What are B2B buying signals?

1. What actually counts as a buying signal

A useful signal has three properties. It is observable, meaning someone else could go and check it. It is time-bound, meaning it happened recently. And it implies something, meaning it points at a need, a new decision-maker, or money that is now available.

A firmographic fact is not a signal. "Series B, 120 employees, fintech, based in Austin" describes fit. "Their head of sales left last week" is a signal. The first tells you who to aim at. The second tells you why to aim now.

Signals also work better in combination than alone. A single job posting might be routine backfilling. A job posting, a new leadership hire, and a funding round in the same quarter is a pattern worth a call.

Leading and lagging signals

Some signals arrive early and are noisy. A role being advertised means the need exists, but the person who will own the decision may not have started yet. Others arrive late and are reliable. The new hire is in the seat, has a mandate, and is reviewing the stack. A good system mixes both: enough early ones to stay ahead, enough late ones to be worth a seller's time.

2. The main categories of buying signals

Almost every signal a lean team can realistically track falls into one of these categories. The examples matter less than the pattern: something public changed, and the change implies a decision.

Specificity is what makes a signal worth acting on. "They raised" is weak. "They raised to expand into a second market, and the announcement named a hiring goal" is a reason to write something a person will read.

3. Buying signals and ICP fit do different jobs

These two get merged into one conversation constantly, and it costs teams money in both directions. Fit decides who belongs on the list. Signals decide who is first on it.

A signal on a bad-fit account is still a bad-fit account. Fit without a signal is a list that quietly ages out before anyone works it. You need both, and they should be maintained by different rules.

4. How to automate signal detection

Manual signal research works until the list gets longer than one person's memory. The goal of automating it isn't to remove judgment. It's to make the same judgment every day without paying someone to browse.

1. Write the rule before you pick the tool. "If this happens at an account in our target market, tell a seller." If you can't write that sentence, the signal isn't ready to automate.
2. Assign a source to each signal. Most have a public or paid one. Apollo is useful for the base list and for filtering on hiring and funding attributes. Job boards, Crunchbase, grant registries, and your own analytics cover the rest.
3. Make it a repeatable pull. A Clay table or workflow can run the same research on a schedule: pull accounts, enrich them, run the lookups, flag which signals are present. Keep the logic visible so someone can debug it when a lookup starts returning junk.
4. Score the combination, not the event. Weight accounts by how many signals fired and how recent they are. A single old signal ranks below two fresh ones.
5. Attach the evidence. The seller should see the signal, the source, the date, and a link to the original. Without the link they can't verify it, and a signal they can't check gets ignored.
6. Map every signal to an action. A sequence, a call, a Slack alert, a nurture list, or an exclusion. A signal that routes nowhere is a report nobody reads.
7. Audit it on a schedule. Signal systems drift quietly. A scraper that stopped working looks exactly like a market that went calm. Sample the flagged accounts weekly and check that the data is still true.

Tools are evidence here, not the plan. Clay is useful when a signal needs several sources stitched together or when raw data has to become a sentence a seller can use. Apollo is useful for volume and contact data. Neither decides which signals matter to your market, and no tool fixes a rule that was wrong to begin with.

5. Choosing the first signals to track

Start with two or three. A short list gets used, checked, and improved. A long one becomes background noise within a month.

Signal checklist

• At least one signal is tied to a reason a buyer would spend money
• At least one is tied to a change in who owns the decision
• You can verify it in under a minute from a link
• The source updates often enough that the signal isn't history
• The output volume is something a seller can actually work
• A seller has looked at the list and agreed they'd use it
• Someone owns the queue and notices when it goes quiet

Volume is the underrated test. If your system flags 300 accounts a week and two people work outbound, you've built a reporting problem, not a pipeline. Narrow the market or raise the bar until the queue is something you finish.

6. Using signals in outreach without sounding automated

This is where good signal work dies. The recipient can usually tell whether the system researched them or merely found them.

• Put it in the first line, plainly. One sentence, no flattery, no throat-clearing. If the reason for the email takes three lines to reach, it's not the reason.
• Don't narrate your monitoring. "Congrats on the expansion" is fine. "We've been tracking your hiring" is not, and it makes the reader wonder what else you know.
• One signal per message. Referencing four facts reads like a dossier. One good detail reads like someone paid attention.
• Connect it to a problem they'd recognize. Not "you raised, so buy this." More like "teams that just opened a second market usually hit the same onboarding wall first."
• Apply the read-aloud test. If the line would embarrass you to have it read back to you, take it out.

The same logic applies to timing. A signal that's three weeks old is a worse opening line than no signal at all, because it shows you were late to something they've already lived through.

7. Common mistakes

• Tracking everything. More signals means more noise, more cost, and a queue nobody trusts enough to open.
• Confusing a signal with qualification. A funding round doesn't mean fit, and a new hire doesn't mean budget.
• Working stale signals. A six-month-old announcement is company history, and the buyer knows it.
• Skipping verification. Provider data is wrong often enough that an unchecked signal will eventually reach a seller who catches it, and then nobody trusts the next ten.
• No owner. Unattended signal systems decay silently. The scraper breaks, the source changes format, and the queue goes empty without anyone noticing.
• Using a signal that shouldn't be used. Some things are legal to see and still unpleasant to be addressed by. Judgment is part of the system.

Related reading

• Outbound email automation for lean B2B teams
• What is lead enrichment?
• What is an ideal customer profile?
• What is GTM engineering?
• What is CRM hygiene?