LinkedIn Prospecting Benchmarks 2026: What Public Reports Measure and How to Read Them

The largest public LinkedIn prospecting benchmarks come from automation platforms and lead-gen agencies. Expandi's 2026 report, built on 13.2 million connection requests, puts acceptance at 28.5% and reply at 10.4% of messages sent. Belkins' study lands at 26% acceptance and a 7.2% reply rate. HeyReach, across 96,051 campaigns, reports a median acceptance rate of 20.75%.

Those numbers are useful, but only if you read them for what they are: averages of the people who use a given tool, each built on a different denominator. Put two of them side by side without checking the fine print and you'll draw the wrong conclusion about your own funnel.

This guide walks through what each report actually measured, why the figures disagree, how much the industry you sell into moves the numbers, and how to build your own benchmark from your data: fixed windows, explicit denominators, and weekly cohorts.

What Public LinkedIn Prospecting Benchmarks Say

LinkedIn prospecting benchmarks are reference rates for each stage of the outbound funnel: connection acceptance, reply, and meeting conversion. The three most complete public reports from 2026 measure similar things, but not the same things:

Three caveats before you use any row of that table:

• Expandi and Belkins are not independent sources. Belkins' large dataset comes from Expandi itself, a different cut (calendar 2025). Similar numbers are the same platform measured twice, not two confirmations.
• "Message" in Expandi's data covers a lot. The 10.4% includes welcome messages, follow-ups, group and event messages, InMails, and emails. It is not just the first touch after a connection is accepted.
• These are averages of automation users. Expandi states its report is a sender-side benchmark: it shows outcomes for users in a given segment, not how recipients in that segment behave.

Why the Numbers Disagree

At first glance, LinkedIn reply rates range from 7.2% (Belkins) to 25.19% (HeyReach). Almost all of that gap is definition, not performance.

The denominator changes. Expandi divides replies by messages sent. HeyReach divides replies by conversations started, which is per person rather than per message. In a three-message sequence where the prospect only answers the third, Expandi counts 1 reply in 3 messages and HeyReach counts 1 reply in 1 conversation. Belkins uses two denominators on the same page: replies ÷ messages on the Expandi data, and replies ÷ connected prospects on its own campaigns.

Mean and median tell different stories. HeyReach's mean acceptance is 23.85%; the median is 20.75%. When the mean sits above the median, a handful of very strong campaigns are pulling it up. The typical campaign looks more like the median.

Who's in the sample decides the result. Users of an automation tool, clients of a lead-gen agency, and an in-house team prospecting its own network are different populations. Belkins' 18.7% acceptance and Expandi's 28.5% don't tell you who prospects better; they tell you different lists, messages, and audiences produce different numbers.

A connection note moves the two rates in opposite directions. On the Expandi data Belkins analyzed (11.5 million requests in 2025), requests with a personalized note had 25.3% acceptance and an 8.2% reply rate; without a note, 27.6% and 5.3%. The details are in our guide on connection notes and acceptance rate.

Your Industry Matters More Than the Platform Average

The most useful advice in Expandi's report is to ignore the platform average and benchmark against your industry row. A few rows from the published table (this is the sender's industry, not the recipient's):

Source: Expandi, May 2025–April 2026. Industries with 15 or fewer accounts are flagged by the report itself as too small to be reliable.

Belkins looked at the recipient side. HR and talent professionals replied at 10.9%, C-level executives at 7.0%, and founders and owners at 6.4%. By industry, Belkins' reply rates ranged from 10.5% in venture capital and private equity down to 4.2% in outsourcing.

Sender title barely matters. In Expandi's data, C-level senders saw 29.4% acceptance and entry-level senders 26.3%, a three-point band.

Do Benchmarks Vary by Country?

Yes, at least in the one public dataset that breaks them out. Belkins' reply-rate-by-country chart (Expandi data, 2025) covers the 17 countries with more than 100,000 contacts. It runs from 5.27% in India to 10.49% in the Netherlands. In the Belkins chart, the US sits at 6.01% and the UK at 6.09%, near the bottom. Belkins' advice to US teams: treat roughly 6% as the baseline rather than the 7.2% global average.

What that chart doesn't cover matters as much. Expandi and HeyReach publish no country breakdown, and large markets such as Brazil are not among Belkins' 17 countries. If you sell into a country that isn't on the list, there is no public benchmark for it, and anyone quoting one should be able to show the sample. Our Brazil guide looks at what is and isn't known for that market specifically.

