A client was drowning in 19,400 support tickets a month. A quarter of them were "where is my order" — entirely preventable. We deployed our CX agent, fixed the infrastructure, and now we run it.
A major UK e-commerce retailer processing approximately 19,400 support tickets per month. CSAT was 11-21 percentage points below industry benchmark. The P90 first response time was 13.7 hours — 14 times slower than the median. More than a quarter of all tickets were customers asking where their order was — entirely preventable with basic proactive communications that the infrastructure already supported but nobody had configured.
Four things, run in parallel. Each addresses a different root cause — together they compound into a step-change in satisfaction and cost.
Configured order lifecycle notifications — shipped, delayed, return received, refund initiated, refund processed. The infrastructure already existed. It just needed trigger configuration and template design. Target: eliminate 26% of ticket volume (~5,000 tickets/month).
The existing chatbot was the #1 source of negative reviews — routing 37% of tickets with minimal true deflection (5%). We rebuilt conversation flows for the top 5 ticket categories: order cancellation, returns initiation, tracking queries, refund status, and image-based product issues. Target: 20%+ true deflection rate.
Added 5 FTEs covering weekends and weekday evenings at £12,500 per FTE all-in — half the cost of UK equivalents. This eliminated the Monday morning backlog of ~766 unattended weekend tickets and brought P90 response from 13.7 hours to under 4.
Targeted coaching for underperforming agents (agent CSAT ranged from 40% to 75%), VIP customer routing (44% of support contacts were VIP customers worth £344/year average — 7.3x overrepresented), and standardised escalation protocols.
| Category | Monthly Volume | Deflection Rate | Tickets Saved |
|---|---|---|---|
| WISMO (Where Is My Order) | ~5,122 | 40% | 2,049 |
| Refund Enquiries | ~989 | 50% | 495 |
| Returns | ~1,959 | 30% | 588 |
| Order Updates & Other | ~7,630 | 25% | 2,232 |
| Total | ~19,400 | 27.6% | ~5,364/month |
44% of support contacts were VIP customers worth £344/year average — 7.3x overrepresented relative to the customer base. These customers had a higher opt-out rate (3.3%) than standard customers. Total revenue at risk from poor CX across all segments: £25.8M.
Every unresolved ticket, every 13-hour wait, every broken chatbot interaction was a chance for a high-value customer to leave. The customer data platform identified exactly which customers were most at risk, enabling prioritised routing and proactive retention. This is not a cost-cutting exercise. It is revenue protection.
These results were measured at a major UK e-commerce retailer processing 19,400 tickets per month. The CX work was one capability we deployed as part of a broader engagement that created £6.4M in annualised value. Our CX agent now monitors CSAT in real-time, identifies underperforming agents automatically, and surfaces emerging ticket trends before they spike. We stay and run it.
Most "where is my order" tickets happen because customers have zero visibility after checkout. By sending automated notifications at each fulfilment milestone — order confirmed, dispatched with tracking, delay alert, delivery confirmation — you answer the question before it becomes a ticket. We saw 40% WISMO reduction from proactive notifications alone.
Revenue protection. Our analysis showed support contactors have 38x higher lifetime value than non-contactors. VIP customers are 7.3x overrepresented in support queues. A 5% improvement in VIP retention pays for the entire CX effort. The cost savings are real but secondary.
Three levers: reduce ticket volume (fewer frustrated repeat-contactors), reduce response time (most dissatisfaction correlates with wait time, not resolution quality), and targeted coaching for underperforming agents. We found a 35-percentage-point spread between best and worst agents — coaching closes this gap.
We do. Our CX agent monitors CSAT in real-time, identifies underperforming agents automatically, and surfaces emerging ticket trends before they spike. Our team reviews the data daily and makes adjustments. That is the forward deployed model — we build it and we stay to run it.
12 weeks for the full scope. Quick wins (proactive communications) deploy in weeks 1-4 and deliver immediate volume reduction. Chatbot optimisation runs weeks 4-8. Offshore recruitment and go-live happens weeks 6-10. Optimisation from week 10 onward. Then we stay and keep running it.
Tell us how many tickets you are processing and we will tell you how much is preventable. No commitment beyond the first conversation.