Your first 30 days of LinkedIn outbound

Most outbound advice assumes you already know the rhythm. This playbook is for the case where you do not. Thirty days is enough time to set up a system, get reliable replies, and decide whether outbound earns its place in your week. Less than that is guessing; more than that and you have wasted a month you cannot get back.

The plan: one week each for setup, first sends, tuning, and scale-or-pivot. Each week has a single goal. Skipping a week does not save time; it costs week three.

Week 1. Setup and calibration

The goal of week one is not to send anything. It is to configure the machine so that when you turn it on in week two, you are not also debugging it.

Day-by-day:

• Day 1: Spin up your workspace. Paste the URL of what you are taking to market; let the AI draft the ICP. Then narrow it. The drafted ICP is a starting point, not a finished one.
• Day 2: Connect your LinkedIn account (and a second sender if you have one). Funkel applies a fourteen-day warmup ramp automatically; the daily cap starts at 20% and rises to full by day fourteen. The ramp window aligns roughly with your first sends in week two.
• Day 3: Pick exactly one signal to start with. Most teams should pick recent_job_changes (broad, easy to read) or competitor_pain (narrower, higher conversion). Avoid running multiple signals in week one; you cannot attribute results.
• Day 4: Write your connection note. One sentence, citing the signal that triggered the outreach. See the connection notes guide for the rules.
• Day 5 to 7: Set up your workflow steps, turn review mode on, leave the agent paused. Resist the urge to launch on day five; let the warmup ramp accumulate.

Week 2. First sends, on conservative caps

Day eight: turn the agent on. Set the per-sender weekly invite cap to forty (half of Funkel’s default of eighty). The warmup ramp is still scaling up; combined with the lower cap, this produces a manageable trickle of invites.

Review mode stays on for the first three days. Every draft message queues for your approval before sending. Read each one. Approve when it sounds like you, edit when it does not, reject when it should not have been written. After roughly twenty approvals, you will know whether the AI mode is in your voice; switch review mode off if it is.

The numbers to expect in week two:

• Invites sent: roughly twenty to thirty across the week per sender, given the warmup factor.
• Accept rate: any number is data. Below 25% is a signal to fix the note in week three; above 35% is a signal you can scale faster.
• Replies: one or two if you are lucky. Replies on twenty-something invites is statistically noisy; do not over-react either way.

The daily digest email lands in your inbox each morning at 09:00 UTC summarizing the previous day. Skim it; do not live in the analytics panel.

Week 3. Fix exactly one thing

Week three is the discipline week. The temptation is to change everything because the numbers are not yet what you want. Resist. Pick the one thing that is most clearly off and fix only that.

The hierarchy of what to fix, in order:

1. Acceptance below 25%: the connection note is the problem. Rewrite it. Do not change the signal or the ICP; you cannot diagnose either with a broken first-impression.
2. Accepts but no replies: the first message after accept is the problem. Cut it to two or three sentences and ask one specific question.
3. Replies but wrong people: the ICP is too loose. Add hard-no clauses (“not students, not consultants”).
4. Zero leads in the pipeline: see the quiet-funnel triage playbook.

Make exactly one change. Wait 48 hours. Read the new numbers. Then either keep the change or revert. This is the boring discipline that separates teams that compound from teams that fiddle forever.

Week 4. Scale or pivot

By day twenty-two you should have enough data to make a clean decision. Three outcomes are possible.

Scale. Acceptance above 30%, replies landing, demos getting booked. Raise the per-sender weekly cap to the default of eighty. Add a second signal to the agent, but only one new one (still no more than two running). Switch review mode off if you have not already.

Pivot. Acceptance below 20%, no replies worth following up. The signal is wrong for your category. Switch to a different one (if you ran job changes, try competitor_pain; if you ran competitor_pain, try an engagement signal). Reset the calibration; week two starts again on a different premise.

Stop. If after thirty days the inbound or the warm-network channels you have are still producing better results, pause outbound and revisit in a quarter. Outbound is a force multiplier for some businesses and a distraction for others. Knowing which one you are is more valuable than running another month of mediocre sends.

What to ignore in the first 30 days

• Open rates on individual messages. Volume is too low to mean anything.
• A/B testing message variations. You do not have enough data; pick one and commit.
• Adding a third or fourth signal “to see what happens”. You will see noise and call it data.
• Comparing your numbers to public benchmarks. Benchmarks are mostly survivor-biased and cohort-specific.

For more on which signal mix to settle on after the first thirty days, see the signal-mix playbook.