Should You Connect First or Message First on LinkedIn?

By Daniel Okoro, Outreach Tactics. Last updated: 2026-05-29

---

A few things that actually come up when reps decide their first move on LinkedIn:

• They fire off InMails to every cold prospect because it feels more "direct," burn through credits fast, and see single-digit reply rates.
• They send connection requests without a strategy for what happens after the accept, then wonder why the follow-up stalls.
• They're already connected to someone who engaged with their content and still wait to see if that person will "reach out first."

The decision isn't about style. It is a cost-and-conversion question. The connection request is free and starts a relationship. The InMail costs a credit and skips the relationship. Picking the wrong one for the situation wastes either money or the warm path to a reply.

---

What does "connect first" vs "message first" actually mean on LinkedIn?

LinkedIn gives you three distinct first moves when approaching a prospect:

1. Connection request (free, with or without a note). You send a request; if accepted, you can message freely. This is "connect first."
2. InMail (paid credit, no connection required). You message a non-connection directly, bypassing the request step. This is "message first" for cold strangers.
3. Direct message to an existing connection. You message someone you already connected with. This is also "message first," and it is the correct default when the relationship already exists.

The choice sets the entire downstream sequence. Connecting first builds a relationship before the ask: the prospect consciously accepted your request, which creates a small but real commitment before you pitch. Messaging first (via InMail) makes the ask before any relationship exists, with no intermediary commitment to warm the exchange.

Is it better to send a connection request with a note or blank?

The note-versus-no-note debate is its own sub-decision inside "connect first." The full breakdown of the 300-character mechanics lives in the LinkedIn connection request note post. The short version for this decision:

A specific, signal-based note can help with cold prospects who need context about why you are reaching out. A blank request sometimes outperforms a generic note that reads as a disguised pitch. The note is a lever worth tuning, but it is not the dominant variable.

Volume is. Reachium's data shows an average connection acceptance rate of 28% across 161,569 requests [PLATFORM]. Acceptance peaks at 34% for accounts sending 10-19 invites a day and falls to 30.6% at 20-29 invites a day [PLATFORM]. The data is clear: over-sending hurts more than a blank note. A mediocre note at the right volume outperforms a perfect note at 50 invites a day.

When should you use InMail instead of a connection request?

InMail wins in a specific set of situations:

• Closed or restricted network. Some senior prospects set their profile so only connections of connections can request. InMail is your only path.
• Out of connection-request capacity. LinkedIn's weekly invitation limit sits around 100 for most accounts (up to 200 for accounts with a high Social Selling Index), confirmed via LinkedIn's Help Center. When you hit the cap mid-week, InMail covers the gap.
• Senior decision-maker where intent matters. At C-suite level, the deliberateness of an InMail can signal seriousness in a way a mass-connect campaign does not.

For most cold B2B prospecting outside those cases, the connection request is the higher-ROI first move. It is free, starts a relationship, and the follow-up message arrives in the recipient's primary inbox rather than the "InMail" tab. The full reply-rate head-to-head between InMail and connection requests is in LinkedIn InMail vs connection request 2026.

Does messaging before connecting convert worse?

For cold prospects, yes in most cases. Asking before any relationship exists raises defensiveness. The connection request earns a small commitment (the accept) before the ask, which lowers resistance to the follow-up message. It is the same psychology as a handshake before a pitch.

The exception is real: existing connections and warm prospects who already engaged with your content. For them, messaging first is correct because the relationship already exists. There is no reason to re-request a connection that is already there, or to make a warm prospect go through an acceptance step they did not expect.

The volume risk works in the same direction. Sending InMails at scale costs credits per message. Sending connection requests at scale is free up to the weekly cap. A rep who defaults to InMail for all cold prospecting is spending budget on a channel that, for most B2B use cases, converts below a well-crafted connection request sequence.

Which first-contact method gets the most replies?

