Most "signal-based outbound" is a filter with a better name. A real signal has to clear three tests: you can detect it reliably, it carries a date, and knowing it changes what the first line of the email says.
If it fails any of the three, it's a segment, not a signal.
The eleven
| Signal | What it tells you | Where it comes from |
|---|---|---|
| Role open 30+ days | Their own hiring has failed | Careers pages, job boards |
| Same role reposted | The first search failed and they restarted | Job posting history |
| Funding in last 90 days | Budget approved, headcount signed off | Crunchbase, press |
| New VP or C-level in seat | New leaders rebuild in their first quarter | LinkedIn job changes |
| Job ad names a tool | They published their own stack, for free | Job description text |
| Detected tech stack | You can price what they already pay | Stack detection on their domain |
| Hiring in an expensive metro | The cost gap is widest and they feel it | Job location |
| New site, plant or office | A whole team is needed, not one hire | Trade press, planning registers |
| Planning permission granted | Fires months before the job ads do | Public planning registers |
| Contract or project award | Staffing follows within weeks | Tender portals, announcements |
| Competitor's community | People already educated, by someone else | Public groups and forums |
The last one is the most underrated. A competitor's user community is a list of people who already believe the problem is worth solving — somebody else paid to convince them. On one campaign that list was 1,928 people; 1,700 contacted, 19 replies, 16 positive. The reply rate looks modest until you notice there was no education call needed on any of them.
The four that waste time
Website visitors. Unless they identified themselves, you are guessing at a company from an IP address, and the match rates are far worse than the vendors imply.
Generic "intent data." Someone at a 4,000-person company read an article about your category. That is not a buying signal, that is an internet.
Social engagement. A like is not intent. It is a thumb.
Technographics with no price attached. Knowing they use HubSpot is only a signal if you can say what it costs them or what it fails to do. Otherwise it's trivia you've paid for.
The test that matters
Write the first line of the email before you build the signal.
If you can write a specific, dated, slightly uncomfortable opening sentence from it — "your Site Manager role in Jebel Ali has been open since July" — it's a signal worth the engineering. If the best you can manage is "I saw you're in SaaS," you've built an expensive filter.
Start with three, not eleven
Building all eleven is a quarter of work and most of them won't fit your buyer. Pick the three with the best coverage-to-effort ratio for your ICP and get a working sequence out of the door.
For most B2B companies that's roles open 30+ days, recent funding, and a detected tech stack. Those three are public, dated, cheap, and between them they cover a surprising share of any list.
Add the fourth once the first three are producing. Signals are only worth what the copy does with them, and copy is where the time should go.