Scaling LinkedIn Outreach in 2026

One LinkedIn account has a hard ceiling. A warmed one sends 15–20 invitations on a sending day and 30–40 on Sales Navigator, and no setting you can reach moves it. Getting past that means running several at once, which is a different job from running one well, and this page is about the difference.

What actually caps one account

The ceiling is per account and it is not configurable. Free and Premium accounts land near 100 invitations a week, Sales Navigator near 150–200, both on a rolling seven-day window. Sent across about five sending days, that is the 15–20 and 30–40 a day in the lead. Run those daily rates seven days a week and you are at 105–140 against a 100-a-week cap, which is why the five-day assumption travels with every daily number on this page. A brand-new account starts near one invitation a day and earns its way to seven or eight within a month, the ramp the warmup engine computes. An aged account that went quiet restarts higher, around 5–10 a day. The full table, including messaging ceilings and the 30,000 cap on total connections, lives on LinkedIn connection limits.

LinkedIn publishes no numeric invitation cap, so those figures are community-observed and corroborated by vendor reporting. They have been stable long enough to plan against, and no plan, proxy or pacing trick gets one account to 500 invitations a week.

What settings do change is whether an account holds the top of its own band. Acceptance rate, invitation age and how mechanical the timing looks decide whether it sits at 20 a day or gets pushed back toward 5. You don’t bypass LinkedIn limits – you earn capacity, and the accounts that send the most are the ones that never hit a limit. That protection runs out at one account’s ceiling.

The four routes to more volume

Each of these is a real option, and only two of them multiply across accounts.

The first two are covered in depth on LinkedIn connection limits and safe LinkedIn automation, and both are worth doing before you add anything. The rest of this page is about the two that multiply.

More accounts means more people

Before the arithmetic, the constraint that decides whether this is worth doing at all. LinkedIn allows one profile per person, which is the same rule that makes a replacement account a second violation for someone already restricted. Scaling this way is your team’s accounts, or your clients’, each belonging to the person whose name is on it. Profiles invented for people who do not exist are the pattern LinkedIn is best at catching, and they take the real accounts down with them when they go.

The part that gets underestimated is the logistics. Every account belongs to somebody who has to sign in, stay signed in, and sign in again when LinkedIn kills the session or they change their password on a Friday. Holding a client’s credentials is a liability, and a shared login is also the fastest way to produce the concurrent-session pattern in the next section.

With that settled, the arithmetic is dull, which is the point. Five warmed free or Premium accounts send 75–100 invitations on a sending day between them, which is 375–500 across a five-day week. Five on Sales Navigator send 150–200 a day, or 750–1,000 a week. Those are ceilings, not forecasts, and the section below is mostly about the gap between the two.

If you do not have that many people

Most teams cannot produce ten colleagues willing to hand over their LinkedIn profile, which is why a category of services exists to rent you warmed accounts. A dozen vendors sell this on broadly similar terms, and two publish theirs in full.

• MirrorProfiles list EUR 100 a month for a European account and USD 150 for a US one. Accounts arrive with three months of manual activity and 500+ connections behind them, warmed in Europe or North America, with a replacement inside 24 hours if one goes down. Disclosure: MirrorProfiles resell the stack this article describes, so read that as a worked example and not as a recommendation.
• Aimfox sell rented profiles as “Outreach Avatar” at $129 a month per profile, billed quarterly, with an outreach seat included and a replacement guarantee. Their agency plan is a separate product: $499 for 20 seats you connect yourself, with the profiles still yours to supply.

Read the arrangement for what it is. The profile belongs to the provider and you operate it, so it differs from the invented profiles above in one way: the history is real. Whether the person named on it is a real, consenting human is what the public terms do not answer, and it is the question to put to a provider directly, since it is what makes an invented profile dangerous. MirrorProfiles’ own site states that it “operates independently from LinkedIn and is neither affiliated with nor endorsed by LinkedIn”.

A rented account also gets restricted exactly like your own, and you deal with the provider when it does, so the identity-verification path on LinkedIn account restricted is not yours to walk. Everything in the next section still applies to it. Renting solves the supply of profiles, and leaves the operating problem where it was.

Why fleets break where single accounts do not

A fleet is not ten copies of a working single-account setup. Four things behave differently once there is more than one account, and each is a way to lose several accounts in the same week.

• Shared fingerprints and shared IPs. The isolation that protects one account is what a fleet gets wrong first. Ten accounts behind one browser profile or one IP pool stop looking like ten people and start looking like one coordinated group, and they tend to be restricted together. GTM API’s own warm-up model charges for it, holding down any account it sees running across three or more sessions and IPs.
• Identical copy and identical timing. Ten accounts sending the same opening line on the same schedule are legible as one operation at the content layer, where perfect browser isolation does nothing for you. Vary the copy per sender and let each account keep its own rhythm.
• Overlapping target lists. Two of your accounts inviting the same person in one week costs both of them the acceptance rate that keeps their ceilings up, and for an agency it is worse than a metric problem: two clients’ campaigns landing on one buyer is a conversation you do not want to have. Deduplication belongs in place before the second account sends.
• Health read in aggregate. One restricted account in ten is a signal about that account. Three in ten inside a month is a signal about how the fleet is being run, and the difference only shows if accounts are measured one by one. The harder case is the account that quietly degrades without ever being restricted: ten limit hits and two blocks in a month is enough for GTM API’s warm-up to hold one near five or six invitations a day, which a fleet total hides.

