but can’t edit it.

Guest users vs. registered brand profiles

Not every brand applying has an account. Guest users have to enter everything manually. Registered users have saved business profiles — application forms prefill automatically from profile data. Strategic conversion prompts were built and timed to the highest-friction moments: the conversion proposition is most tangible exactly when the friction is highest.

After two years

The product went from feeling unfinished in sales demos to holding its own against well-funded competitors. The design system gave engineering the room to ship faster and more consistently. The shift to a data-forward visual language positioned the product correctly for the enterprise buyers it was targeting.

Most importantly: the product started to reflect the strength of the data behind it. Cohere’s vendor intelligence was genuinely strong — the design’s job was to make that evident to anyone who landed on the platform.