← Blog

Sell ·

LinkedIn Limits, Explained: What Is Published, What Is Folklore, and What Gets Accounts Restricted

LinkedIn publishes almost no numbers. What its rules actually say, which caps are community folklore, and what it says gets accounts restricted.

By , Founder, Baljia AI

The short answer

LinkedIn publishes rules, not numbers. The User Agreement bans bots and other unauthorized automated methods, invitations, messages, likes and comments included, and bans scraping. The help pages confirm that invitation limits exist, and state no weekly figure anywhere. The numbers you see everywhere else, 200 invites a week, 50 messages a day, are community folklore: unofficial observations LinkedIn has never confirmed, which can change without notice. And the things LinkedIn says will get an account restricted make a short, specific list: automated activity, scraping, hitting the invitation limit, pitching strangers through connection requests, engagement pods, and comments posted by tools.

The honest playbook follows in three lines. Stay under every limit LinkedIn does publish. Slow down at the first warning instead of pushing through it. And do manually whatever a tool offers to automate through your account.

We make Baljia, which drafts LinkedIn and email outreach and, by default, sends nothing without a founder's yes, with LinkedIn campaigns on the Growth and Scale plans. We publish this page because our own product documentation leans on the same line we draw here, and you should read LinkedIn's actual rules rather than any vendor's retelling of them, including ours. Every claim below comes from LinkedIn's own pages, all listed in the sources.

Bucket one: what LinkedIn actually publishes

The User Agreement: the banned methods

Section 8.2 of the User Agreement, effective November 3, 2025, bans using "bots or other unauthorized automated methods" across the actions outreach tools sell: accessing the service, adding contacts, sending messages, and creating, commenting on, liking or sharing posts. It separately bans scraping software and bypassing access controls or use limits. What that list means for a working weekly prospecting system is walked through in our guide to getting customers from LinkedIn; for this page, one observation matters most. The section contains no number. The method is what it bans, not the pace.

The Professional Community Policies: what invitations are for

The policies tell members not to use invitations to send promotional messages to people they do not know. They also tell members not to agree ahead of time to like or reshare each other's content, which is the working definition of an engagement pod, and they ban lotteries, contests, sweepstakes and giveaways outright.

The March 13, 2026 newsroom post

LinkedIn's newsroom said it actively limits the reach of engagement pods and automated comments, defined an automated comment as one posted by a tool rather than a person clicking the button, and warned that accounts can face restrictions. It is the most recent, most direct statement LinkedIn has made about automation and enforcement, and almost nobody selling "LinkedIn limits" advice links to it.

The help pages: the few numbers that exist

Four concrete things live on LinkedIn's help pages:

  • Invitation limits exist for every account. Cross one and inviting is typically off for a week, and pulling back the invitations you already sent will not shorten that.
  • Free accounts get a small monthly allowance of personalized invitations, the ones with a note attached. One help page says three per month; another says five. Plan for three, and notice the deeper lesson: LinkedIn's own pages disagree with each other, because these numbers move without announcement.
  • A personalized invitation note is capped at 200 characters. One of the few hard numbers LinkedIn does publish, stated on its invitation help pages.
  • Scheduling posts is supported, not suspicious. LinkedIn ships its own post scheduler, from 10 minutes to three months ahead.

The spam help page adds one more rule without a number: content designed to artificially increase engagement, including chain-letter style requests for likes and shares, can be limited or removed.

That is close to the complete set of published specifics. Everything else you have read belongs in the next bucket.

Bucket two: the folklore numbers

The number you readWhere it comes fromWhat LinkedIn publishes
"200 connection requests a week"Automation vendors, sales blogs, forum threadsNothing. Limits exist; no weekly figure is stated anywhere.
"50 messages a day"The same sourcesNothing.
"20 to 25 invites a day to stay safe"Pacing guidesNothing. "Safe" is not a published category.
Sales Navigator caps (results per search, saved leads)Tool vendors' own documentationWe have not verified these against a LinkedIn page, so we will not repeat them as policy.

Three things to understand about folklore numbers, stated without drama:

They are observations, not permissions. Somebody's account sent some volume without a restriction, they wrote it down, and ten blogs rounded it into a rule. That is real information of a weak kind: it tells you what did not trigger enforcement for one account, somewhere, at some point.

They change without notice. An unpublished limit has no changelog. The communities that track these figures revise them every year or two, so a tool or a habit tuned to last year's folklore is tuned to nothing. LinkedIn's own help pages disagreeing about three versus five personalized invites is the published version of the same drift.

They answer the wrong question. What section 8.2 bans is the unauthorized automated method itself, not a particular daily volume, so no folklore cap can make a banned method permitted.

