Agent Case Study · Ecommerce
Support that resolves,
while the team sleeps
A home-goods ecommerce retailer was drowning in where-is-my-order tickets and losing overnight shoppers to unanswered questions. We deployed an AI support chatbot trained on their catalog, policies, and order data — it resolves the routine, hands off the rest with context, and turns midnight browsers into morning leads.
Industry
Ecommerce (home goods)
Company size
~40 employees, 3 support agents
Region
United States
Time to live
3 weeks
Client details anonymized under NDA — industry, size band, and timeline are accurate; identifying details are not shared.
Where they started
The challenge
Three support agents handled everything: order status, returns, product-fit questions, and the occasional genuinely hard problem buried under fifty easy ones. First-response times stretched past twelve hours in busy weeks, and the after-hours inbox was a graveyard — shoppers with a question at 11pm simply bought elsewhere.
A previous chatbot attempt had backfired: a rules-based widget that deflected with FAQ links and made customers repeat everything to a human afterward. The bar this time was resolution, not deflection.
The build
What we built
01Grounded in their truth
The chatbot answers only from their help docs, product catalog, and live order API — real tracking status, real return windows, no invented policies.
02Resolution over deflection
It processes routine actions end to end: return initiations, address changes before shipment, and order lookups — not links to forms.
03Handoff with the whole story
Conversations that need a human arrive in the support queue with a summary and full transcript, so customers never repeat themselves.
04Overnight lead capture
Pre-purchase questions after hours end with an answer plus an optional follow-up — name, email, and the product they asked about, waiting for the team at 9am.
Ninety days later
The results
The support team now works the queue that deserves humans — damaged items, edge-case returns, the customer who needs a phone call — while the chatbot absorbs the repetitive majority. The overnight lead capture surprised everyone: questions that used to evaporate at midnight became a steady trickle of warm morning follow-ups.
Want this outcome in your business?
Start with a free AI audit: we map your workflows and show you where this agent pays for itself — before you commit to anything.