What you'll be able to do after this guide: put a standard set of checks in front of every outreach action, choose between human-sent and automatic sending, and respect both the person whose account is used and the platform's rules.
Every outreach agent runs the same checks before acting, in order of cost. A candidate stopped by an early check costs nothing: no browsing, no drafting.
1. Status gate. The company is at Review (or Contacted, for a follow-up) and has a research date. This is a CRM read only.
2. Deal gate. No open deal past its first stage. If a person is already in a conversation, the agent stays out. Check deals linked to the company and to the person, because duplicate company records mean a deal can hang off the "other" one. If the deal lookup fails, skip the candidate. A failed check must never be read as "no deal".
3. Reply check. If the prospect has ever replied, stop and flag them for a person. Not "replied since our last message": any reply at all. Don't snooze them; a snooze would hide them from the person who needs to answer. The agent checks replies the CRM knows about: synced emails and replies a person has logged.
4. Recent conversation. If there's any message in the last 30 days, from anyone, wait. This catches messages a person sent by hand. Where the agent can't see the conversation (on a social platform, for example), this check and the reply check become the first line of the person's task: "Before sending, check the thread. If they have replied, or anyone has messaged in the last 30 days, don't send; note why and complete the task."
5. Cooldown. At least 14 days since our last message, keyed on the person's profile URL rather than the CRM record, so duplicate person records can't slip a second message through.
6. Claim check. Re-verify any fact the message will mention, and confirm every link opens. Facts go stale: a tool confirmed in research can be gone a few weeks later. A page you planned to publish may never have shipped.
Only then does the agent act, and it records the action straight away, before moving to the next candidate. Deferring writes to the end of a run leaves a window where a message has been sent but the cooldown doesn't know about it.
Put this sentence in every outreach spec: "Never resolve a conflict in the direction of sending a message." If the spec and the prompt disagree, the more restrictive one wins. If a check can't be completed, the candidate is skipped.
We recommend a human-send queue as the default, especially on social platforms.
How the human-send queue works:
Each morning the agent selects candidates, runs every gate it can check from the CRM, and drafts the message.
It creates one CRM task per candidate containing the profile link, a short history, the research summary, the thread checks for the person to make, and the draft.
A person checks the thread, reads the draft, edits it if needed, sends it themselves, and completes the task. If they changed the message or didn't send it, they add a short note saying so.
On the next run, the agent reconciles: for each completed task, it reads the task and any note, and only if the person confirmed the send does it write the message date and set the company to Contacted. It records the message as actually sent, which may differ from the draft.
If a task was completed without a send (for example, the prospect had already replied), it records that and writes no message date.
The agent never writes a "sent" date until the send is confirmed. That keeps cooldowns and deal stages accurate even when a person changes their mind.
Check the platform's terms before automating anything on it. LinkedIn's User Agreement prohibits using bots or other automated methods to add contacts, send messages or collect data from the service. Use the human-send queue for any platform with rules like these. Reserve automatic actions for systems you own, like your CRM, or channels with an official, permitted integration.
Two runs given the same data should pick the same candidates. Write the ranking order into the spec, for example:
Fit score, highest first.
Research date, oldest first.
Record creation date, oldest first.
Other selection rules that matter:
Filter by status before ranking, not after. Otherwise an excluded company can win a slot and then be thrown away.
Share slots between segments. For example, at most 3 of 5 candidates from any one segment, so a busy segment doesn't starve the others.
Check candidates against each other within a run. CRM guards only see what's already saved. If a run picks two candidates, make sure they're different people at different companies.
A limit is a ceiling, not a target. "Up to 5" means don't lower a standard to reach 5. An empty run is a normal outcome.
Give every agent a per-run cap, and tell it what hitting the cap means:
| Agent | Example cap | If hit |
|---|---|---|
| First touch | 5 candidates | Normal |
| Acceptance check | 60 status changes | Unusual: first run after a long gap, or a matching bug. Stop and report |
| Message | 2 candidates | Normal |
| Email task | 10 tasks a day | If this binds every day for two weeks, raise the cap or source less |
| Deal sync | 30 conversations | Unusual for a small team: report |
When outreach happens from a person's own account, it should fit their working day and their reputation:
Create tasks within working hours, set in the scheduler, so actions happen when the person is at work.
One contact per company at a time. Several requests from one person to a five-person company reads as spray.
React only to safe content. If the person likes a prospect's post as a light first touch, give them a simple rule to apply: only posts about a company launch, a hire, an award or useful professional content. Never political or divisive content, personal life events (good or bad), charity, memorial or religious posts. The test: would you be comfortable seeing this reaction in a screenshot next to your company's name? If that needs any thought, skip it.
No suggested comments. A comment is a public statement and needs the person's own words.
If the platform shows a limit, stop. Don't look for a way round it.
Only treat a profile as the right person when the name, job title and company all match, ideally with one more detail such as a start date. Match profile URLs exactly; "jo-smith" is not "jo-smith-uk". A missed match only means a prospect waits. A wrong match puts a stranger into your outreach.
Next: The Email Task Agent.
