Support costs grew faster than
the business did.

CartLoop sells homeware across five markets.

When order volume quadrupled in a year, the support queue followed it almost exactly. Nine agents who had comfortably handled the previous year’s volume were now working through a backlog that reset every morning before anyone finished reading it.

The obvious fix was an expensive one.

Hiring proportionally meant fourteen more agents, most of them answering questions the company had already answered thousands of times. It would also have consumed the margin the growth was supposed to create, which made the decision harder than it looked on paper.

Where the volume was coming from

Before changing anything, the team audited a full quarter of tickets. The distribution was lopsided.

• 64% were order tracking, returns, or delivery questions.

• 21% needed a judgment call on refunds or damaged goods.

• 15% were genuinely complex or specific to one account.

Almost two thirds of the queue could be answered from systems the company already ran. The people best equipped for the last group rarely reached it before the end of the day.

What they automated first

CartLoop connected ResolveAI to their order platform and stopped there for the first month.

Only the top three intents were automated. There was no broad rollout and no attempt to deflect everything, because deflection was never the goal. The goal was removing questions where a correct answer already existed somewhere the AI could read it.

Within two weeks it was handling:

• Order status and tracking

• Return eligibility and labels

• Delivery date changes

• Address and account updates

• Routing for everything else

Nothing about the team’s workflow had to change.

Results after 90 days

The second phase mattered more than the first. Explicit rules sent refunds above a set value, damaged goods claims, and any third contact within a week straight to a person.

Support cost per order - Reduced by 60%.

First response time - Improved from 4 hours to under 2 minutes on automated conversations.

Customer satisfaction - Increased from 4.2 to 4.5 out of 5.

Automated resolution - 61% of total volume closed without an agent, with no measurable rise in repeat contacts.

Quality moved in the same direction as cost, which was the part the team had been least confident about.

Why it worked

CartLoop never tried to automate everything, and they never reduced headcount. Three agents moved onto proactive outreach and returns recovery, work that had needed an owner for a year and never had one. The 60% saving is measured against the team they would have had to hire, not the team they already had.

What they would do differently

The team’s own verdict is that they waited too long to write the handoff rules. For the first few weeks the AI was making reasonable calls on refunds that a person should have owned, and while nothing went badly wrong, the rules removed a category of risk that had been quietly sitting there the whole time. Writing them on day one would have cost an afternoon.

Ready to grow without growing the queue?

If your support volume scales directly with your order volume, the pattern CartLoop found is probably sitting in your own tickets too. ResolveAI connects to the systems you already run, automates the questions that repeat, and routes everything else to the people best placed to handle it.

Book a personalized demo to see how ResolveAI fits into your existing workflow.