A support team of three handles around 400 conversations a week at one of our customers. Six months ago their median first response took almost two days. Today it is under two hours, with the same three people. Here is what changed, in the order they changed it.

Routing before anything else

The biggest delay was not writing replies; it was deciding who should write them. Every new conversation now runs through a small set of rules before a human sees it.

• Billing questions go straight to the person who owns invoicing.

• Messages mentioning “bug”, “error” or “broken” open a ticket for engineering and get a holding reply.

• Everything else lands in a shared queue sorted by how long it has waited.

Connecting the tools that already existed

None of this needed new software. VEYA sits between the help desk, Slack and the CRM, so a conversation can carry its customer context with it. When a reply arrives, the agent already sees the plan, the last invoice and any open tickets in one place.

Templates that still sound human

The team wrote twelve short templates for the questions that make up most of the volume. Each template is a starting point, not a script, and the agent edits before sending. Response quality went up because people stopped retyping the same three paragraphs.

We did not hire anyone. We just stopped making people search for things they already had.

What the numbers look like now

• Median first response: 1 h 50 m, down from 46 hours.

• Conversations resolved in one reply: 61%, up from 38%.

• Time spent on manual routing: effectively zero.

If your response times are long, start by measuring where the time actually goes. In almost every team we have worked with, the waiting happens before anyone starts typing.