Intent data promises to tell you who is in market. What it actually reports is narrower, and the gap between the two is where budget goes to die.
What the signal really says
Third-party intent typically means: content consumption associated with a company's IP range or a cookie pool, matched to a topic taxonomy, compared against that company's own baseline.
So a "spike" means: someone, possibly, at that company, read something the vendor classified under a topic, more than usual.
Each of those hedges matters:
- Someone. Not necessarily your buyer. Could be a graduate doing research,
a competitor, or someone in an unrelated department.
- Possibly. IP matching is imperfect, especially with remote work.
- Classified. Topic taxonomies are broad, and adjacent topics collapse
together.
- More than usual. A baseline comparison, so small companies produce noisy
spikes on very little activity.
None of that makes it worthless. It makes it a prioritisation input.
The account-versus-person gap
The most consequential limitation. Intent is nearly always account-level, and you sell to people.
Knowing that someone at a 4,000-person company read about your category does not tell you which of the eleven relevant people to call, or whether the reader has any authority. On a 40-person company that inference is much safer, which is why intent works better downmarket than the vendors' enterprise case studies suggest.
The test before buying any source
One question: what does a rep do differently because of this signal?
Not "prioritise". A specific different action: call today instead of sequencing, open with this angle, work this persona rather than that one.
If nobody can answer, you are buying a dashboard. That test is the same one in trigger based prospecting, and it disqualifies most generic intent purchases.
First-party beats third-party
Your own signals are more specific and cost nothing extra:
- Pricing page visits, repeated.
- Documentation or integration page visits, which indicate evaluation rather
than awareness.
- A second person from the same account arriving within a short window, which
is a strong buying-committee signal.
- Reopening an old thread.
- A champion changing jobs, which is the highest-converting signal most teams
never watch.
Instrument these before buying third-party intent. Most teams have not, and they are strictly better signals.
Combining beats any single source
One signal is a reason. Two unrelated signals inside a window is a much stronger one, and counting distinct live signals per account is the cheapest scoring improvement available. Do that before building a weighted model.
Set decay windows
Every signal expires. An intent spike is live for days, not the quarter it sits in a dashboard. Without a defined window, reps work stale spikes and the message lands as obviously behind.
Where this sits
This is the Signal pillar of the Tenbound Pipeline Architecture Standard, the pillar most often skipped and most often bought rather than built. A team with excellent sequences and no signal layer produces volume without meetings, because the sequences are aimed at accounts with no reason to care this week.