From one account to a fleet, staged

A workable sequence, assuming the first account is already running and healthy.

1. Get one account to its ceiling first. If one account is not holding 15–20 invitations on a sending day at a healthy acceptance rate, adding accounts multiplies a broken process.
2. Sort out logins before volume. Decide how each person signs in, what happens on a second factor, and who re-authenticates when a session dies. A process that depends on somebody’s password in a password manager breaks in week three.
3. Add the second account, and build deduplication before it sends. A suppression list per client, checked at send time, plus a way to spot one prospect appearing under two clients without putting either list in front of the other. Teams retrofit this, and by then the acceptance rate has paid for it.
4. Give each account its own session and proxy from day one. One browser profile per account, one residential IP fitting the geography of the person named on the profile, and the pair stays together.
5. Onboard in waves of two or three. Staggering does not make the fleet mature faster. It is that onboarding is where accounts get killed, by a rushed proxy assignment or the same opening line pasted across four senders, and two or three a week is what a small team can set up carefully.
6. Instrument per-account health before you need it. Warm-up state, limit hits, blocks, acceptance rate and session count, per account, on a schedule, so you catch drift before it hits something.

Around 20 accounts the manual version stops working, because somebody is now maintaining a spreadsheet of which account is warm, which proxy belongs to whom and who has already been contacted, and that spreadsheet becomes the thing that breaks.

What a platform does with that

This is the part GTM API is built for, across what it reports as 20,000+ LinkedIn accounts running at under a 1% monthly ban rate. It covers the isolation, the pacing and the health reading. Target deduplication is not in it, so that one stays on your side of the line whatever you run this on.

• Per-action limits, written per account, automatically. Sixteen buckets are created the moment an account connects, each with its own daily cap and spacing, so a heavy engagement day cannot spend the invitation allowance. Every interval in the resulting schedule is drawn at random, because a fixed tick is itself a detectable signature.
• A warm-up allowance computed from the account’s own history. Six drivers, multiplied: account age, limit hits and blocks in the last 30 days as separate terms, activity level, session and IP spread, and profile completeness. A brand-new account is allowed roughly one invitation a day, and a month of moderate activity takes it to seven or eight. The allowance ramps in steps and snaps down when a driver worsens.
• The owner signs in, once. An account joins by its owner following a one-time link and signing in inside the cloud browser, so you never hold the password or handle the second factor. That is what makes a client’s account workable at all.
• One operational browser per profile, enforced platform-wide. A second operational browser for the same profile is rejected outright, so two operators cannot drive one account into a concurrent-session flag. Each browser owns its proxy, pickable by country.
• Fleet state you can be told about and can query. limit-reached fires when an account exhausts a bucket and limit-released when the clock frees it, so a stop arrives as an event, while snapshots every 22 hours carry the warm-up state for a dashboard to poll. Both read over the same REST and MCP contract.

Keep two clocks apart. The platform’s buckets refill on their own schedule, invitations about an hour after a quota hit and InMail nearer three, with daily counters clearing at each account’s local midnight. LinkedIn’s weekly ceiling is none of those: it releases as your oldest invitations age out of the rolling seven-day window, and nothing speeds that up.

None of this raises what LinkedIn allows one account to do. It keeps each account inside its own ceiling while you add more of them, which is the only kind of scaling that survives a quarter.

What the alternatives do well

Operator tools solve a real part of this. HeyReach sells multi-seat sender rotation on a flat fee, $999 a month for 25 senders or $2,999 for unlimited, and for a team that wants campaigns running this week without building anything, that is the shorter path. Aimfox bundles rented profiles into the same product. Both are closed UIs, which buys you less to build and leaves you less to change. An API layer earns its keep when your own product or agent goes in front of it, and its honest cost is the sequencer, the inbox and the reporting you now own.

FAQ

Sources

• MirrorProfiles, pricing and terms (read 7 August 2026). Backs the EUR 100 and USD 150 figures, the three months of activity and 500+ connections, the 24-hour replacement, and the quoted affiliation statement.
• Aimfox, pricing (read 7 August 2026). Backs the $129 per month Outreach Avatar figure and the separate $499 for 20 seats agency plan.
• HeyReach, pricing (read 7 August 2026). Backs the $999 for 25 senders and $2,999 for unlimited figures.
• PhantomBuster’s “LinkedIn connection request limits” write-up (on their blog). Corroboration for the weekly and daily figures. LinkedIn publishes no numeric invitation cap, so vendor reporting here is corroboration and not authority.
• Related reading on this site: LinkedIn limits, LinkedIn connection limits, LinkedIn account restricted, safe LinkedIn automation.

Michael SalamatovGTM engineerRuns the GTM API blog. Covers LinkedIn limits, account health and the API side of outbound.in

Last updated: August 2026 · Per-account caps are community-observed and shift over time