The 5-Touch LinkedIn Outreach Playbook

Five touches across 10 days. The first three are LinkedIn-native; the last two introduce a second channel (email) because anyone reaching touch 4 has either accepted but ignored, or never accepted at all, and they need a fresh surface to engage on.

The whole sequence is built around one principle: reply rate compounds on the back half of the sequence, not the first message. Most teams write a great touch 1 and then template touches 2-5 — that is where most of the lost pipeline is.

Touch 1 — Connection request (Day 0)

Goal: acceptance. Not reply. Acceptance.

Why: LinkedIn weights accepted-request rate heavily into account health. Trying to "sell" in the connection note drops acceptance from ~30% to ~10% and quietly poisons the account's reach for the next 30 days of requests.

Pattern: one specific, verifiable observation about the prospect's work or company that took at least 60 seconds to find. No pitch. No "loved your post" generic.

Template (do not paste, rewrite per prospect):

Hi {{first_name}}, saw your team just announced {{specific_thing}}.
{{One sentence on why that is interesting to you in a way that
matches your actual experience.}} Connecting.

Anti-patterns to avoid:

• "I'd love to learn more about what you do at {{company}}"
• "I help companies like yours with {{thing}}"
• Anything that mentions a meeting, demo, or 15 minutes

Measurement: acceptance rate. Healthy band 25-35%. Below 15% the account is being de-prioritised by the algorithm; pause for 7 days and revisit the note before re-sending volume.

Touch 2 — First DM after acceptance (Day +2)

Goal: a reply, any reply. Even "not interested" is a reply.

Why: the gap between accept and first message is where most teams waste the warmest signal in B2B outbound. A prospect who just accepted is paying you 30 seconds of attention. Burning that window with a generic "thanks for connecting!" is a fumble.

Pattern: ONE sentence with one specific question.

For a founder / VP:

{{first_name}} — quick question. {{One sentence naming the operational
problem you suspect they have, framed as a question they have an
opinion on.}}

For an SDR / individual contributor:

{{first_name}} — saw you run outbound at {{company}}. Are you {{one
specific aspect of their workflow you suspect is broken or unusual}}?

Anti-patterns:

• Multi-paragraph intro
• "I noticed you visited our profile" (creepy even when true)
• "Here is a 2-minute video on what we do" (we will never click)
• A Calendly link in touch 2

Measurement: reply rate to touch 2. Healthy band 8-15%. Below 5% the question is generic or the ICP segment is wrong; rewrite, do not just keep sending.

Touch 3 — Soft follow-up (Day +5)

Goal: the polite re-poke that respects you are not the only salesperson messaging them this week.

Why: 70% of B2B replies on LinkedIn arrive between touches 2 and 4. Skipping touch 3 because "they did not reply to touch 2" leaves the majority of the pipeline on the table.

Pattern: acknowledge the silence, do not apologise for it, leave one fresh hook.

{{first_name}}, I know you are probably swimming. One thing I forgot
to mention — {{a new piece of information that is genuinely valuable
even if they never reply: a stat, a case study one-liner, a public
benchmark}}. If it is interesting, happy to send the deeper version.

Anti-patterns:

• "Just bumping this to the top of your inbox"
• "Did you get a chance to look at my last message?"
• "Following up to make sure this did not get lost"

Touch 4 — Channel switch (Day +7)

Goal: move the conversation off LinkedIn before LinkedIn becomes fatigue territory.

Why: by touch 4 in their LinkedIn inbox you are competing with every other SDR who hit them this week. The same message in their email — which is far less crowded for senior B2B targets — gets read.

Pattern: find the work email (Hunter, Apollo, etc.) and send a 4-sentence email that is OBVIOUSLY a continuation of the LinkedIn conversation, not a fresh cold email.

Subject: Following up from LinkedIn — {{specific thing you mentioned}}

Hi {{first_name}},

We exchanged a hello on LinkedIn last week about {{specific topic}}.
You are probably stretched thin, so this can be a one-line reply.

{{The one question or value-add line — same as touch 3 but
recontextualised for email.}}

If now is not the time, totally fair — say the word and I will stop
clogging the inbox.

{{Your name}}

Touch 5 — Permission close (Day +10)

Goal: end the sequence with a self-respecting close, leave the door open, do not nag.

Why: the difference between a polished outbound machine and a slimy one is touch 5. The slimy version sends "FINAL CALL." The polished version says "I am stopping here, your choice."

Subject: Last note from BIGM

Hi {{first_name}},

This is the last time you will hear from me unless you ask me to keep
going.

{{One concrete observation that you would only have if you had really
thought about their company, not a templatable line.}}

If any of this is wrong or worth a conversation, the door is open.
If not, sincerely, no hard feelings — I will not message again.

{{Your name}}

This touch alone often lands 15-20% of sequence reply volume because B2B buyers reply to "I am stopping" more reliably than to "I keep following up."

Operational rules across all 5 touches

1. Volume per account: ≤ 20 NEW outbound connections per day per account. Above this LinkedIn flags accelerated.
2. Acceptance threshold: if account dips below 15% acceptance for a week, pause new requests.
3. Account pool: never run all your outbound from a single account.
4. IP rotation: automation through a single static IP is the single fastest path to a 30-day restriction.
5. Personalisation tier: touch 1 + touch 5 MUST be per-prospect. Touch 2-4 can use AI-rewritten per-prospect openers but not pure {{first_name}} templates.

What good looks like at the sequence level

Per-gap fix sheets

One short page per gap our diagnostic flags. Each one explains what the gap actually means, three things to do this week, and what "fixed" looks like.

