LinkedIn reply rates fell 37% in a year. Your templates did that.

Somewhere in your sent folder is a message that starts with "I came across your profile and was impressed by..."

Delete the template it came from. Here's the data.

The numbers nobody sending sequences wants to see

Expandi, a LinkedIn automation company, published benchmarks this year from 13.2 million connection requests across 70,000+ campaigns. Read that source carefully: an automation vendor measuring its own users, publishing numbers that hurt its own pitch.

What they found:

Connection-request reply rates fell 37% in twelve months. From 3.5% in May 2025 to 2.2% in April 2026. Their own explanation: "increasingly generic connection-note templates as more senders converge on similar phrasing." Everyone bought the same tools, loaded the same scripts, and hit the same prospects.

Template-based campaigns underperform, and it's measurable. Templated campaigns averaged 8.62% replies against a 10.3% overall average. The shortcut costs you responses. That's the whole trade.

Here's the part most people miss: message reply rates held steady at 10 to 11%. LinkedIn outreach is not dying. People still answer messages. They stopped answering a specific kind of message: the one they've already seen nine times that week from nine different senders.

The channel is fine. The template is dead.

Why this is happening now

AI happened. Not to outreach in general, to everyone's outreach at the same time.

When writing a message costs nothing, everyone sends more of them. When everyone's messages come from the same handful of models with the same handful of prompts, they converge on identical phrasing. Your prospect doesn't need an AI detector. Pattern recognition does the job: same opener, same flattery, same ask.

The hiring world is the clearest picture of where this ends. The Markup posted a single job listing and got 400+ applications in 12 hours, most of it machine-generated. Fortune reports 53% of job seekers got ghosted by employers in 2026, up from 38% two years ago, as AI-automated pipelines replaced human responses. Some tech employers have stopped reading inbound applications entirely.

That's what saturation looks like at the end state. Volume goes up, trust goes to zero, and everyone stops reading.

Your prospects' inboxes are on the same curve. The question is what you do while your competitors keep feeding it.

If you run a staffing or recruiting firm, this is your BD problem

StaffingHub's benchmark survey found that finding new clients is now the top challenge for staffing agencies: 23% of firms name it their number one problem, up from 16% the year before. Candidate-side pain is falling at the same time. Bullhorn's industry survey of 2,300 firms says the same thing: winning new clients is the top priority, ahead of finding candidates.

The industry flipped. The bottleneck moved from talent to clients. And the default answer, load a sequence tool and blast hiring managers, is the exact behavior the Expandi data says is collapsing.

You know this from your own inbox. You can spot a templated message in half a second, because you've sent thousands of them. So can the VP of Engineering you're trying to reach.

What still gets replies

The same benchmarks that show templates dying show the alternative working. Independent 2026 studies put personalized connection requests around 45% acceptance versus roughly 15% for generic ones. Treat the exact numbers as directional, but the gap shows up in every dataset, every time, in the same direction.

What "personalized" means, because the word got ruined:

1. A reason to message this person this week. A posted role, a funding announcement, a team they're visibly growing. Not "I saw we're both in staffing."
2. Written for one reader. If the message still makes sense with a different name pasted at the top, it's a template, whatever tool produced it.
3. A human reads it before it sends. Not a human who wrote the prompt six weeks ago. A human who reads this message, to this person, today, and asks: would I send this if my own name were on it?

That third step is the one nobody wants to pay for, because it doesn't scale like software scales. It's also the only step the data says still matters.

The math has changed

The old playbook: send 1,000 templated messages, book meetings on volume. That worked when reply rates were high and inboxes were quiet. Both conditions are gone, and LinkedIn's own sending caps killed the volume game on top of it.

The new math: send 100 messages a real person would answer. Fewer sends, more conversations, and none of the account restrictions and burned goodwill that come with the blast.

For a staffing firm, one reply that turns into one client isn't a small win. It's recurring revenue on every contractor hour you place there, for as long as you hold the account.

What we do about it

This is the entire reason Inroads exists. We run LinkedIn outreach for founders and agency owners, and a person reviews every single message before it sends. Not a spot check. Every message. Across 65,000+ requests, 46.6% get accepted against a 26.4% industry figure, and 28.4% of those reply.

Your prospects are drowning in automation. Be the one message that reads like a person.

Sources: Expandi LinkedIn Outreach Benchmarks 2026 (13.2M connection requests); StaffingHub State of Staffing 2025; Bullhorn GRID Industry Trends; Fortune, March 2026; The Markup, January 2026.