Buying intent data: sources, quality and how to act on it

Intent data promises the closest thing outbound has to mind-reading: which companies are researching your category right now. The promise is real - research behavior is observable and it does predict purchase windows - but the category is also where pipeline budgets go to produce dashboards nobody acts on. The difference between the two outcomes is rarely the vendor; it is whether the buyer understood what the data observes, judged its quality honestly, and routed it into action while it was fresh. This guide covers all three, as the intent chapter of the signal-based selling pillar.

⟨ KEY TAKEAWAYS ⟩

Intent data observes research behavior - content consumption on a topic, aggregated to the company - and infers that someone inside is studying the problem.
The families differ structurally: third-party co-op data is broad but noisy; first-party signals from your own properties are narrow but true.
Quality lives in three vendor questions: how is the company identified, what taxonomy defines the topic, and what baseline defines a surge.
Trajectory beats spikes: sustained rising research across weeks means a process; one spike often means one curious person.
Intent decays in weeks - routing latency is the difference between a warm signal and a stale list.
Intent alone justifies a relevant touch; intent converging with hiring or funding justifies priority - the combination rule from the pillar.

What intent data actually observes

Strip the marketing and the mechanism is simple: networks of publishers and data co-ops observe content consumption - articles read, comparisons viewed, docs visited - resolve the traffic to companies, and aggregate it into topic-level signals. "Acme Corp is surging on 'sales intelligence'" means people whose traffic resolves to Acme consumed meaningfully more sales-intelligence content than their baseline. That is genuinely useful, and it is also three inferences deep: the resolution could be wrong, the topic taxonomy could be loose, and the reader could be an intern writing a term paper. Good intent practice is just managing those three inferences deliberately.

The source families

WHERE INTENT SIGNALS COME FROM

Family	What it sees	Honest character
Third-party co-op	Content consumption across publisher networks	Broadest coverage; noisiest resolution; the default "intent data"
Review-site activity	Category and comparison research on review platforms	High purchase correlation; small volumes; late in the journey
First-party	Your site, docs, pricing page, product signals	Ground truth; tiny reach; the anchor to calibrate everything else against
Search-derived	Question and comparison query demand around the category	Market-level rather than account-level; sizes the water, not the fish

The families are complements, not substitutes: first-party signals calibrate what a real in-market account looks like; third-party extends that shape across the market; search-derived data sizes the demand the AI advertising plan will buy against. Treating third-party co-op data as ground truth is the original sin of most failed intent programs.

The quality questions to ask any vendor

Three questions separate usable intent from expensive noise. How are companies identified? IP resolution quality varies enormously, degrades with remote work, and small-company resolution is systematically worse - ask for the method and the claimed accuracy, then spot-check against accounts you know. What defines the topic? Taxonomies differ: "marketing automation" might include adjacent content that poisons the signal, or exclude the comparison pages that matter most - read the actual topic definition, not the label. What baseline defines a surge? A surge relative to the account's own history means something; a raw volume threshold mostly measures company size. Vendors fluent in these three answers are selling data; vendors who redirect to dashboard screenshots are selling dashboards.

Trajectory beats spikes

The single upgrade that most improves intent programs costs nothing: read trajectories instead of moments. One week's spike is one curious human; four weeks of rising, broadening research - more topics, more sessions, comparison content appearing - is a buying process with multiple participants. Score trajectory (direction and breadth over a window), not presence, and route the sustained risers. This is also where agentic execution quietly matters: computing trajectories across thousands of account-weeks is exactly the sandbox arithmetic a mission does exactly and a spreadsheet afternoon does approximately.

Acting on intent: the routing patterns

Intent expires in weeks, so the program is only as good as its routing latency. Three patterns cover the field. The weekly sweep: rank the market's risers, dedupe against accounts already in motion, verify contacts for the implicated roles, and land the top N in the CRM with the evidence attached - the intent-to-meetings playbook end to end. The trigger route: an account crossing the trajectory threshold posts to the owner's channel within the day, with context and a suggested first touch per the personalization rules. The convergence queue: intent converging with hiring or funding - the pillar's combination rule - jumps every other queue, because independent evidence agreeing is the strongest signal outbound gets.

The failure modes

How intent programs actually die

Dashboard death: signals reviewed monthly in a meeting, acted on never - intent is a perishable, and a month-old surge is archaeology. Spray-on personalization: blasting "I saw you're researching X" at every surge - creepy when wrong, lazy when right; the signal should shape the offer, and only public evidence gets referenced. Solo-signal worship: treating intent as sufficient rather than as one vote that wants corroboration. The common repair: wire intent into a weekly routed motion with combination scoring, and measure the program on meetings from routed accounts - never on signals collected.

Frequently asked questions

What is buying intent data?

Aggregated research behavior - content consumption on a topic, resolved to companies - used to infer which accounts are actively studying a problem category right now. It observes research, and infers the buying interest behind it.

How accurate is intent data?

It varies with three things vendors should answer plainly: company-resolution method, topic taxonomy definition, and the baseline behind "surge". Calibrate any third-party feed against your first-party signals before trusting it at scale.

What is the difference between a spike and real intent?

Trajectory: one spike is often one curious person, while weeks of rising, broadening research across related topics indicates a multi-person buying process. Score direction and breadth over a window rather than single-week presence.

How fast does intent data go stale?

Research windows run weeks, so signals decay on that scale. Programs win or lose on routing latency - a same-week touch on a rising account versus a monthly review of expired surges.

Is intent enough to justify outreach on its own?

One strong, fresh, specific signal justifies a relevant touch. Priority treatment belongs to convergence - intent plus hiring or funding on the same account - because independent evidence agreeing is the strongest predictor available.

Sources

⟨ RUN IT INSTEAD OF READING IT ⟩

Every playbook on this blog ships as a runnable mission.

Open a workspace and the playbook library is waiting - describe the outcome and the agents carry it end to end, on your plan's monthly credits.

Open a workspace →Read the docs