Most B2B prospecting still starts with a filter: an industry, a headcount band, a job title. That produces a list of people who match a description and gives no reason to contact any of them today.
A buying signal is the other starting point. It is an action somebody took, so it arrives with a reason and a date attached, and both of those are things a filtered list can never give you.
What is a B2B buying signal?
A B2B buying signal is an observable action suggesting a company or a person is closer to buying than they were last week. A visit to a pricing page. A funding round. A job posting for the role that owns the problem you solve. A comment on somebody else's post about that problem.
The action is the signal. The person or company behind it is the lead. Keeping those two things separate matters, because the quality of a signal has little to do with the quality of the list it produces: a strong signal attached to the wrong person is still the wrong person.
What are the types of B2B buying signal?
Four families, which differ in where the data comes from and in whether it resolves to a company or to a named individual.
First-party
Your own site, product, docs and emails
Warmly, RB2B, Dreamdata
Third-party topic intent
Aggregated content consumption across a publisher network
Bombora, 6sense, G2
Company events
Funding, hiring, leadership changes, tech stack changes, mergers
Clay, Max, Satellyte
Public person-level activity
What a named person posted, reacted to or commented on
WarmDogo, Traxy, Common Room
The split that matters most is the last column. Three of the four families tell you an account is warm and leave you to guess which of its forty employees to write to. Person-level activity names the individual, because the signal was something that individual did in public under their own name.
Both are useful and they answer different questions. An account-level signal is a good reason to run a campaign at a company. A person-level signal is a good reason to write one message to one person today.
Which B2B buying signals actually convert?
Most writing on this subject ranks signal types by intuition. The numbers below come from production: every finished WarmDogo run up to 30 September 2026, across the signal kinds with enough runs to say something honest.
"Kept" below is the share of leads a person reviewed and chose to keep rather than skip, so it measures whether a lead was worth having and not merely whether it existed.
- Engaged with a trusted voice's post3.9 a run80%
- Engaged with a competitor's post0.25 a run66%
- Fits the ICP and posted recently6.6 a run62%
- Engaged with posts about a phrase3.5 a run49%
- Wrote a post using a phrase1.1 a run43%
- Company posted a job for the buying role0.57 a run34%
Three things fall out of that table, and two of them are uncomfortable.
Volume and quality are different axes. The kind that fires most often is a person who fits the ICP and posted this month, and it sits in the middle of the order above. The kind people keep most often produces well under half as many leads a run. A tool tuned for leads per run and a tool tuned for leads worth having would make different choices, and most marketing quotes the first number.
The rarest signal is among the best. Engagement with a competitor's post fires in a seventh of runs and produces a quarter of a lead when it does. Judged on volume it looks like a failure, and judged on what a salesperson does with the leads, it does not.
The company event is the weakest of the six. A job posting for the buying role is the signal most commonly recommended in articles about intent, and it sits last here at roughly half the keep rate of the engagement signals. A company hiring a role is evidence about the company. It is weak evidence about any particular person at it, and somebody still has to pick one.
One caveat worth stating plainly. These are LinkedIn signals, read for the customers of one product, and they are not a general benchmark for B2B intent. They are published because almost nothing in this category publishes anything. The nine kinds and how each is read are on the LinkedIn signals guide.
What makes a buying signal worth acting on?
Three properties, and a signal missing any of them is closer to a list entry.
It is recent. A signal describes a moment and decays as the moment passes. A reaction last week is a reason to write today. The same reaction from March is a name with a stale footnote.
It names a person. An account-level signal tells you a company is in motion and leaves the hardest part undone. Somebody still has to decide which human inside it to approach, and that decision is where most intent data quietly fails to turn into pipeline.
It is specific enough to repeat back. The test is whether you can state the signal in the first line of a message without sounding like you ran a report. "You commented on the thread about onboarding drop-off" passes. "Your company showed intent on the topic of customer success software" does not, and a reader who cannot tell what you observed assumes you observed nothing.
How do you act on a buying signal?
The order matters more than the tooling.
- Score the person, not only the signal. A strong signal from somebody outside the ICP is a distraction with a good reason attached.
- Write from the specific thing. The signal is the first line. If it cannot be the first line, it was too vague to be worth acting on.
- Go within days. Recency is doing most of the work, and a queue that takes a fortnight to clear has already spent it.
- Record what you kept and what you skipped. Keep rate per signal kind is the only honest measure of which sources deserve more of your week, and the table above exists because that was recorded.
- Leave the senders alone. Signals change which people enter a sequence. They do not change the sequence, and a tool that sources well pairs with whatever you already send from.
Where to start
Pick one signal family and measure it for a month before adding a second. Running four of them badly produces a queue nobody on the team trusts, and the keep rate per kind is the number that tells you which one has earned the next month of your attention.
For the LinkedIn family specifically, the nine kinds each have a page with what they read and what they produced. For the wider picture of finding buyers who are already interested, start with LinkedIn prospecting, or see what WarmDogo costs.