Case study — the first implementation
What building the concierge inside a real furniture store taught us
ShowroomOps wasn't designed on a whiteboard. It was built inside our own high-end furniture store — one large destination showroom with a deep new-and-consignment catalog — on a live catalog, real policies, real staff, and real customers. This page is what happened, with the caveats attached. We keep the store unnamed publicly; on a teardown call you get the store, the dashboards, and the methodology.
The starting problem
A busy showroom, a website that couldn't sell
The store had what most independent furniture retailers have: a deep, fast-moving catalog spanning new and consignment inventory, a sales floor that closed the overwhelming majority of revenue, and a website that generated questions it couldn't answer. Shoppers searched for pieces the search bar couldn't parse, asked delivery questions after hours, and left.
Off-the-shelf chat tools were support-first: built to deflect tickets, unable to search the catalog, and blind to the showroom. So we built the system we couldn't buy — product-search-first, floor-aware, and honest about what it didn't know.
What was installed
The build
- An AI concierge trained on the live Shopify catalog — thousands of SKUs including consignment pieces, floor models, and deep variant trees
- Product cards in chat: image, title, price with sale strikethrough, availability, and a link to the product
- Showroom routing — the concierge knows what's on the floor and invites shoppers to come sit on the exact piece
- Policy training: delivery tiers, white-glove vs. front-door, returns, financing, custom-order lead times
- Human handoff routed by type — designer, sales team, support, or the owner
- Auto-answer with guardrails: unknowns get an honest 'I don't know' and a handoff, never a fabricated spec
- Conversation-to-order matching, so reporting runs on matched revenue instead of chat counts
The conversation pattern that emerged:
Store Concierge
Answers from your catalog, policies, and floor inventory
Hartley Performance Sectional
112" · L-shape
$3,295 $4,695
In stock — on the showroom floor
See productMarlow Chaise Sectional
108" · reversible chaise
$2,850
In stock — ships in 5–7 days
See productEverett Track-Arm Sofa + Ottoman
104" combined
$2,390 $3,190
Custom fabrics — 4–6 weeks
See productWhat changed after launch
The numbers, with the caveat attached
- outsold in one, at pace
- 5 yrs
- 2026 store revenue is pacing past the previous five years of sales combined.
- June revenue, year over year
- 4.05×
- $628K in June 2026, with July projected near $765K.
- chat-assisted revenue in 14 days
- $27.5K
- Orders matched to concierge conversations after auto-answer went live — chat volume roughly doubled, and matched revenue ran at a ~$60K/month pace.
Read this before quoting any of it
We report the store's numbers as the store's numbers and assisted revenue as matched orders — never claimed lift from any single tool. The growth came from the whole operating stack working together: buying, marketing, catalog, concierge, and measurement. We'd rather under-claim and show you the methodology. What we will say: the concierge made the website a genuine sales surface for the first time, and the matched numbers were strong enough that we productized the system.
Lessons
Seven lessons that became the product
Product search first, support deflection second
The tools we tried before building were support-first, and furniture shoppers ignored them. The moment chat could actually search the catalog and show products, it became a sales surface. This is now the product's first design rule.
Product cards are the money surface
Prose answers about products underperformed cards with image, price, and availability every time. Cards first, then a short recommendation recap — the order matters, and it's now fixed in the system.
The showroom dual-path outperforms pure ecommerce
Roughly 90% of the store's sales close on the floor. Conversations that ended with 'come sit on it — it's on display' became visits and floor sales that no web-conversion metric would ever have credited. Showroom CTAs are now first-class conversions.
Anxiety moments decide high-ticket sales
Delivery, returns, financing, white-glove details, lead times, custom-order rules — precise answers at these moments moved conversations forward. Vague answers ended them. Policy precision is implementation work, not an AI afterthought.
Never gate the first question
Requiring an email before answering killed conversations. Frictionless first answers, contact capture later when it serves the shopper — that ordering roughly aligns with when chat volume doubled.
Route handoffs to the right person
A generic 'an agent will follow up' handoff loses the thread. Routing design questions to designers and big-ticket buyers to sales kept conversations alive — and staff actually engaged because the handoffs were relevant.
Measure matched orders, not widget events
Chat-widget events inflated GA4 engagement and proved nothing. Joining conversations to backend orders and POS produced numbers worth making decisions on. Dashboards now lead with assisted revenue, card clicks, design leads, and showroom CTAs.
Every ShowroomOps engagement inherits these lessons on day one — that's the point of buying an implementation from operators instead of a widget from an app store. See how the full system works or what an engagement costs.
Your store's version of this case study starts with a teardown
We'll apply everything above to your site — search, PDPs, policies, showroom path — and show you what the first 90 days would target.
The teardown is free and delivered live on a 30-minute call by the operator who built the system. We onboard a small number of retailers at a time — each system is built from real catalog and policy data.