Inside one minute. My DM system answers the typical lead 30 seconds after they message, and 9 out of 10 inside 56 seconds. The research says a lead starts dying within the first hour. In a DM it dies faster, because the person is still holding their phone and the next reel is one swipe away.

I don’t guess at this. I run my business on an AI agent fleet, and every reply it sends is logged with the number of seconds since the person wrote. This article gives you those numbers, the research worth trusting, and the setup that gets you under a minute without you living in your inbox.

Why does speed matter more in a DM than in a web form?

A web form is a request. A DM is a conversation that is still happening.

When someone comments a keyword on your reel and then replies to your DM, they are in the app, with intent, right now. You have their attention for the length of a scroll. Answer while they’re still there and you are part of the conversation they started. Answer in an hour and you are a notification they have to remember the context for.

People don’t leave because they stopped wanting the thing. They leave because the moment passed and nobody was there when it was happening.

What does the research actually say about lead response time?

Most speed-to-lead articles recycle the same stats. You’ve probably seen “respond in 5 minutes and you’re 21 times more likely to qualify the lead.” I’m not using it. I couldn’t trace it to a primary source I would put my name on, so it stays out.

Here is what I can cite.

Harvard Business Review, 2011. Researchers audited 2,241 US companies by sending each one a web lead and timing the response. 37% responded within an hour. 23% never responded at all. Firms that tried to contact the lead within an hour were nearly 7 times as likely to qualify it as firms that tried an hour later, and more than 60 times as likely as firms that waited 24 hours or more. (The Short Life of Online Sales Leads, HBR) That study is 15 years old and it measured web forms, not DMs. The direction still holds.

Workato, first published 2022. Workato filled out demo requests at 114 B2B companies. Only 1 of the 114 sent a personalized email within 5 minutes. The average email response took 11 hours and 54 minutes. (Workato lead response time study)

So the bar is low. Most businesses answer in hours. Answering in under a minute is not a small edge. It puts you in a different category.

What does my own reply data show?

These are from my DM responder’s log over six days, October 7 to October 13, 2026. It sent 1,125 replies in that window.

What I measured	Result
Typical reply (median)	30 seconds after the person wrote
90% of replies sent within	56 seconds
Replies sent within one minute	91%
Replies that took longer than 2 minutes	3%
Checkout links, typical time to send	29 seconds
Replies sent outside my one scheduled human shift	More than 2 out of 3

Read that last row twice. More than two out of every three replies went out at night, early in the morning, or on a weekend. That is when your leads are actually on their phones. If a human has to be awake for a lead to get an answer, most of your leads are waiting.

And it isn’t just fast. When I audited the replies to my opt-in question over a 36 hour window in September, 98% of the people who asked for the link got it. I went looking for the 2%. Two of them were real misses. One person typed the keyword and then “thank you” a minute later, and the system read the thank you as a goodbye. The other typed the keyword with a typo. Both are fixed. That’s the job: find the miss, name it, close it.

Why is the average the wrong number to watch?

Because a handful of slow threads hide inside it.

My average reply time in that same window was 527 seconds, almost 9 minutes. My median was 30 seconds. Same week, same system. The average is dragged up by a small number of threads that were held for a human on purpose, a complaint, an objection, someone upset, and answered hours later by design.

If I watched the average, I would think the system was slow. If I only watched the median, I’d miss the people stuck in the tail. Watch both the median and the 90th percentile. Then look at the slowest few by name every week and ask why.

Where do slow replies come from?

In my system, three places.

Threads that should go to a person. Objections, refunds, anger, crisis, and “are you a bot” get escalated to a human instead of answered by the machine. Those are slower on purpose. I’d rather be slow than wrong with someone who is upset.

Platform lag. In early October I watched some replies wait 2 to 5 minutes. The cause was not my code. The messaging platform’s search index was behind on some threads, so my responder couldn’t see the new message yet. I added a direct check on every thread where I had just asked a question, so the answer gets caught before the index catches up.

Silent failure. A job can say “ok” and send nothing. That is why a separate watcher checks every job in my fleet every 15 minutes. Speed you can’t verify is a guess.

How do you get under a minute without living in your inbox?

You stop treating every DM as a writing task. Most of them aren’t.

1. Give people one clear word to send. A keyword on a reel, then one question in the DM. A clear yes is easy to recognize and easy to answer instantly.
2. Answer clear requests with rules, not a model. When someone says yes to the link, send the link. No AI needed. That’s why my links go out in a typical 29 seconds.
3. Let AI write only the bounded replies. A short, voice-checked answer to a real question. No prices it wasn’t given. No promises. If the reply fails a check, it doesn’t send.
4. Escalate the hard ones. Money, anger, refunds, and anything sensitive go to a human with the full thread attached.
5. Cap the machine. My responder stops ordinary replies after 8 in one thread in one day, and a failed send gets at most 3 attempts before it goes to a person. Caps keep a fast system from becoming a spam system.
6. Watch it from the outside. A second system checks the first. Every 15 minutes.

Is faster always better?

No. Fast and uninvited is spam.

My rule is absolute: the system only messages people who asked. Someone who typed the keyword, replied to the question, or wrote to me first. Nobody else gets a DM, no matter how fast I could send it. Speed is a reward for intent. It is not permission.

And a reply in 30 seconds that says the wrong thing is worse than a good reply in 3 minutes. Get it right, then get it fast. In that order.

What should you measure starting this week?

Three numbers, measured from the moment the person wrote to the moment your reply was delivered:

• Median reply time. Your typical lead’s experience.
• 90th percentile. Your slow end. This is where sales leak.
• Share answered outside business hours. If you can’t answer when your leads are awake, this shows you how much you’re missing.

Measure the delivered reply, not the time your tool says it ran. A green dashboard is not a receipt.

Want this built inside your business?

This is what Operator OS does. Three connected engines installed in your business in 90 days, plus ascension, carrying every client from first comment to delivered result, so only the real exceptions reach you.

Operator OS isn’t bought. It’s applied for. If your leads are waiting on you, apply for Operator OS here.

FAQs