Choose Keywords That Find Buyers

The same workspace, the same ICP, and the same scoring can produce anywhere from 0% to 47% high-fit leads. The difference is the source. This guide explains what separates a source that finds buyers from one that wastes credits. Everything in it comes from measured fit rates on real Signado sources, not theory. Sample sizes vary and are noted where they are small.

What makes a good keyword

A keyword tells Signado which LinkedIn posts to look at. The people who wrote those posts, or commented on them, become your potential leads. So the real question behind every keyword is: who talks about this on LinkedIn? Not who you want to find. Who actually writes and comments on posts containing these words.

The keywords that work are phrases your buyer would type into a post when their problem is bothering them. The keywords that fail are labels: the name of your product category, or the job title of your target. Labels feel natural because that is how you describe your market internally. But your buyer does not talk in labels. They talk in problems.

Measured on real sources in engager mode:

Keyword	What kind of phrase it is	High-fit rate
cold email is dead	Something a buyer would say	14.6%
lead generation	A product category	9%
n8n	A tool name	2%
hiring manager	A job title	0 of 5

Not every conversational phrase is a winner (reply rate dropped ran at 8.2%), but the shape sets your ceiling: every top source we have measured is a phrase a buyer would say, never a label.

Some rewrites, from label to buyer phrase:

• outbound sales (category) becomes cold email is dead (an opinion buyers debate)
• real estate broker (job title) becomes listings sitting too long (a pain that broker posts about)
• employee onboarding (category) becomes new hires quitting in 90 days (a complaint that buyer writes)

The next two sections explain why the two label types fail, because each fails for a different reason.

Why job titles find job seekers, not buyers

Ask yourself who writes or comments on a LinkedIn post containing the words "hiring manager". Almost never a hiring manager. A person who holds a job does not post about their own job title. The people who do talk about a job title are the people circling it: candidates who want that job, career coaches selling advice to those candidates, resume writers, and recruiters trying to fill the role.

We measured this directly. A source for hiring manager in engager mode returned five leads. None scored high fit. The crowd was job seekers and a resume coach, and the preview had shown exactly that before the source was ever activated.

The fix is to stop targeting who your buyer is and start targeting what your buyer says. If your buyer holds the hiring manager title, they never post "hiring manager", but they do post about the problems of the job: candidates ghosting after the offer, time to hire keeps slipping. Those phrases find the person holding the title, because those are the words the title-holder actually writes.

Why your product category finds competitors, not buyers

The second label trap is subtler. Suppose you sell lead generation services, so you add lead generation as a keyword. It describes your market perfectly. The problem: who posts about lead generation on LinkedIn? Mostly people who sell lead generation. Agencies, tool vendors, consultants, growth gurus. Your competitors and peers, having a conversation with each other.

The mechanism is a vocabulary split. Buyers describe problems in problem words. Sellers describe those same problems in category words. A founder who needs leads does not post "lead generation"; they post pipeline dried up or referrals dried up. "Lead generation" is the seller's word for the solution to that complaint. When you use the seller's word as your keyword, you collect sellers.

You can see this in the data. In our scored leads, roughly 15% of everyone discovered were capped as likely competitors by the ICP scorer. Signado's scoring catches them, so they will not clutter your high-fit view. But scoring happens after discovery, and discovery costs 3 credits per lead. The scorer protects your attention; it cannot refund the credits. It is cheaper never to invite competitors in the first place.

Rewrites for this trap: lead generation becomes pipeline dried up. cold email tools becomes cold email is dead. In each case you swap the solution's name for the complaint that makes someone need the solution.

Post authors vs post engagers: which mode to use

Every keyword source runs in one of two modes, and the mode decides who becomes a lead when Signado finds a matching post:

• Post authors collects the person who wrote each matching post.
• Post engagers collects the people who commented on each matching post.

