LinkedIn InMail vs connection request — which converts better?

A connection request and an InMail are not two versions of the same move. One spends a credit from an allowance LinkedIn publishes, with a documented refund rule that makes many of them free. The other spends account standing that has no price list, no balance you can read, and no refund at all. Asking which converts better compares a metered resource against an unmetered one. The useful question is which budget you can afford to spend on this particular person, and the answer changes by how reachable they are.

Two different currencies, and only one of them has a published price list

Start with what each move actually consumes, because that is where the comparison usually goes wrong.

An InMail is a message to a member you are not connected to. It draws on a credit balance tied to your subscription. LinkedIn publishes the allotments, the accumulation caps and the refund rule, so you can compute the cost of a send before you make it.

A connection request consumes no credit. What it consumes is the account's standing with LinkedIn's invitation system, and LinkedIn describes that system in behavioural terms rather than numeric ones. Its page on types of restrictions for sending invitations names three triggers: that you have sent many invitations within a short amount of time, that many of your invitations have been ignored, left pending or marked as spam by the recipients, and that an excessive number of invitations may lead LinkedIn to suspect the use of an automation tool and suspend or restrict the account.

Read the middle trigger again. Two of the three named triggers are about how recipients reacted, not about how many you sent. That makes an invitation a bet against a balance you cannot query, settled by strangers, with no statement at the end of the month.

What a credit actually costs, and the rule that makes some InMails free

LinkedIn's help page on InMail message credits and the renewal process gives the monthly allotments directly: 5 for Premium Career, 15 for Premium Business, 50 for Sales Navigator Core, 30 for Recruiter Lite. It also gives accumulation caps — 15, 45, 150 and 120 respectively, plus 5 per job for Hiring Pro — so credits build up to a ceiling and stop.

Then the rule that changes the arithmetic entirely: *"Every InMail message that is accepted/declined or responded to directly within 90 days of it being sent is credited back."*

Sit with that for a second. A reply refunds the credit. A decline refunds the credit. Only silence costs you anything. So the effective price of an InMail is not a fixed unit — it is a unit multiplied by your ignore rate. Send fifty InMails on a Sales Navigator Core seat, get twenty responses of any kind inside ninety days, and you have spent thirty credits, not fifty.

This has a strategic consequence most InMail advice misses. The expensive InMail is the one nobody bothers to decline. Writing something a disinterested person will answer with "not for us" is cheaper than writing something they scroll past, and it is also the more honest message.

Open Profile: the third path, where the message costs nothing at all

There is a route that is neither an invitation nor a paid InMail, and almost nobody budgets for it.

LinkedIn's Open Profile feature lets a Premium subscriber accept messages from members who are not in their network, without an InMail credit being spent. LinkedIn's own page on sending an Open Profile message states that the Open Profile Premium feature allows members to contact a Premium member for free. Two constraints come with it: LinkedIn says it limits the number of Open Profile messages you can send in a given time period, and its InMail page states that you cannot message a member who has chosen not to receive InMail messages in their message preferences settings.

The practical effect is that a slice of your target list is reachable at zero credit cost and zero invitation risk, and you cannot tell which slice by looking at a list of names. It has to be checked per profile. That check is exactly the kind of thing worth automating, and it is the reason the routing section below exists.

What an invitation spends, and why no invoice ever arrives

An invitation's cost is deferred and diffuse, which is why teams underprice it.

LinkedIn's restriction page says most restrictions are automatically removed within one week, and gives graduated guidance: wait a few hours after a first restriction, a few days if you are restricted repeatedly, and up to one month if the restriction is for having too many outstanding invitations. It also notes that after you withdraw an invitation you cannot resend one to the same recipient for up to three weeks — so mass-withdrawing a stale backlog silently removes those people from your reachable list for most of a month. The mechanics of writing an invitation people accept are a separate craft, covered in the invitation notes that actually get accepted.

There is one hard number on that page, and it is the ceiling rather than the rate: 30,000 first-degree connections is the maximum network size for a LinkedIn member. Everything else is described, not quantified.

