Why a 22% LinkedIn Accept Rate Means Your Targeting Works
A 22% LinkedIn connection accept rate sits mid-band for cold B2B lists. Here is what actually moves it, how to test your AI layer, and the meeting math.
Is a 22% LinkedIn connection accept rate good?
It is mid-band. Across our campaigns accept rates land between 20 and 30%, so 22% sits at the bottom end of normal for a cold B2B list. Good enough to keep running. It is not proof that the AI layer did anything, because the same list sent by hand usually lands in the same range.
What actually moves that number is narrow. Title match, whether you share connections, and what your headline says when the person clicks through before deciding. Expandi ran 13.2 million connection requests across 13,302 accounts and found acceptance varies by industry, which is another way of saying the list decides most of it. If you are sending to VP Engineering at Series B SaaS companies and getting 22%, swapping the copy generator will not take you to 35%. Swapping to Heads of RevOps at companies that just posted a hiring ad for an SDR might.
Did the AI produce the 22%, or did the targeting?
Targeting, almost certainly. A connection request with a note gives the reader three lines and your profile. Most people accept or ignore based on who you appear to be, not on how well the opener was personalised. The AI earns its keep after the accept, in the 97 conversations that follow.
We wrote this out in more detail in why a 22% accept rate is a targeting win. The quick test: run the same list with a plain one-line note and no personalisation for a week. If accepts hold near 22%, the AI was writing openers nobody read closely. If they drop to 12%, keep paying for it.
If accepts are under 15%, the profile is usually the problem before the list is. A headline that says "Founder | Helping B2B teams scale" tells a prospect nothing, and the LinkedIn headline rewriter is free and takes two minutes.
What does 22% actually turn into in meetings?
One warm account sending 20 invites a day across 22 working days is 440 invites a month. At 22% that is 97 new connections. At a 12 to 18% first-message reply rate plus 5 to 8% on follow-ups, you get roughly 19 to 22 replies. Meetings land at 3 to 5% of invites.
| Stage | Published rate | At 440 invites a month |
|---|---|---|
| Connection accepted | 20-30% | 88 to 132 connections |
| Reply to first message | 12-18% | 12 to 17 replies |
| Reply to follow-ups | 5-8% | 4 to 7 more replies |
| Meeting booked | 3-5% of invites | 13 to 22 meetings |
The meeting row assumes replies get answered the same day. If a prospect says "send me something" on Tuesday and you reply Friday, that one is gone. We broke the full funnel down in what a 22% accept rate turns into.
Is AI LinkedIn outreach better than cold email?
Per message, yes. Cold email averages around 3.5% reply, 5 to 8% for good campaigns, and LinkedIn roughly doubles that. AI-assisted outreach gets 10.3% response against 5.1% for traditional cold email. Per month, email still wins on volume.
The ceiling is the difference. One LinkedIn account tops out near 500 invites a month before you start risking restrictions. One email domain with three mailboxes sends 3,000 and nobody notices. So if you need 40 meetings a month and your deal size is $6,000, LinkedIn alone will not get you there on one account. We argue the sequencing in whether to automate LinkedIn before cold email, and the short version is that LinkedIn is the better first channel and the worse only channel.
Why do some sites claim 70-90% response rates?
Because they are describing a different activity. One benchmark page lists hyper-personalised AI outreach at 70 to 90% response, and a white-label guide reports platform averages of 40% acceptance and 42% response. Those numbers come from small warm lists, not cold B2B SaaS prospecting at 440 invites a month.
Thirty people you met at a conference, messaged by hand, referencing the session you both sat in, will reply at 60%. That is real and it is also not automation. When you scale the same approach to 400 strangers a month, replies settle at 12 to 18%. A founder who reads 70% and budgets for it will conclude after month one that their tool is broken when it is performing normally.
What part of LinkedIn outreach is worth automating?
The replies. Sending is the part every tool already does well, and it is also the part that takes the least time. Twenty invites a day is about fifteen minutes. Answering 22 conversations, each with a different objection, across a week where you are also shipping product, is the part that quietly stops.
Expandi, Meet Alfred, Closely and Salesforge all send well. Expandi runs cloud-based sequences and offers a seven-day free trial, Salesforge covers LinkedIn and email with 21+ languages, Closely claims personalised messaging lifts replies by up to 35%. In all four cases you or someone on your team still writes the copy and answers the replies.
Gelee is the other side of that. It replies in your voice, handles objections from your playbook, and books the meeting. There is no self-serve signup, setup takes 15 minutes, and pricing is shared on a demo call.
How many invites a day is actually safe?
20 to 25 a day on a warm account, 5 to 10 while a new account ramps over the first two or three weeks. LinkedIn does not publish a limit and it moves with account age, profile completeness, and how many people ignore you. Accounts that get ignored a lot get throttled faster.
Two practical consequences for a founder at 22%. First, your maximum is roughly 500 invites and 13 to 22 meetings a month from one account, so the only ways past that are more accounts or a second channel. Second, pushing to 50 a day to hit a quarter target is how accounts end up restricted, and a restricted account takes the pipeline with it.
When is this the wrong thing to automate?
Four situations. Deal sizes under $2,000 do not carry the cost of any of these tools. Buyers who are not on LinkedIn, like most trades, local services and plant managers, cannot be reached there. No product-market fit means 97 new conversations produce 97 polite declines. And phone-first sales needs a dialler.
One more, and it matters for a team already at 22%. If you have two SDRs with time on their hands, buy them better tooling, not an AI SDR. A multi-account sending platform at a few hundred a month plus people who already answer replies beats paying for a system that replaces work your team is doing anyway. Gelee is for the founder answering LinkedIn DMs at 11pm, or the team of two who cannot hire a third.
So is AI LinkedIn outreach the best automation you can run?
For a B2B SaaS founder with a $10,000+ deal size whose buyers are on LinkedIn, it is the highest-return channel to automate first, and 22% accept is a working starting point. The AI is not why it works. The list is why it works, and the AI is why the 22 replies get answered.
What would change our answer: if your accept rate is already 22% and your reply rate is 4%, do not add tooling. Fix the first message, because the people saying yes to the connection are saying no to the pitch, and no amount of automation improves an offer nobody wants. Overloop puts it fairly in their own comparison when they say there is no universal winner among these tools. The channel is the win. The software is the part you can change next month.