What Happens in the 30 Minutes After a Prospect Says "Interested"

11:04am — a prospect replies to a cold email with four words: "Sure, what does this involve?"

11:04am is the moment that reply is worth the most it will ever be worth. The prospect is at their desk, in their inbox, thinking about your product for the only time this week. Every hour after that, the value decays — not because they change their mind, but because the context around your email gets buried under everything that arrives after it.

Most teams have a system for sending and no system for the thirty minutes after someone replies. That asymmetry is where a large share of qualified pipeline quietly disappears.

The two clocks that matter

There are actually two distinct response windows, and conflating them is why teams measure this badly.

Clock one: classification. How long between a reply arriving and someone knowing what kind of reply it is? Interested, not interested, wrong person, has a question, out of office. A workable target is 15 minutes — fast enough that the routing decision happens while the conversation is still live.

Clock two: response. How long between knowing it's a positive reply and your actual answer landing in their inbox? A workable target during business hours is 30 minutes, with a meeting-booking attempt triggered within the first hour.

Teams that track only the second clock miss the real problem, which is usually the first. A reply that sits unclassified in a shared inbox for six hours has already lost the window, regardless of how fast the eventual response was written.

Why "wrong person" replies are the most wasted

Every category of reply has a defined correct next action, and most teams only have one for "interested."

• —Interested → book the meeting, immediately, with concrete slots rather than "when works for you?"
• —Has a question → answer it directly and accurately; a deflection here reads as evasion and reliably ends the thread
• —Wrong person → this is the one teams throw away. Someone just told you they aren't the buyer, which means they probably know who is. The correct action is asking for the right contact, not marking the lead dead
• —Not interested → pause the sequence immediately and update the record; continuing to send after an explicit no converts a neutral non-buyer into someone who marks you as spam
• —Out of office → reschedule against their return date rather than continuing the cadence into an empty inbox

The "wrong person" case is worth dwelling on because it's a free referral inside your own pipeline that most teams treat as a rejection.

The classification confidence problem

Automated reply classification works well on clear cases and poorly on ambiguous ones. "Thanks, but we're all set for now" could be a soft no, a timing objection, or a polite brush-off that a well-timed follow-up would reopen.

The safeguard is a confidence threshold. When classification confidence falls below 80%, the reply routes to human review rather than to an automated action. This is unglamorous and it's the thing that keeps automated reply handling from producing embarrassing outcomes — an automated "great, let's book time!" sent in response to a reply that was actually a complaint does more damage than a slow response ever would.

The same principle applies in reverse: never send an automated reply to a message that already received one. Duplicate automated responses in a single thread are the most recognizable tell that a human isn't reading.

Why speed beats polish here

There's a persistent belief that a considered response outperforms a fast one. For inbound-positive replies specifically, the evidence generally runs the other way — response speed correlates more strongly with meeting conversion than response length or sophistication does.

The intuition is simple: a fast, adequate reply reaches a prospect who is still thinking about you. A perfect reply four hours later reaches a prospect who has moved on to eleven other things. The second email is objectively better and performs worse.

This doesn't mean careless. It means the bar for "send it" should be accurate and specific, not optimally crafted.

What to actually build

If you're fixing this, the order that works:

1. 1.Get every reply into one place. Replies scattered across four mailboxes, two reps' inboxes, and LinkedIn cannot be responded to in thirty minutes by anyone.
2. 2.Classify on arrival, not on review. Classification should happen when the reply lands, not when someone opens the inbox.
3. 3.Define the next action per category — including for "wrong person."
4. 4.Set a confidence floor below which a human looks at it.
5. 5.Log everything with timestamps, so you can actually measure both clocks instead of estimating them.

Most teams discover, on measuring this for the first time, that their median classification time is measured in hours and their "we respond fast" belief was based on the times someone happened to be watching.