I’ve worked in support, built chatbots and voicebots, and now design contact-center AI products.
We’ve gone from IVR and self-service to agent assist and LLMs. The technology changes, but we keep asking similar questions.
How many calls can we automate? How much handling time can we save? How much can we reduce operating costs?
Those goals matter. I’ve worked toward them and seen results. But I started wondering what we were actually trying to solve.
This series focuses on inbound support, when customers contact us after running into trouble with a product or service.
1. Why do customers keep calling about the same problems?
“How many calls can we automate?” begins with an observable symptom: high call volume.
It asks how to handle the calls, not necessarily why they’re happening.
Product teams keep improving services, and contact centers keep investing in automation. Yet some issues still drive calls year after year.
A stable call count doesn’t mean improvements have failed. Customer growth and new services bring new demand. But are the original causes of recurring contacts actually disappearing?
If not, what have we actually solved?
2. Is handling an inquiry the same as solving the problem?
Working on these systems taught me they aren’t.
Even when customers ask the same question, what they need can depend on the situation that brought them there.
During tax-filing season at a card company, customers kept asking for transaction records. Four self-service channels already offered transaction history, yet agents still issued 82.5% of the documents.
The issue wasn’t the absence of a transaction history feature. Customers needed records for a specific tax period, in Excel, with the details required for filing.
We were discussing “transaction history inquiries.” Customers were trying to complete their tax filings.
We had the right category and a working feature. What we lacked were the options customers actually needed.
Support can track rising inquiries. Product teams need to know where customers got stuck and what should change.
Contact centers are designed to handle inquiries. Tracing causes across products and policies, and assigning work to other teams, isn’t always within their remit.
3. If you started a business, what would you expect from customer support?
Would you want accurate, cost-efficient responses?
Or would you also expect the team to uncover why customers needed help and turn those conversations into product improvements?
Both matter. The question is whether support is expected only to manage the cost of contacts or also to help the business fix what caused them.
In this series, I want to explore why that second role is difficult to deliver, and what AI might change.
#CustomerExperience #ContactCenter #AIProductManagement