Role
End-to-end product design, prototype testing, Branding, Art Direction: Elliot Farmer
Industry
Real estate
01 / OVERVIEW
B&C's advantage was never their listings. It was that they act as a concierge – and none of that reached the website.
B&C operates in Monaco's luxury market, where the differentiator isn't inventory — the same properties circulate between agencies — but service. They handle relocation logistics, know the local property law, maintain long relationships with clients, and actively hunt for properties matching a client's criteria even when nothing suitable is listed.
Their website did none of that. It listed properties. Every part of the business that made B&C worth choosing happened in person, over WhatsApp, or on the phone.
02 / HOW I WORKED
The workshop established what the business actually sells before we discussed what the website should do.
I ran a discovery workshop with the client team, mapping their strengths and their goals separately.
Redesign goals were identified
01 — A more luxurious and mobile-friendly site
02 — Personalised property search
03 — Engaging content communicating their expertise
04 — Stronger brand image
05 — Better lead capture
06 — Conversion optimisation
Service blueprint
Half of this service happens where a website can't see it.
B&C isn't a digital product with an offline component, or an offline business with a website attached. A Monaco property search genuinely runs across both for months — browsing at midnight, viewings in person, decisions made over dinner with a spouse, paperwork with a notary — and the handoffs between the two are where things get lost. So before deciding what the product should do, I mapped the whole service: what the client does, what the agent does visibly, what the agent does invisibly, and which systems support any of it.
I was looking for 3 things:
01 — Gaps
02 — Places where digital could make the offline experience easier
03 – Moments where offline knowledge was leaving no trace. The most business value of the three. A client's reaction after a viewing – price too high, wrong district – lives in an agent's head and nowhere else. Capturing it in the product turns messy agent knowledge into data the agency can analyse and act on.
Everything that was scattered across 5 channels, in one account
Before this, a property search in Monaco was spread across whatever channel each step happened to use. Listings arrived as WhatsApp links that scrolled out of reach within a week. Tours were booked by phone, so they existed only in the agent's calendar and the client's memory. Offers went by email. And because these decisions are never made alone, the client forwarded everything to a partner or an advisor, where the conversation continued somewhere neither they nor the agent could see it.
The account collects all of it.
Share is built into every section for the same reason. If the decision involves a spouse or a lawyer, that conversation should stay inside the search rather than fragmenting into screenshots — so nothing gets lost, and the agent can still see where things stand.
Lead generation planning
Preferences that pay off immediately
Preferences arrive pre-filled from the search the user has already done, so the form is a confirmation rather than a questionnaire. Saving them triggers a visible generation state — the recommendations are being built for this person, not pulled from a shared list. The signal matters as much as the mechanism: personalisation the user can't see doesn't feel like personalisation.
It also changes the agent's side of the relationship. Preferences are visible to the agent, so the first properties a client receives are relevant — instead of the standard opening round of near-misses that costs an agency credibility before the relationship starts.
Popularity as quiet urgency
"Saved 16 times this week" gives a property page the one thing a static listing lacks: evidence that other people are interested. In a market where the same properties circulate between agencies, that's a real signal rather than a manufactured countdown — and it's stated as a fact rather than a warning, which keeps it on the right side of the brand.
Guides that prove expertise and capture an email
The buying, renting and valuation guides demonstrate the local knowledge B&C sells — Monaco property law, the actual steps of a purchase — which is the most credible form of marketing for an agency whose differentiator is expertise. And the agent sits alongside the content the whole way down, named, with a direct WhatsApp line. Downloading the full guide requires an email, which makes it the softest possible lead capture: the visitor gets something genuinely useful, and B&C gets a contact who has self-identified as a serious buyer.
Mobile wasn't an adaptation at the end. It was a requirement from the first workshop.
The client raised it in discovery: a significant share of their traffic comes from phones and tablets, and their existing site handled it poorly — which matters more in this market than most, since these clients browse between meetings and while travelling, not at a desk.
03 / VALIDATION
Testing
I validated the prototype at several stages with three B&C partners and agents. They weren't end buyers, but they were the people running these transactions daily — which made them the right check on whether the model matched how a Monaco deal actually works: the sequence, the terminology, the edge cases like multi-property tours and off-market listings.
What that testing couldn't tell me was how a first-time visitor experiences the product, so decisions about the registration moment stayed reasoned rather than validated.
04 / DESIGN SYSTEM
Design system with Claude Code
A self-directed experiment after the project wrapped: using Claude Code and the Figma MCP server to audit B&C's existing components and turn them into a documented, tokenized system — shipped to both Figma and Storybook.
05 / COMPROMISE
What didn't ship
2 of the more ambitious ideas didn't survive contact with the budget.
AI-driven recommendations and search
Two of The original design leaned far more heavily on AI — generating recommendations, shaping the search, interpreting intent. The client was enthusiastic and signed off at design stage. When the developer costed implementation and ongoing usage, the economics didn't work for an agency of this size.
The lesson was about sequencing, not about AI. Design approval isn't feasibility approval, and I'd now bring engineering into the costing conversation before the client falls in love with a direction — because withdrawing an approved idea costs more trust than never proposing it.
Free-text search
The plan was to let people type what they wanted — "a property in Larvotto to rent" — and have the system interpret it. It was simplified to structured filters, partly for implementation cost and partly because free text was unpredictable: when interpretation fails on a €5M search, the user has no idea what went wrong or how to correct it.
The controlled version turned out to have a genuine advantage: filters show their state, so a user can always see why they're getting these results and adjust. If I designed it today I'd propose the hybrid — free text as an accelerator that populates visible filters the user can correct — so interpretation errors stay recoverable.