By I pulled the timestamps on 2,511 outbound LinkedIn messages sent through Linkdify this summer. The median reply arrived about 14 hours after the message went out. Most of the replies that were ever going to come landed inside the first day. After that, the curve goes flat fast.
That one number changed how I think about follow-up cadence and about what an "AI SDR" is actually for. The full numbers are in our outbound benchmarks report, but the 14-hour figure is the one I keep coming back to.
Silence after two days is an answer
The standard playbook says wait three to five business days, then follow up. That advice comes from email, where your message gets buried under two hundred others and a nudge can genuinely resurface you.
LinkedIn is not email. A message sits in a short inbox, on a phone, with a notification attached. If someone was going to answer, they mostly already did. When you follow up on day four, the person you reach saw the first message and chose not to respond. That is a completely different situation from a busy person who forgot, and it needs a different message.
Why the "bump" follow-up is dead weight
"Just bumping this to the top of your inbox" assumes the problem was visibility. The data says visibility was fine. The problem was the message.
So a follow-up has one job: give the person a reason that didn't exist the first time. A specific thing you noticed about their company works. So does a question that is easier to answer than whatever you asked first. If you can't add anything new, send nothing. Silence costs you zero. A weak second touch costs you the option of coming back in three months with something real.
We built this into how Linkdify writes follow-ups. Before it drafts the next step, the AI reads what was already sent in the thread, so the second message doesn't restate the first one with a friendlier opener. That sounds obvious. It took watching a lot of campaigns to see how much automated outreach skips it.
Want LinkedIn outreach that runs itself?
Linkdify finds your leads, writes personalized messages, and sends them at safe, human-paced limits.
The bottleneck was never sending
Here is the part founders get wrong about AI SDRs. The pitch is volume. Ten times the outreach, no headcount. Now look at the funnel.
Across 8,400 connection requests we saw a 25.2% acceptance rate, roughly one in four. Across those 2,511 messages, 14.6% replied. Those are healthy numbers. They also mean that if you send 100 messages, you get about 15 conversations, and those 15 land in your inbox within a day, bunched together.
Fifteen real conversations in one day is a lot for a solo founder. Every one of them needs a considered answer from a human who understands the product and can hear what the prospect is actually asking. Automation got them to the door. It cannot sit in the meeting for you.
So when a tool promises to triple your volume, ask what happens to the 45 replies that arrive Tuesday morning. If the answer is "the AI handles those too," be careful. The first automated reply that misreads a question is the moment a warm lead becomes a story someone tells about how bad your outreach was.
My view after building this: automation earns its keep by sending fewer, better-targeted messages, timed so the replies arrive when you can actually answer them. Volume is the easy part. Nobody has ever been bottlenecked on their ability to click send.
What this means if you're doing outbound alone
Send in the prospect's morning, not yours. Replies cluster in the first day. You want that day to be one where you are at your desk, in their working hours. Linkdify has a per-campaign send window for exactly this reason. It is the least glamorous feature we have and one of the most used.
Cap volume at what you can hold conversations about. Twenty connection requests a day is plenty for one person. It also happens to be a pace LinkedIn tolerates without complaint, which is not a coincidence. The platform is tuned for human behavior, and human behavior is bounded by how many conversations one person can carry.
Treat day-two silence as feedback on the message. If a whole batch goes quiet, the pitch is off. Rewrite it before the next batch instead of bolting a fourth follow-up step onto a sequence that already isn't working.
Write the follow-up as a new message. If the first note was about a problem, the second should be about a result you got somewhere similar, or a question. Never a paraphrase of the first with "circling back" in front of it.
Want LinkedIn outreach that runs itself?
Linkdify finds your leads, writes personalized messages, and sends them at safe, human-paced limits.
What I'd change tomorrow
If you have a five-step sequence with bumps on days 3, 7, and 14, cut it to two. One opening message. One follow-up on day two or three that says something genuinely new. Then stop and let the pipeline breathe.
I don't have a clean number on what the later steps cost you, and I won't pretend to. But I have watched enough campaigns to believe those steps mostly reach people who already decided, and each one spends a little of your reputation with someone you may want to talk to next year.
The 14-hour median is the most useful number I own. It tells me LinkedIn outreach is closer to a conversation than a campaign. Conversations don't improve when you talk louder or more often. They improve when you say something worth answering, and then you are there when the answer comes.