How to design a seven-day experiment that proves a GPT assistant reduces reopen rates without hurting csat

When I run pilots for new automation—especially AI that interacts directly with customers—I want one thing above all: clear, defensible evidence that the tech improves outcomes people care about. Recently I built a seven-day experiment to prove a GPT-powered assistant could reduce reopen rates...

How to design a seven-day experiment that proves a GPT assistant reduces reopen rates without hurting csat
Oct 15, 2026 • by Claire Moreau

Featured

Latest News from Customer Carenumber Co

practical checklist to reduce average handle time without hurting first contact resolution

practical checklist to reduce average handle time without hurting first contact resolution

I often get asked whether it's possible to cut average handle time (AHT) without eroding first contact resolution (FCR). In practice, these two metrics are deeply connected: shave off a few minutes in one place and you can accidentally send customers back into the queue. Over the last decade I’ve worked with support teams who needed to reduce AHT to meet cost targets while protecting — and...

Read more...

how to cut support costs by 30% using a tiered self-service funnel and simple analytics

how to cut support costs by 30% using a tiered self-service funnel and simple analytics

I want to show you a practical way to cut support costs by about 30% in under a year by combining a simple tiered self-service funnel with lightweight analytics. I’ve done this for teams that range from 10 to 200 agents, and the pattern is the same: fewer repetitive contacts, faster resolution for customers who prefer self-help, and better triage for agents when human intervention is...

Read more...

step-by-step playbook to migrate from email-first support to true omnichannel operations

step-by-step playbook to migrate from email-first support to true omnichannel operations

Moving a support organisation from an email-first mindset to true omnichannel operations is one of the highest-leverage changes a team can make. I’ve led and advised several migrations like this, and what consistently separates successful programs from stalled projects is a clear, pragmatic sequence of decisions — not aspirational roadmaps. Below is a step-by-step playbook I use with teams:...

Read more...

designing an SLA framework that aligns engineering, product and support teams for faster fixes

designing an SLA framework that aligns engineering, product and support teams for faster fixes

I’ve spent more than a decade helping support teams bridge the messy gaps between product, engineering, and customer-facing teams. One of the clearest levers I’ve seen for getting faster fixes and fewer escalations is a well-designed Service Level Agreement (SLA) framework that’s owned and respected across disciplines. Done poorly, SLAs become finger-pointing tools or a backlog of...

Read more...

how to evaluate knowledge base quality in 20 minutes and prioritize fixes that boost deflection

how to evaluate knowledge base quality in 20 minutes and prioritize fixes that boost deflection

I often get asked by support leaders: “How do I know if our knowledge base is actually helping customers — and how quickly can I spot the biggest wins?” If you’ve got 20 minutes and a cup of coffee, you can run a pragmatic, evidence-based audit that surfaces the highest-impact fixes for deflection. Below I walk you through a repeatable 20-minute process I use with teams to evaluate...

Read more...

5 automation rules that reduce repetitive work for agents and increase velocity (with examples)

5 automation rules that reduce repetitive work for agents and increase velocity (with examples)

When support teams ask me where to start with automation, the conversation quickly moves from flashy AI demos to the boring-but-critical question: which rules actually reduce repetitive work for agents and increase velocity? Over the years I’ve seen teams pour time into automations that sound clever on paper but don’t move the needle on response times, agent load, or customer experience. Here...

Read more...