Why Contact Center Investments Fall Short (2/2)
Part 2 looked at what we chose to automate. Here I want to ask who decides what to automate, and why.
1. When we choose what to automate, we inherit the existing categories.
When a contact center plans automation, it usually starts with the inquiries customers make most often. The shortlist often comes straight from its existing call categories.
Labels like “credit limit” and “transaction history” help route and track calls. They weren’t designed to explain why customers got stuck.
Two customers may ask to raise their card limit. One has used it all; another recently lowered it and may need a separate review to raise it again. Same question, different responses.
More detailed categories aren’t the answer. I’ve seen centers with hundreds or thousands of labels, many rarely used. More choices slow agents down, create inconsistent judgments, and add work for managers. This is why automated classification is a common request for agent-assist systems.
But what if we automate “limit increases” based on that category, and the calls keep coming? Different problems may still be hidden under the same label. We may have automated handling the inquiry, not fixed what caused it, then used those recurring calls to justify more automation.
2. Who is at the table when we decide what to build?
Contact centers do investigate recurring inquiries. But the cause may be in a product flow, policy, notification, or process owned by another team.
Customers get stuck using a product or service, then contact support. Yet in many projects I worked on, the contact center defined the requirements for chatbots, voicebots, and AI.
Support can lead these projects. Not every call can be prevented, and bots can help. But results depend on what we use them to solve after launch. If a customer couldn’t finish a task in the original service, a bot may need changes to the product, policy, or process.
The contact center often pays for the tools and the calls, even when the cause sits elsewhere. Should it also be solely responsible for running and improving them?
Who decides what the tool should solve after launch? If AI answers quickly and accurately, but customers keep calling for the same reason, have we solved the underlying problem?
Faster answers and lower costs matter. But to judge the investment, we also need to know whether customers finished their task and whether the same cause keeps bringing them back.
If support can’t fix the cause alone, the teams that can must help define the purpose and stay involved after launch. Success isn’t just deploying AI. It’s what the customer can now do.
And even when we agree on the problem, who is responsible for fixing it?
#ContactCenter #AIProductManagement #CustomerExperience