The honest summary is that an invitation is cheap in cash, expensive in optionality, and priced by people who never see your budget.

The message budgets are not comparable either: 200 characters against 2,000

Format is the difference people notice last and feel first.

LinkedIn's page on sending an InMail message states that InMail messages can have up to 200 characters in the subject line and up to 2,000 characters in the body, with the subject optional. A personalised invitation note gets 200 characters total — LinkedIn's own help pages disagree on how many free-account members get per month and on whether Premium raises the character budget, which is a documentation inconsistency worth knowing about rather than a fact worth quoting.

Two hundred characters is roughly two sentences. Two thousand is roughly four paragraphs. These are not the same writing problem:

• At 200 characters you cannot make a case. You can make a reason. The note has to justify the connection, not the meeting, and any sentence that would also work as an email opener is wasting a tenth of your budget.
• At 2,000 characters you can make a case, and the temptation to use all of it is the main way InMail gets ignored. The subject line is doing the work of an email subject line, and the body is competing with a real inbox.

The InMail also arrives with a visible marker that it is an InMail. The recipient knows a credit was spent. That is worth something and it costs something, and which one dominates depends entirely on whether the message reads as researched or as scaled.

When an InMail is the right instrument, and when it is a purchase of impatience

Here is the decision rule that survives contact with a real list.

Reach for an InMail when the person is structurally hard to reach. No mutual connections, a locked-down profile, a title senior enough that invitations from unknown names are declined as policy, or a recipient in a market where your company has no recognition. In that situation, an invitation is likely to sit pending, and a pending invitation is precisely the thing LinkedIn's restriction page names as a trigger.

Reach for a connection request when acceptance is plausible. Shared groups, shared employer history, a mutual connection, an obviously adjacent role, or a profile that has interacted with your company. Acceptance opens an ongoing channel with no further credit cost, which is the compounding asset in this comparison.

Check for Open Profile before doing either, because when it is available it dominates both.

And the anti-pattern, stated plainly: sending an InMail because your invitations were restricted this week is buying your way past a signal rather than acting on it. The restriction was information about your targeting. Spending credits to keep the same targeting running converts a copy problem into a billing problem.

Why "which converts better" cannot be answered from public data

This section exists because the question in the title deserves an honest refusal.

There is no authoritative public dataset comparing InMail response rates against connection-request acceptance rates. LinkedIn does not publish either. Every figure in circulation comes from a tool vendor measuring its own customers, with an undisclosed sampling frame, and the two metrics are not even the same event — an acceptance is not a reply, and a reply is not a meeting.

The measurements also run on different denominators. An invitation's outcome is observable to you only as accepted, pending or withdrawn. An InMail's outcome is observable as replied or silent, and LinkedIn's own refund mechanism means a decline registers as engagement. Any single blended "conversion rate" across both is an average of things that are not alike.

What you can measure on your own account, per segment:

• Acceptance rate on invitations, split by whether a note was attached, tracked weekly rather than per campaign.
• Response rate on InMails, counting declines separately from replies, because they cost differently.
• Credits actually consumed, which is sends minus refunds and is the only real cost figure.
• Conversations reached per week per account, which is the number that constrains a pipeline regardless of instrument.
• Pending invitations outstanding, since that is the balance LinkedIn's own page ties to the longest restriction.

Segment those by seniority and by whether you have any prior relationship. The gap between segments will be larger than the gap between instruments.

How Autocloz routes between the two, including the diversion you did not configure

Autocloz runs LinkedIn as sequence steps rather than as a separate tool, with distinct step kinds for linkedin_invite, linkedin_dm and linkedin_inmail, alongside view, like, follow, endorse and withdraw actions. Three behaviours are worth knowing before you build a sequence, because two of them will change what your steps do.

Accepted invitations and replies on either instrument land in the same inbound queue as email, SMS and WhatsApp, which matters more than it sounds: the two instruments produce one conversation stream, so a rep answering an InMail reply does not have to remember which move produced it.