The two modes produce very different crowds, and the reason is self-selection. Writing a LinkedIn post about a problem is a strong signal: the author chose the topic, thought about it, and put their name on an opinion. People who do that are almost always living the problem professionally: operators, founders, consultants, practitioners. Commenting is a much weaker signal. A comment thread on a popular post gathers buyers, but also peers, job seekers, students, and people who simply enjoy arguing. The author self-selected; the commenters just showed up.

The measured difference is large. In one workspace, cold email in author mode ran at 47% high fit (13 of 17 scored leads) and outbound at 28%. Engager sources on strong phrases ran between 8% and 15%. Author mode consistently produces fewer leads per post (one author versus potentially many commenters), but a far higher share of them fit.

How to choose:

• Choose Post authors when your buyers write posts about the problem. This is true for founders, consultants, and operators in professions that post publicly. You get fewer, better leads, and your outreach can reference something they wrote themselves.
• Choose Post engagers when you want volume, or when your buyers react rather than publish. Some buyers never write posts but do comment on them. Engager mode also hands you the comment text, which gives outreach a natural opening.
• Use both when both signals matter. Add the same keyword twice, once per mode. Each uses its own source slot and can be previewed and activated separately.

The full mechanics of the two modes are in Post Authors vs Post Engagers.

How to choose a creator or competitor profile

A creator or competitor source watches a LinkedIn profile and surfaces the people commenting on that profile's posts. One question decides whether it works: is this profile's audience made of your buyers?

The measured spread is wide. Creator sources ran at 15 to 22% high fit when the profile's audience matched the workspace's ICP, and about 3% when it did not. Same feature, five times the yield, and the only variable is whose audience it is.

The common mistake is picking profiles you follow: the big names in your own industry. But a peer's audience is more peers. If you sell to SaaS founders, another agency owner's followers are mostly other agencies. The profile you want is the one your buyers read: the growth voice founders follow, the industry figure your customers quote. Ask where your existing customers spend their LinkedIn attention, and put your source there.

Check a source with a preview before spending credits

A preview costs 1 credit and shows you the real crowd before discovery starts: matching posts, plus the actual authors or commenters the selected mode would surface, with their names and headlines. Signado adds a one-line crowd check comparing that sample against your ICP. Active discovery costs 3 credits per lead, so the preview is the cheap moment to find out a source is wrong.

Read a preview with one question: would you want a meeting with these specific people? Not "is the topic relevant" but "are these humans my buyers".

Two patterns from our data are worth repeating:

• The worst source we measured (166 leads, 1.2% high fit) was never previewed. One credit would have shown the mismatch before 160+ leads were paid for.
• Believe what the preview shows you. In one measured case, the preview for a keyword meant to find hiring managers surfaced a resume writer as a top potential lead. The source was activated anyway and delivered zero high-fit leads. The preview was right, and it usually is: it is drawn from the same crowd discovery will pull from.

Read the fit stats and replace weak sources

Once 10 of a source's leads have been scored, its card on the Discovery Sources page shows what share of them scored high or medium fit. This turns keyword choice from a one-time guess into a feedback loop: activate, read the numbers, keep what works, rewrite what does not.

For calibration, from our measured data:

• About 10% high fit is the normal baseline across workspaces.
• Sharp, conversation-specific keywords reach 20% or more.
• The best measured sources, author mode on a phrase buyers post about, ran above 40%.

A source stuck under 5% after 20 or more scored leads is telling you something: the crowd it attracts is not your crowd. Pause it, rewrite the phrase, or spend the slot on a different source. This matters more than it looks, because every active source takes an equal share of your daily budget. A weak source does not just underperform; it starves your good sources of budget.

Checklist before you activate

1. Is it a phrase your buyer would post when the problem bites?
2. Is it a job title? It will find job seekers, not the people holding the title.
3. Is it your own product category? It will find competitors, not buyers.
4. Do your buyers write posts about this problem? If yes, try author mode.
5. For a profile source: do your buyers follow it, or do you?
6. Preview it. Would you take a meeting with the people it shows?
7. After 10 scored leads, check the fit rate on the source card. Replace what stays low.