What an AI SDR actually does in the first 60 seconds of a lead

The single largest controllable variable in inbound conversion is how fast the first meaningful response arrives. Not an autoresponder — a real reply that engages with what the person asked. Most teams measure this in hours. The gap between minutes and hours is where the majority of inbound pipeline is lost.

Second 0 to 5: acknowledge and enrich

The lead submits. Before any message goes out, the record is enriched — company, size, industry, existing CRM history, whether this is a returning visitor or a new one, and which pages they read before submitting. That context is what separates a useful first reply from a template.

Second 5 to 30: the first reply

• Answers the actual question asked in the form, in the channel the lead used.
• References what they were looking at, so it reads as a response rather than a broadcast.
• Asks exactly one qualifying question — not a form's worth.
• Offers a concrete next step with real availability, not 'someone will be in touch'.

Second 30 to 60: route and prepare

In parallel, the lead is scored against your criteria and routed. A high-fit lead alerts the right salesperson immediately with a one-paragraph brief: who they are, what they asked, what they read, and what the AI already told them. A poor-fit lead goes into nurture without consuming a salesperson's afternoon.

What it should never do

• Quote prices or commit to timelines it cannot verify.
• Keep talking when the lead asks for a human. The handover should be one message away, always.
• Pretend to be a named human being. Say it is an assistant; the conversion difference is negligible and the trust difference is not.
• Continue a sequence after the lead has replied to a salesperson. Nothing kills a deal faster than the robot and the rep talking over each other.

How to tell whether it is working

Track median time to first meaningful reply, reply rate to that first message, and the share of leads that reach a booked meeting. If reply rate is high but meetings are flat, the qualification question is wrong. If meetings are up but close rate falls, the scoring is too generous. Both are fixable in a week — but only if you are measuring at that granularity.