The decision comes down to your relationship to the prospect at the moment you reach out:

Prospect situation	Recommended first move	Why
Cold stranger	Connection request	Free, earns the accept before the ask, opens a direct message channel
Engaged with your content (liked, commented)	Direct message	Relationship exists; skip the request step
Existing connection (not recently active)	Direct message	Already connected; no need to re-request
Senior / C-suite prospect	InMail or connection request	InMail signals deliberate intent; connection request works if the profile is open
Closed network (can't request)	InMail	Only available channel
Out of weekly invite capacity	InMail	Covers the gap until capacity resets

The reply rate that matters is downstream: in Reachium's data, 29% of accepted connections go on to reply [PLATFORM]. That means every accepted connection is a future DM channel. The connection-first path compounds because each accept is a low-cost, permanent warm channel to that prospect, even if the first follow-up does not convert. See the full funnel math in the LinkedIn outreach benchmarks 2026.

For a deeper look at researching prospects before you choose your approach, see how to research a prospect on LinkedIn fast, which covers the signals that tell you which method to use before you click send.

How do you run the connection-first approach at scale safely?

Since the connection request is the default first move for most reps, the operational question shifts to volume safety. More requests is not more replies. Reachium's data shows the volume tax clearly: acceptance peaks at 34% for accounts at 10-19 invites per day and falls as volume climbs [PLATFORM]. The safe ceiling the platform calibrates to is roughly 25 invites per day. The stop sending 100 connection requests per day post covers the mechanics and the math behind that ceiling.

The system answer is to run the connection-first sequence at safe daily volume on the verified API, with the follow-up scheduled automatically after acceptance rather than manually tracked in a spreadsheet. That is the operational piece most reps skip: the accept fires, and nothing happens for four days because the rep forgot to check.

FAQ

Can you message someone on LinkedIn without connecting?

Yes, through InMail, which is LinkedIn's paid messaging feature that lets you reach non-connections directly. InMail requires a Premium subscription and consumes a credit per send. Some profiles set to "Open Profile" allow InMail without credits. For free accounts without Premium, the only option for cold strangers is the connection request.

How many connection requests can you send before LinkedIn limits you?

LinkedIn caps weekly invitations dynamically based on account signals. Most accounts land around 100 requests per week; accounts with a high Social Selling Index (70+) can reach up to 200. New accounts (under three months old) are typically capped closer to 50-80. LinkedIn's Help Center notes that hitting the cap results in a one-week cooldown before capacity resets. Working inside roughly 25 invites per day keeps accounts well within safe territory for most established accounts.

Does an InMail count against your connection request limit?

No. InMail and connection requests are separate quotas. InMail draws from your monthly InMail credit allocation (typically 20-50 per month depending on your Premium plan), not from the weekly connection-request limit. This is one reason InMail is the backup when you hit your weekly invite cap, not an alternative to the connection request for normal volume.

Should you wait after a prospect accepts before sending your follow-up?

A short gap of 24-48 hours is generally better than sending the follow-up within minutes of acceptance. Instant follow-ups read as automated (because they usually are) and break the rapport the accept created. A brief delay makes the message feel like you noticed the acceptance and responded. The content of the follow-up matters more than the exact timing: reference the reason you connected and make a specific, low-friction ask.

Is it better to connect first or follow first on LinkedIn?

Following a prospect before connecting is a warm-up tactic, not a substitute for the connection request. A follow gives you visibility into their content and lets you engage with their posts before reaching out, which can increase acceptance rates on the subsequent request. The follow-first approach works best for senior prospects or accounts where you want to build familiarity before the ask. For most cold outreach at volume, following first adds friction without enough payoff; the connection request remains the faster path.

Sources

• Reachium - Platform data: 28% acceptance rate, 29% reply of accepted, across 161,569 connection requests and 316,703 sequences.
• Linked Insider: LinkedIn outreach benchmarks 2026 - Full funnel benchmarks including acceptance, reply, and meeting rates.
• LinkedIn Help Center: Invitation limits - Official documentation on invitation caps and the weekly reset mechanic.
• Linked Insider: LinkedIn InMail vs connection request 2026 - Head-to-head reply-rate comparison.
• Linked Insider: LinkedIn connection request note - Note vs no-note mechanics and the 300-character breakdown.
• Expandi: LinkedIn outreach benchmarks 2026 - acceptance and post-connection reply benchmarks across 13.2M data points.