Chattie does not publish platform usage data, so you won't find a "Chattie benchmark" in this post.

End-to-End Funnel Math: What You Can and Can't Multiply

It's tempting to assemble a funnel by taking acceptance from one source, reply from another, and meeting rate from a third. It doesn't work: each number has its own denominator and population, and multiplying rates from different samples produces a result no real operation ever measured.

The only public source with all three stages measured on the same population is Belkins' own client data: 18.7% of invited prospects accepted, 17.6% of connected prospects replied, and 1.3% of connected prospects booked a meeting. Running the math on those figures (18.7% × 1.3%) gives about 0.24% of invited prospects, or roughly 2 to 3 meetings per 1,000 requests. That's arithmetic on Belkins' agency campaigns from 2025, not a forecast for your team.

For the bottom of the funnel (show rate, opportunity, close), we found no public LinkedIn-specific data with a stated methodology. Use your CRM. Our B2B sales cycle guide helps structure those stages.

How to Build Your Own Benchmark (How to Measure It in Your Operation)

The benchmark worth having is your own, measured the same way every time. Six steps:

1. Write the definitions before you count.

The 15-day acceptance window isn't arbitrary for Chattie users: that's when the product automatically withdraws pending requests. The other windows are suggestions; what matters is never changing them mid-stream.

2. Count per person, not per message. If you count replies per message sent, every extra follow-up drags the rate down even when it produces more conversations. Per person, the question is simple: of this list, how many replied? The cadence touchpoint guide explains why per-message and per-person views diverge.

3. Group by weekly cohorts. Every request sent in week 1 is cohort 1, and each of its metrics is measured inside the fixed windows, even if the reply lands in week 3. That keeps weeks comparable and stops you from confusing "this week's list was worse" with "LinkedIn changed."

A made-up spreadsheet, just to show the format:

4. Keep apart what shouldn't mix. Cold outreach and outreach to people who already engaged with you (warm) have different rates by nature. The same goes for segments, languages, and sender accounts. Give each its own cohort.

5. Know how much your number wobbles. With a small sample, the measured rate can sit far from the true one. Using the standard margin-of-error formula (95% confidence), in a hypothetical example with 28% acceptance:

• with 100 requests, the margin is about ±8.8 points (the true rate could be anywhere from 19% to 37%);
• with 400 requests, about ±4.4 points;
• with 1,000 requests, about ±2.8 points.

For a 10% reply rate, the margin is ±5.9 points with 100 people and ±3.4 with 300. These are arithmetic, not market data.

6. To compare two versions, size the test first. Using the two-proportion formula (95% confidence, 80% power), in hypothetical examples:

• acceptance of 28% vs. 35%: about 690 requests per version;
• reply of 10% vs. 15%: about 690 connected prospects per version (at 28% acceptance, close to 2,500 requests per version);
• meeting rate of 1% vs. 2%: about 2,300 connected prospects per version.

The takeaway: acceptance tests take weeks; reply and meeting tests need far more volume. To pick what to test first, see how to increase your LinkedIn reply rate, the message structures that get replies, and how to personalize messages at scale.

Using Your Benchmark to Find the Bottleneck

Once you have three or four cohorts measured the same way, compare each stage against your own baseline, not a platform average from a different population:

Change one thing at a time. If the copy and the list change in the same week, you can't tell which one made the difference. The B2B prospecting cadence guide covers the sequence structure that makes this kind of test possible, and the technical guide to LinkedIn B2B prospecting covers the full setup.

Where Chattie Fits

Chattie doesn't publish usage data, so it doesn't offer a "platform benchmark." What it does is make volume predictable, which helps you plan measurement:

• Fixed daily caps: 25 connection requests, 40 conversations, and 200 LinkedIn actions per day, sending only during business hours. Those caps are fixed by default — above 25 connection requests per day, raising them requires the user's own risk consent — and they're what keep account restriction risk low (never zero).
• Automatic withdrawal of pending requests after 15 days, which matches the acceptance window suggested above.
• Three sending modes: in Autopilot the AI sends the message; in Semi-pilot the AI sends until the lead replies, then a human approves each response; in Copilot the AI drafts and a person approves before it goes out.