Quick jump: volume · acceptance · reply rate · restricted · single account · IP rotation · templates · young account · spend vs results

Sending too many connection requests per day

LinkedIn's algorithmic ceiling sits at roughly 20-25/day per account before penalty kicks in, AND only when acceptance is under ~30% at that volume. Combine "high volume" with "low acceptance" and you enter a decay loop: requests get shown to fewer people, acceptance drops more, restrictions follow.

Fix this week:

1. Throttle daily new-request volume to 15 for 7 days, regardless of pipeline math. Acceptance recovers in week 2.
2. Audit touch 1. If it contains a meeting ask / "I help companies like yours" / generic "saw your post" — it is too sales-y. Use the touch 1 template above.
3. Split volume across accounts. Two accounts at 15/day each beats one account at 30/day on every dimension.

Connection acceptance rate is too low

Acceptance rate is the metric LinkedIn weights most heavily into your account's reach. Below 15% acceptance, future requests start getting shown to fewer people; the lower it goes, the less reach.

Fix this week:

1. Targeting first. If your audience filter is broader than "people with title X at companies of size Y in industry Z," acceptance is being killed by mis-targeting.
2. Touch 1 rewrite. Read your note out loud. If it pitches in the first sentence, rewrite. Reference one specific verifiable thing about the prospect.
3. A/B test ONE variable at a time. Send 30 with template A, 30 with template B (one specific change between them), measure over 5 days.

Cold reply rate is below baseline

Cold LinkedIn reply rate baseline is 5-12%. Below 5%, the sequence is leaving most of the pipeline on the table. The bottleneck is almost always touch 2 and 3, not touch 1.

Fix this week:

1. Audit gap between touch 1 and touch 2. If touch 2 starts with "thanks for connecting!" you wasted the warmest 30 seconds you will ever get.
2. Add touch 3 if it does not exist. 70% of B2B replies arrive between touches 2 and 4.
3. Move to email by touch 4. Same conversation, different channel. Use the touch 4 template above.

Account was restricted in the last 90 days

A LinkedIn restriction marks the account in LinkedIn's risk model. The interval between restrictions shrinks each time; the second is usually 7-30 days, third often permanent.

Do THIS WEEK:

1. Stop outbound from that account for ≥ 14 days. Every accepted request right now is one the algorithm is watching.
2. Provision a backup account NOW. 4-6 weeks of warmup is required; do it before the next restriction, not after.
3. Audit the trigger. Common causes: volume spike, acceptance crash, duplicate templates, automation fingerprint.

Pipeline depends on a single account

Most B2B sales teams that survive past $10k MRR via LinkedIn outbound learned the hard way: a single 30-day restriction on the one account = pipeline goes to zero overnight. Ban probability on a single account doing real volume is 15-30% over 12 months.

Fix in 30 days:

1. Provision a second account NOW. 4-6 weeks to warm; zero cost to do when you do not need it, one month of zero pipeline to do when you do.
2. Split volume the day account #2 is warm. Same total outbound, half per-account visible volume.
3. Plan for 4 accounts inside 90 days. Pool aggregate matches one account at full speed; restriction is invisible to pipeline.

No IP rotation on automated outreach

LinkedIn tracks IP and browser fingerprint of every action. Real humans log in from 1-3 places; automated outreach from a static cloud IP looks nothing like a real human. When acceptance is high and volume is reasonable, LinkedIn often ignores this; once acceptance drops or volume spikes, fingerprint is what tips it from "grey zone" to "bot."

Fix in 14 days:

1. Get residential IPs (Bright Data, Oxylabs, IPRoyal). $20-100/mo per account.
2. If running manually, stop flipping between 3+ networks. Stability is good.
3. Disable browser automation extensions that touch LinkedIn. Cheap ones are easy to fingerprint.

Using pure templates on outbound messages

LinkedIn's bot classifier got dramatically better at "this exact string was sent to 200 people last week" in 2024-25. Templated outbound correlates with both lower acceptance AND faster account flags.

Fix this week:

1. AI-rewrite touch 1 per profile. Use a per-prospect AI rewrite to generate the connection note from the profile's recent posts + role + company.
2. NEVER reuse touch 5 verbatim. The "I am stopping" line is the highest-conversion artefact; do not fingerprint it.
3. Audit duplicate strings. Count phrases like "I help" or "Hope you are well" — these are the canonical bot tells.

Outbound account is too young

LinkedIn weights account age heavily. Under 6 months gets throttled at thresholds that would not register on a 4-year-old account. Real outbound from a fresh account is fighting the algorithm.

Do in 60 days:

1. Slow-warm 60 days. Days 1-30: 3-5 requests/day to existing graph. Days 31-60: scale to 10/day, mix cold.
2. Run outbound from a borrowed older account. If a teammate has 3+ year account, use theirs with permission.
3. Never push volume to catch up. That is the trap; first restriction hits day 22.

High tool spend, low reply rate

$500+/mo on outbound tools with reply rate under 5% means the bottleneck is targeting + sequence quality, not tools. Teams that consistently land 10%+ reply rate at low tool cost use 2-3 tools, not 5-6.

Fix this month:

1. Audit each tool — if you cancelled tomorrow, would reply rate drop? Cancel the 30-50% that wouldn't.
2. Hire a B2B copywriter for 5-10 hours to rewrite touches 1-5. Highest-ROI use of outbound budget.
3. Track the funnel by touch, not by month. The touch with the worst ratio is what needs the next 10 hours.

When you would just have BIGM do it

If three or more of these gaps are open AND you have already rewritten touch 1 twice, the problem is operational, not positional. That is what BIGM ships: managed account pools, per-prospect AI openers on touches 1-3, channel switch automation on touch 4, and inbox handling so a reply at 2am does not wait until tomorrow morning.

See how BIGM works →