Which buying signals actually predict pipeline
We ranked the common intent signals by how much they move conversion. Job changes and funding are not equal, and one widely-sold signal barely registers.
"Intent data" is sold as a category, which obscures that the individual signals inside it differ from each other by an order of magnitude in usefulness. Some are close to a purchase decision. Some are noise with a confidence score attached.
Here is how they rank in practice, roughly by lift over a baseline cold list into the same ICP.
Tier 1: a person changed jobs
The strongest signal available, and it is not close.
A new VP or director in a function you sell to has three properties at once: budget authority, an explicit mandate to change something, and no loyalty to the incumbent vendor. The window is real and it is short — roughly the first ninety days, with the sweet spot between weeks three and ten. Before that they are still learning the organisation; after that the decisions are made.
Two refinements that matter more than the signal itself:
- A move within your customer base is worth several times a cold move. Someone who used your product at their last company and has just landed somewhere new is the highest-probability conversation in outbound, full stop.
- Seniority gates it. A new individual contributor is not a buying signal. A new head of function is.
Tier 2: funding, with a caveat
A funded company has money and a mandate to spend it. That is genuinely predictive, but the naive version of this signal is badly timed.
The week a round is announced, the inbox of every executive at that company is unusable. Everyone selling anything has the same alert. The signal is real; the timing everyone uses is wrong. Reaching out at six to ten weeks post-announcement puts you in front of the person at the point where the hiring and tooling decisions are actually being made, and after the noise has cleared.
Stage matters too. Series A and B are the band where net-new tooling decisions are made quickly. Later stages have procurement.
Tier 3: hiring surges
Job postings are a leading indicator with a useful property: they are specific. A company posting six SDR roles is going to need sequencing, data and dialling infrastructure within a quarter, and they have told you so publicly.
The signal degrades when read too broadly. "This company is hiring" means nothing. "This company is hiring for the function my product supports, at a rate materially above its baseline" is a real signal with a clear reason to reach out.
Tier 4: technology installs
Knowing that an account runs a particular stack is useful for two things: qualifying out (they have a competitor on a three-year contract) and positioning (they run the thing you integrate with).
It is weak as a timing signal, which is how it is usually sold. Someone installing a tool today tells you very little about whether they are in a buying cycle for an adjacent one. Use technographics to shape the message, not to decide when to send it.
Tier 5: anonymous web intent
The category most heavily marketed and the one with the weakest evidence behind it.
The mechanism — third-party content consumption, IP-resolved to a company, aggregated into a topic score — has real coverage problems and no visibility into who was reading. A "surge" at a five-thousand-person company might be one analyst doing research for a report.
It is not worthless. It is worth roughly what a well-defined ICP filter is worth, which is much less than its price.
How to combine them
The mistake is treating signals as a queue. A ranked list of every account that fired any signal, worked top to bottom, converges on working the noisiest signal most often.
Use them as a multiplier on ICP fit, not as a substitute for it:
priority = icp_fit_score * signal_weight * recency_decayAn account that fits badly should not reach the top of a rep's day because it raised money. An account that fits perfectly and just hired a new head of the function you sell to should be contacted this week.
The point of a signal is not to find someone to talk to. It is to decide which of the people you already wanted to talk to should hear from you today.
That reframing removes most of the disappointment teams have with intent data, because it stops asking the data to do the targeting work that belongs upstream of it.