At 25 requests per business day, one account sends about 125 requests a week. An acceptance test with 690 requests per version (1,380 total) takes about 11 weeks on a single account. The Growth plan covers 3 LinkedIn accounts; randomizing versions within each account, the same test fits in about 4 weeks.

FAQ: LinkedIn Prospecting Benchmarks 2026

What is a good LinkedIn connection acceptance rate for B2B outreach?

In the public 2026 reports, acceptance sits between roughly 21% and 29% depending on the source and on mean versus median: 28.5% at Expandi, 26% in Belkins' study of Expandi data, and a 23.85% mean (20.75% median) at HeyReach. Industry moves it more: among Expandi's highest-volume industries, it ranges from 26.0% in financial services to 36.5% in staffing and recruiting. The best reference is your own rate, measured with a fixed window.

What is a realistic LinkedIn reply rate for cold B2B outbound?

It depends on the denominator. Per message sent, Expandi reports 10.4% and Belkins 7.2%. Per conversation started, HeyReach reports a 22.22% median. Before comparing your rate with any of them, check that you're dividing by the same thing.

Why do LinkedIn benchmark reports disagree so much?

Mostly because they define the rates differently and measure different populations. One counts replies per message, another per conversation; one reports the mean, another the median; one covers automation users, another an agency's client campaigns. Expandi and Belkins also share data, so their agreement isn't independent confirmation.

Do LinkedIn reply rates vary by country?

In Belkins' analysis of 2025 Expandi data, reply rates across 17 countries ranged from 5.27% in India to 10.49% in the Netherlands, with the US at 6.01%. Countries outside that list, including Brazil, have no public benchmark with a stated sample.

How many connection requests do I need before trusting my acceptance rate?

At around 28% acceptance, 100 requests give a margin of error of about ±8.8 points, 400 requests about ±4.4, and 1,000 about ±2.8. To compare two versions and detect a move from 28% to 35%, the standard calculation calls for about 690 requests per version. These are hypothetical calculations; redo them with your own baseline.

Should I optimize acceptance rate or reply rate first?

Start where your own cohorts show the biggest drop relative to your baseline. Acceptance is the fastest to test because it needs the fewest sends, so it's a practical first lever. Watch both together, though: in the Expandi data Belkins analyzed, adding a note lowered acceptance (25.3% vs. 27.6%) but raised reply rate (8.2% vs. 5.3%).

Conclusion: Calibrate, Don't Chase

Public LinkedIn benchmarks are a useful range, as long as you read them with the denominator in hand: roughly 21% to 29% acceptance in the large automation datasets (Expandi, Belkins, HeyReach), 7.2% to 10.4% reply per message sent (Belkins, Expandi), and an industry spread wider than the gap between sources. None of them is a target, and none of them measures your audience.

What to do with that: use the industry tables as a starting point, define your metrics with fixed windows and denominators, track weekly cohorts, and compare each stage against your own baseline. Within a few weeks you'll have a benchmark that tells you more than any platform average.

If you want to prospect with predictable daily volume and caps that never go up, explore Chattie here.

References

External sources cited in the body of this article, in order of appearance.

• Expandi — Stefana Zarić, "LinkedIn Outreach Benchmarks 2026: Acceptance, Reply, and Message Response Rates from 13.2M Outreach Attempts," May 19, 2026: platform averages, industry and sender-seniority tables, methodology (https://expandi.io/blog/linkedin-outreach-benchmarks-2026/)
• Belkins — Michael Maximoff, "LinkedIn B2B outreach statistics: Belkins' 2026 study," updated June 29, 2026: Expandi 2025 data (15.1 million contacts), own campaigns (14,077 contacts), connection notes, reply by role, industry, and country (https://belkins.io/blog/linkedin-outreach-study)
• HeyReach — Bojana Vojnović and Anam Hassan, "LinkedIn Outreach Benchmarks 2026: 96,051 Campaigns Analyzed," September 17, 2026: mean and median acceptance and reply rates (https://www.heyreach.io/blog/linkedin-outreach-benchmark-report-2026)

See also

• Best AI SDR Platform for B2B SaaS 2026: 17 Tools
• LinkedIn prospecting benchmarks for Brazil: what's known and how to build your own
• LinkedIn B2B sales cycle by deal size
• LinkedIn connection acceptance rate: when to add a note
• Which LinkedIn cadence touchpoint gets the most replies
• LinkedIn reply rate: fixes that work
• LinkedIn SSI and reply rates