Track the phrases your buyers post

Every other agent waits for someone to touch your content. Keyword tracking does not. It searches public LinkedIn posts for phrases you choose, and the person who wrote the post becomes the lead.

Before you start

• Have your ICP saved. Buying signals and product wording feed both the suggestions and the relevance check.
• Bring a handful of phrases your buyers actually type, not the words you use in your own marketing.

Add keywords

1. Create the keyword agent

   Add the Keyword Tracker from the agents screen. There is one keyword agent per workspace, and it holds all of your keywords.

2. Type each phrase

   Press Enter or a comma to add each keyword. Each one is capped at 85 characters, which is LinkedIn’s search limit. Quotes and boolean operators such as AND and OR are allowed, so "looking for" CRM for sales teams is a valid single keyword.

3. Or let Embers suggest them

   Suggest from your ICP generates search phrases from your product and buying signals, each with a one line reason. Add the ones that fit and edit the rest.

4. Manage them later

   Keywords are not write-once. The Tracked Keywords card on the agent, and the same card on the Keywords page (/keywords), let you add and remove them at any time, with the plan cap shown as a count.

Picking keywords that produce buyers

A keyword search returns any post containing the term, including jokes, news shares, and competitors selling the same thing. The difference between a useful agent and a noisy one is entirely in the phrasing.

• Write buyer language, not seller language. “Looking for”, “recommend”, “alternative to”, and “switching from” are the phrases a person in-market writes. Your own category slogan is what your competitors write.
• Name the pain, not the industry. “Manual data entry” beats “operations”. “ERP migration” beats “software”.
• Use your competitor names, paired with switching words where you can.
• Avoid single broad words. Terms such as marketing, sales, SaaS, AI, hiring, or growth are flagged in the app with a warning, because they return mostly unrelated posts.
• Add quotes for exact phrases when a loose match would drag in the wrong conversation.

What Embers does with a match

Each scan runs one batched search across all of your active keywords, on a 24 hour cadence. Matched posts then go through a relevance check before anyone reaches your queue:

• Off-topic mentions, competitor self-promotion, and recruiter or staffing job ads are marked not relevant and never surface as leads. Job ads stuffed with tool names are the single biggest source of noise in keyword search.
• Every surviving post gets a match strength of strong, partial, or lowest, an intent category such as seeking a solution, seeking a recommendation, or describing a pain point, and a one sentence match reason you can read on the Keywords page.
• Posts that name one of your ICP buying signals are boosted, so “looking for a tool like X, migrating off our current one” ranks far above a passing mention.
• Only authors of relevant posts are matched against your ICP, enriched, and scored.

Embers does not collect the reactions or comments on a matched post. The author is the signal.

Removing a keyword is a soft removal. It stops being searched and stops counting toward your cap, and the posts it already matched stay attached to their leads.

Related