LinkedIn connection acceptance rates: what 59% means and what it does not

What is a connection request acceptance rate?

A LinkedIn connection request acceptance rate is accepted requests divided by requests sent, over a stated window, on a stated segment. All three parts of that sentence matter and two of them get dropped constantly.

The first measurement decision almost nobody states is how pending requests are treated. A request sent on Friday and accepted on Tuesday spends four days pending, so a denominator counting every request from the day it left understates your trailing rate, and understates it more the faster you send. The second is withdrawals: depending on the tool, withdrawn invitations either drop out of the denominator or stay in it, and dropping them raises the reported rate without producing a single extra acceptance.

Two teams running the same tool against the same list can publish different rates on identical activity because of those two choices. Ours, so the rest of this page is readable: accepted divided by sent, rolling four week window, withdrawn requests left in the denominator.

What is a good LinkedIn acceptance rate for cold outreach?

There is no good acceptance rate that holds across audiences, and the useful benchmark is your own trailing four week rate, not anybody's published figure. Acceptance moves with seniority, industry, geography, how recognisable the sender is, and whether shared connections show on the request. A founder writing to founders in a market where the company name means something is running a different experiment from a sales title writing to the same people, and that spread is wider than the spread between a good note and a bad one.

Then there is who publishes the numbers. Almost every acceptance benchmark on the open web comes from a company selling LinkedIn automation seats. We sell campaign management instead of seats, which is a different bias and not the absence of one, so the condition goes next to the number and you decide whether your situation resembles it.

What did we measure, and under what conditions?

We recorded 59% connection request acceptance on the control cell of a white label programme running across a set of advisor workspaces, counted as accepted requests divided by requests sent. On that programme LinkedIn and email ran from the same senders, against the same target population, in the same window, and the sender was someone the audience read as a peer, not as a vendor.

That last condition is doing most of the work in the number. It is not a well written note beating a badly written one. It is a recognisable person contacting people like themselves, the most favourable setup the channel offers.

The total connection requests sent and the exact date window are not stated here, because they have not been pulled from the campaign record yet. The condition is stated instead, because the condition is what makes the number usable, and a sample size we have not confirmed would be a worse thing to publish than a gap we have named.

What does 59% not mean?

It does not mean 59% is the LinkedIn benchmark. It is one figure from one vertical with one sender profile, and a broader or colder list produces a materially lower number. We would expect ours to fall on a broader list too.

It is not a target. Building a plan around a rate produced by the most favourable sender and audience combination available is how programmes get declared broken in month three, when the rate they get is the one their actual list supports.

It is not a positive outcome. An accepted connection is a permission, not an interest signal, and plenty of people accept because accepting is the socially cheap option and never open the message.

And it says nothing about reply rate. Different denominator, different behaviour, and the gap between an acceptance and a booked meeting is where the real work sits.

Why is an aggregate acceptance rate the most misleading number in the channel?

An aggregate acceptance rate averages away the exact information you need to act, which is why we never report one on a client campaign. Call this the Segment Split Rule: acceptance rate is reported by segment or it is not reported. The figures below are illustrative arithmetic, not campaign data.

A 35% aggregate reads like a campaign that needs a rewrite. The split says something else entirely: A is working, B is acceptable, and C is burning a third of your weekly capacity on people who will not accept, while dragging the sending account's quality signals down for everything behind it. The aggregate tells you to rewrite. The split tells you to cut.

Splitting acceptance by segment is the single most useful reporting change we make on an inherited campaign, and it usually takes an afternoon. It also feeds the decisions. The group runs fixed kill and scale thresholds on positive replies per send, published in full with their denominator by the property that owns them. Those thresholds need positive reply volume to fire, which takes weeks. Segment level acceptance is available in days and tells you which way the decision is heading.

Why is my LinkedIn acceptance rate dropping?

A falling acceptance rate has four common causes, and checking them in a fixed order saves you the three weeks most teams spend rewriting. Call this the Acceptance Rate Diagnostic. The order is the point.

List exhaustion is the one that gets misdiagnosed most often. The good fit segment gets worked through in the first weeks, the campaign keeps running at the same volume, and the remaining sends are drawn from progressively worse fit prospects. The rate does not fall because the copy decayed. It falls because the list did.

Copy iteration is deliberately not on that list. It is what you do after all four come back clean. On our own client accounts one of the group's internal monitoring desks watches segment level acceptance week over week, because a drift caught in week one is a filter change and a drift caught in week five is a rebuild.

Why is acceptance mostly a targeting problem and not a copy problem?

Acceptance is mostly a targeting problem because the recipient decides in a few seconds based on who you appear to be relative to them, and copy cannot change who you appear to be. Run identical request text against a list of comparable seniority in the sender's own industry and against a list assembled by job title alone across four countries, and the gap between those two rates will be larger than any gap a rewrite produces.

The least popular sentence here: most teams spend three weeks on wording when a segment filter would have fixed it in an afternoon. Where the list was built badly to begin with, the fix is list construction, not another copy round.

How does a falling acceptance rate feed back into your sending ceiling?

A falling acceptance rate reduces how much you can send, which makes this a capacity problem, not a reporting problem. Acceptance rate is the input the platform's quality check watches most closely, because it is the cheapest available proxy for whether your volume is wanted. Low acceptance makes the platform more restrictive, restriction makes the programme feel behind, feeling behind makes teams push volume, and pushing volume makes acceptance worse.

The recovery move is counterintuitive and almost nobody makes it: cut volume, fix the targeting, let the rate recover, then rebuild. Teams do the opposite, because the plan says a number and the number does not care that the list stopped working.

The ways out are all slow, which is the part planning skips. Adding accounts is a preparation decision measured in weeks, not a purchase, and the group's onboarding and email warm up windows are published on the Outbound Pros site. The client approves the lead list before anything goes out, which is the one gate where a bad segment gets caught cheaply and the gate people most want to rush. Neither of our two motions raises the ceiling either. WideNET tests angles systematically across the full addressable market, Spearhead runs signal triggered work on the hottest slice, and both run inside the capacity you already have.

What acceptance rate should you plan a campaign around?

Plan around a range, model the low end, and treat anything above it as headroom, not as the plan. A planning tool that guesses this number for you is a sales tool, which is why the calculator refuses a default acceptance rate. If you have four weeks of your own data, use it. If you do not, model at a rate low enough that you would still fund the programme.

Where this approach fails, plainly. Segment level acceptance needs volume to mean anything. If your addressable market is 300 named accounts, splitting acceptance three ways gives you three numbers built on noise, and the right instrument is a human working the list. If your buyers do not open LinkedIn in a normal working month, none of this applies and the honest answer is a different channel. And optimising for acceptance in isolation optimises for a permission instead of for revenue: you can raise the rate by targeting people who accept everything, end up with a larger connection base, and book fewer meetings than before.

One definitional note. Everything here is a rate, so the denominator is sends. A ratio, such as positive replies divided by total replies, has a different denominator, and placing one beside the other produces a comparison that looks impressive and means nothing. That argument sits with our sibling property AllboundPros, which owns rate definitions across the group.