How do you find out whether your queue has adjacent demand?

Read what the agent could not resolve as a demand map. IKEA found that the unresolved contacts were people asking whether a sofa would suit their living room, which is a sales conversation. If your unresolved contacts are integration failures and billing disputes, there is no purchase in them.

What is a value center?

Canon's term in the same Salesforce piece. Bill Duval describes agents taking the questions that do not need a person, customer effort going down, and the voice of the customer feeding product development. It is a real return and it is not revenue.

What goes in the business case if there is no revenue line?

The cost number, one tightly defined customer outcome, and the name of the person who reports it. At Crisis Text Line every proxy I tried, such as session length and messages sent, failed the question of whether the person ended up safer, so the outcome has to be the thing itself.

Is the Canada Goose result typical?

I have no data on how typical it is, and the article does not claim it is. It is the lead example in a vendor's customer roundup. It fits a pattern, luxury retail where buyers want advice, that most B2B support queues do not share.

Can AI Turn a Support Cost Center Into a Revenue Engine

Can AI turn a support cost center into a revenue engine?

THE SHORT ANSWER

Only when there is unserved demand sitting next to the work you automated. Canada Goose has it: per Salesforce, Agentforce resolves 89% of routine messaging inquiries and 15% of calls, and the former call center now sells through personal shopping. IKEA had it: Billie resolved about 47% of 3.2 million interactions, 8,500 staff became remote design advisers, and the channel produced €1.3 billion in FY22. A queue of password resets and failed integrations has no purchase attached. For those, the ROI is the cost number plus a customer outcome someone is named to report. I ran product at Crisis Text Line with no revenue at all, and the question that replaced it was whether the person ended up safer.

Salesforce's customer roundup of October 1, 2026 carries the headline that AI turns cost centers into revenue engines, and Not Every Cost Center Has a Revenue Engine Inside is my case for retiring that as the default. The test it fails for most teams comes from The IKEA Story Does Not Prove What You Think It Proves. The scoreboard to use when there is no revenue to point at comes from Product Management Without Revenue Made Me Better.

It belongs to the AI Business Models argument: AI removes a labor cost, and what the freed labor becomes has to be stated as honestly as the saving.

Last reviewed 2026-10-01 · 1 min read

Can AI turn a support cost center into a revenue engine?

SOURCES

THE LONG VERSION

RELATED ANSWERS