An InMail step without a subject fails loudly. A linkedin_inmail step with no subject_tpl returns a failed dispatch with the message "LinkedIn InMail requires a subject_tpl" rather than sending a subject-less message. That is deliberate — a silent send would be worse than a visible refusal.

A DM step will not fire before acceptance. A linkedin_dm step checks for a prior invite touch on that lead in an accepted or replied state, and skips with "DM steps require a prior accepted linkedin_invite for this lead" when it finds none. You cannot accidentally build a sequence that messages people who never connected.

And the one that surprises people: an invite step can go out as an InMail. When a step is a linkedin_invite, a real provider is connected, and the rendered note is at least 60 characters, Autocloz fetches the target's profile first and — if it is an Open Profile — sends the note as an InMail instead, deriving a subject from the first sentence capped at 80 characters. The 60-character floor exists so a one-line "let's connect" is never promoted into an InMail, where it would read as spam. This bypasses the invite cap for that lead, which is the point, but it means the action recorded is not always the action you configured.

Around all of that sits the pacing layer, and it is worth being precise about what those numbers are. Autocloz defaults to a weekly invite ceiling of 80, a first-day floor of 5 actions growing on a smoothstep curve to the daily target over 21 days, weekend allowance dampened to 30% of the weekday figure, a burst refusal when five actions land inside ten minutes, a hard stop at 1,000 pending invites, an operator-run sweep for invitations still pending after 14 days — nothing schedules it — and a halving of the daily ceiling when the trailing accept rate falls below 30%. Every one of those is a product default chosen to sit well inside an undocumented system — none of them is a LinkedIn-published limit, because LinkedIn publishes none. There are also two deduplication gates most teams do not think to build: a 90-day refusal to touch a lead another sender in the same workspace already engaged, and a 30-day refusal at company level so five reps cannot converge on one account in one month.

Autocloz's free plan covers 5 users and 10 mailboxes with all five channels included and a LinkedIn daily cap of 50 actions — start free and run invitations, DMs and InMails as steps in the same sequence as your email. The mechanics of the LinkedIn channel itself, including how the accept-rate throttle behaves, are set out on the LinkedIn channel page.

What neither instrument buys, and what Autocloz does not do

Several things, stated plainly, because a comparison that concedes nothing is not worth trusting.

Neither instrument buys relevance. An InMail to the wrong person is a credit spent on being ignored by someone better qualified to ignore you, and a note attached to a badly targeted invitation raises the odds of the "I don't know this person" response that LinkedIn's restriction page names directly. If invitations are being ignored at scale, no amount of credit spending fixes the underlying list.

Neither is a substitute for email. The comparison of the two platforms as channels — where each one's ceiling actually sits, and what each can observe about you — is a different question, worked through in LinkedIn against email for B2B outreach, and the safety mechanics across LinkedIn, WhatsApp and SMS together are in running multiple channels without getting banned.

And Autocloz specifically. It cannot tell you whether a message was read; LinkedIn exposes no read receipt to an integration, so an unanswered InMail and an unopened one are indistinguishable. It does not sell LinkedIn accounts, credits or Sales Navigator seats — the subscription, and therefore the credit balance, stays yours and is bought from LinkedIn. It cannot see your credit balance either, so it will not stop you configuring more InMail steps than you have credits to run; the provider returns the refusal and the step records a failure. The Open Profile diversion depends on a live provider session, so in a dry-run or stub configuration it does not happen and the step behaves as a plain invite. LinkedIn's User Agreement prohibits using bots or other unauthorised automated methods to access the Services, add or download contacts, or send or redirect messages, at clause 8.2.13, and prohibits scraping at 8.2.2. Any tool that acts on your behalf operates against that language however carefully it paces, so nobody — including us — can tell you the enforcement risk is zero. What careful pacing buys is a lower probability, not immunity. If that trade is not one you want to make, the honest answer is to run LinkedIn manually and put the automation budget into the channels where the platform actually invites integration.