Content pipeline
Content pipeline running on autopilot
Keyword research and publishing pipeline for an in-house SEO stack. Indexed pages generated, linked, and maintained without a writer on payroll.
Publishing without an in-house writer
Ecommerce ops
One CX person handles 400 tickets/day by clicking 'approve'
Every ticket arrives with full order context pulled, policy checked, and a draft response waiting. The human reviews, tweaks if needed, and sends.
Client
UK DTC skincare brand
Build time
3 weeks to production
Headline outcome
Support ops without a tier-1 team
The problem
A growing DTC brand hits a wall around £2m ARR: enough volume to need a CX team, not enough margin to hire one. Tickets sit in Gorgias or Zendesk for 12+ hours, customers churn, and the founder ends up triaging on weekends. Canned responses help 10%, the other 90% still need real judgement.
The system
A ticket classification + drafting layer on top of the helpdesk. Every ticket is tagged (return, delivery, product question, wholesale, complaint), order context is pulled from Shopify, the returns policy engine makes a refund/replace/escalate decision, and a response is drafted in-brand. One CX person reviews and sends, no typing from scratch.
How it runs
The workflow, end to end.
01
Ticket classification + priority
Every inbound ticket is classified by intent, sentiment, and urgency. VIPs and complaints skip the queue.
02
Order context pull
Shopify order, tracking, return history, and customer LTV are pulled and attached to the ticket before a human looks at it.
03
Policy decisioning
Return eligibility, refund window, and replacement-vs-refund logic runs against the brand's configurable policy. Output: recommended action + reasoning.
04
Draft response
Claude drafts a response in brand voice, using the customer's first name, the order details, and the policy decision. No templated copy.
05
Approve + send
CX clicks approve. Shopify refund/replacement is processed automatically. Everything logs back to the ticket.
Outcomes
What it moves.
12h → 8min
First-response time
80 → 400
Tickets / agent / day
~85%
One-touch resolution rate
+18pts
CSAT
Stack
What it's built on.
- Gorgias / Zendesk API
- Shopify API
- Claude (drafting + classification)
- ShipStation / Royal Mail tracking
- Postgres
EcommerceCXShopify
Doing £1m+ on Shopify and drowning in tickets? Let's scope the first pass.
20 minutes. No slides. One process, one prototype.
Book a 20-min introMore builds
View all