Turn reactions on your posts into leads

This is the agent most accounts start with. It watches the public posts on a profile you added to the Watchlist with the Own role and collects everyone who reacts to them or comments on them. Those people already know who you are, which is why they open a conversation better than a cold list does.

Before you start

• At least one profile on the Watchlist with the Own role, added by pasting its public URL. Own profiles count toward the watched profile cap like any other entry, and you can add several within it.
• An ICP saved, so the people found have something to be scored against.
• Something published. If the Own profile has no public posts in the lookback window, there is nothing to read.

What gets captured

Each scan collects the profile’s recent public posts, then the engagement on each one:

• Reactions. Like, praise, empathy, interest, appreciation, love, insightful, and funny are each stored as their own engagement type, so a considered reaction outranks a reflex like.
• Comments, with the comment text kept as evidence you can quote in outreach.
• Reposts, which carry the most weight of any single interaction.
• The timestamp of each engagement, which drives recency in the score.

Every post is classified into a topic category with a relevance weight, and every comment is classified into an intent category. That is why two people who both commented can score very differently: one wrote “congrats”, the other described the problem you sell into.

What happens after collection

1. People are resolved

   Each engager is stored once per workspace, so the same person reacting to five posts is one lead with five signals, not five leads.

2. ICP matching

   New people are matched against your ideal customer profile using their role, company, and what they actually did.

3. Profile enrichment

   Matched people, plus the strongest non-matching engagers, have their public profile read for role, company, and work history.

4. Scoring and ranking

   Everyone is rescored, and the queue re-sorts. Repeat engagers, recent engagers, and people who comment rather than react rise to the top.

Cadence and collection limits

The agent runs on a 24 hour cadence. There is one post engagement agent per Own profile, and its schedule is fixed at every 24 hours.

Each scan is bounded so a viral week cannot run away with your data or your bill:

A very popular post is also damped in scoring, so a thousand drive-by likes on one post do not outrank a handful of comments on a post that mattered.

Troubleshooting

No leads after the first scan. Check that the Own profile actually has public posts in the window, and that people engaged with them. Embers tells you which of those two was missing rather than showing a blank queue.

People show up who are clearly not buyers. That is expected on a post that travelled outside your niche. Mark them as not a match on the lead, and the ICP classifier learns from it.

A scan says it failed. Open the scan history on the agent and read the message. The two common causes are a profile LinkedIn could not read and a temporary data outage. Both are safe to retry.

Related