Bucket three: what LinkedIn says triggers restrictions

Pulled together from the same pages, the published triggers are:

  1. Bots and other unauthorized automated methods acting through an account: adding contacts, messaging, liking, commenting, sharing (User Agreement 8.2).
  2. Scraping, which the agreement bans at the software level, whoever presses the button.
  3. Hitting the invitation limit, which typically brings a one-week invite restriction (help pages).
  4. Using invitations to pitch strangers (Professional Community Policies).
  5. Engagement pods and automated comments, which get reach limited and can bring account restrictions (the March 13, 2026 newsroom post).
  6. Content built to artificially inflate engagement, which can be limited or removed (the spam help page).

Note what this list is not: mysterious. Every trigger on it is an action you can simply not take, which is why this page needs no scare stories and no survival kit. The plain list is enough to plan around.

The honest playbook

Stay under the limits LinkedIn does publish. Three personalized invitations a month on a free account is a real number; budget it, and keep every other request bare. Beyond that, send at the pace of a person actually working, because that is what you are.

Treat the first warning as the answer. If LinkedIn restricts your invitations, the week off is the message. Withdrawing pending invites does not lift it, and returning at the same volume asks the same question again. Use the week to reread your targeting: a list where most people accept needs fewer invitations in the first place.

Do manually what the folklore tools automate. The arithmetic is on your side. A founder working a hand-picked list of 20 to 40 people, the system in our guide to getting customers from LinkedIn, sends fewer invitations in a month than the folklore says one account can send in a week. Keep the list small enough to be honest about, and the limits question mostly dissolves.

Let software stop at the draft. Research, drafting, and scheduling through official routes make up the lane LinkedIn itself supports; its own post scheduler is proof the lane exists. Software that sends invites, messages or likes through your account sits in the territory section 8.2 describes: it bans "bots or other unauthorized automated methods", and a pacing dial does not grant permission. The wider version of that argument, covering email law and your own reputation, is in Is it safe to let AI send messages in your name?, and the content-side version lives in the founder brand guide.

Where Baljia fits

Baljia finds prospects and drafts LinkedIn and email outreach. By default nothing goes out without your yes: a request for a draft stays a draft, and switching sending on is an explicit choice you make. When you do switch it on, messages are delivered through the account you connected, and everything this page says about automated sending applies to tools that work that way, ours included. That is why drafts are the default, why volumes stay low, and why the lowest-risk way to use Baljia on LinkedIn is to keep messages as drafts and send them by hand. LinkedIn campaigns come with the Growth and Scale plans, and our own site's caution holds here: sending limits reduce volume; they do not guarantee account safety or platform permission, and LinkedIn can restrict accounts that use automation, including accounts connected to Baljia. Costs, risks and limits are on the AI sales outreach page, and plans are on pricing.

Frequently asked questions

How many connection requests can I send per week on LinkedIn?

LinkedIn does not say. Limits exist on every account, and no weekly figure appears on any LinkedIn page. Every specific number you have read is a community observation, not a rule, and it can change without notice.

What happens when I hit the LinkedIn invitation limit?

Inviting is typically shut off for about a week, and pulling back the invitations you already sent will not shorten that. The useful response is to wait it out, then return with a smaller, better-chosen list rather than the same volume aimed at the same silence.

Is it against LinkedIn's rules to use AI to write my messages?

Writing is not the banned act; automated sending is. Section 8.2 bans unauthorized automated methods for adding contacts and sending messages, among other actions. A message an AI drafted and a human read, edited and sent with their own click is that human's message. A message a tool pushed through your account is automation, whoever wrote the words.

If everyone follows the folklore numbers, are they not basically safe?

Staying under a folklore number is not permission. The User Agreement bans unauthorized automated methods at any volume, so the folklore only describes what some accounts got away with at some point, under limits that are unpublished and can move underneath you. A person sending each message by hand rarely gets near any of it.

Does scheduling LinkedIn posts count as automation?

Not the kind the rules target. LinkedIn ships its own scheduler for posts, from 10 minutes to three months ahead, so scheduling is supported behavior. What the rules target is unauthorized automation: invitations, messages, likes and comments sent by a tool through your account, and scraping.

Keep the sources, skip the folklore

LinkedIn's numbers move quietly, so bookmark the pages below and reread them before you change how you prospect. If you want outreach that stays a draft until you say yes, Baljia offers a 7-day free trial, no credit card, at https://baljia.ai/. The trial does not include LinkedIn campaigns, which come with Growth and Scale, but it shows you how the draft-and-approve loop feels.

Sources