The four types of intent data are search intent, topic (research) intent, engagement intent, and firmographic-change signals. I learned to split them the embarrassing way. In 2023, our dashboard had ONE column called “intent,” fed by four totally different things. An SDR called a webinar attendee and opened with “saw you’re hiring.” Wrong evidence, wrong opener, dead conversation. So let’s pull the four types apart, with examples, and figure out which one your team actually needs.
📌 TL;DR: Search intent = what people type into search engines (anonymous demand). Topic intent = accounts reading category content across third-party networks (early but fuzzy). Engagement intent = behavior on YOUR properties (warmest, narrowest). Firmographic-change signals = what the company itself just did: hires, funding, headcount moves. Most vendors sell the first three. The fourth needs snapshot infrastructure, which is why so few offer it.
What Is Intent Data?
Intent data is behavioral or situational evidence that an account may be getting ready to buy. The plain-English definition lives in our intent data wiki entry; this page is the working map. Because here’s the thing: “intent” has become a bucket word. Four different data streams get poured into it, and each one justifies a different action. That’s also where intent fits inside the wider world of B2B buying signals: it’s one family among several.
So instead of one abstract definition, you’ll get intent data examples per type, where each comes from, and the blind spot each one carries. And one promise up front: I’ll be honest about the type we sell too, limits included.
Why bother separating the types of intent data at all? Because the label decides the follow-up. A search spike justifies content. A topic surge justifies soft outreach. A demo request justifies a call within the hour. And a leadership change justifies a note about THEIR change, not your feature list. Mix the labels and you get my webinar-opener story. Every mislabeled signal turns into a mismatched email.
Type 1: Search Intent Data
Search intent data captures what people type into search engines around your category. It’s demand in its purest written form: questions, comparisons, “pricing” queries, competitor names.
Where it comes from: search platforms, SEO tools, and search-partnership datasets. Typical examples look like:
- A spike in searches for “email verification tool” in your target region
- Rising queries comparing your competitor’s name with “alternatives”
- Growing volume on a problem phrase your product solves
The strength is honesty: nobody types “CRM migration checklist” for fun. The limit is anonymity. Search data is aggregate. It tells you the MARKET is asking, almost never which account. One caution before we move on: this is not the same thing as “search intent” in SEO, which classifies queries as informational, navigational, commercial, or transactional. I’ll untangle that in the FAQ.
How to act on it: feed content and paid campaigns, not your SDR queue. Search intent tells marketing what to build this quarter. Handing a keyword report to a rep and calling it “intent” is how dashboards lie to sales teams.
Type 2: Topic (Research) Intent Data
That’s the market’s questions. Now the accounts doing the reading. Topic intent data flags companies consuming category content across third-party networks at unusual rates. This is the type most people mean when they say “intent data,” and it’s usually sold at the account level.
Where it comes from, in plain words: publisher co-ops (B2B media sites pooling reader behavior) and bidstream data (the exhaust of real-time ad auctions, where page URLs and rough identities pass by). Review-site category activity often gets bundled in here too. A classic example: an account reading “email deliverability” articles at four times its own baseline this week gets flagged as surging.
The strength: it’s the earliest wide-market view of research you can buy, and buyers really do research long before they talk to anyone. The limits: the person is unknown, the topic is not your product, freshness lags behind aggregation, and the collection still leans on cookies that browsers keep restricting (Chrome’s Privacy Sandbox has wobbled for years while Safari simply blocks). I walk through the plumbing in how intent data is collected, and the ownership split in first-party vs third-party intent data.
🧠 Fun Fact: Most "intent" pixels never know WHO is reading. They infer the account from IP addresses and cookies, and Safari has blocked third-party cookies by default since 2020. The person behind a surge stays a guess.
How to act on it: treat a surge as a WHERE, not a WHO. Add surging accounts to research-stage nurture, have an SDR find the likely buyer inside each one, and keep the opener soft. “Noticed teams like yours are rethinking deliverability” survives scrutiny. “I saw you reading articles” does not, and yes, people write that.
Type 3: Engagement Intent Data
Engagement intent data is first-party behavior on properties you own: your website, your product, your emails. It’s the warmest type because the account already found you.
Examples worth acting on: pricing-page visits, demo-form fills, webinar attendance, email replies, and product usage spikes on trial accounts. Collection is consented and controlled by you, which keeps it on the friendly side of cookie-consent rules. And conversion rates on this type embarrass the other three, because the evidence maps directly to your product.
But the blind spot is brutal: engagement data only sees accounts that already arrived. If your traffic is thin, this type is thin. My webinar-opener disaster came from mislabeling exactly this: the attendee HAD engagement intent, and my SDR read it as a hiring event. Label the bucket and the opener writes itself.
How to act on it: speed and specificity. Route hot engagement to a human the same day, and reference the actual behavior honestly (“you asked about pricing tiers on Tuesday, here’s the straight answer”). This is the one type where being direct is a courtesy, because the buyer started the conversation.
Type 4: Firmographic-Change Signals (The Type Most Vendors Can’t Offer)
This type is different in one sentence: it tracks what the COMPANY did, not what its people read. A new sales leader. A funding round. Headcount up 12% in a quarter. A hiring freeze. An office opened in a new country. Each is a dated event in the company’s own reality, a buying signal rather than a research trace. Sellers chased these sales triggers by hand for decades; job change tracking is the best-known slice of the family.
How detection actually works, mechanically: crawl public company and people profiles on a schedule, store each crawl as a snapshot, diff consecutive snapshots, and fire deterministic triggers on what changed. Quantitative changes get a magnitude bucket: low is a 1 to 5 percent change, moderate is 5 to 15 percent, high is 15 to 30 percent, and hyper is 30 percent or more. Some triggers carry exact math: a layoff signal fires when headcount drops 10% or more in a single interval, and a jobs-open spike fires when open postings hit twice the rolling baseline of the previous four crawls.
📌 Example: → Crawl Monday: 54 employees → crawl Thursday: 61 → +13% between snapshots → a moderate-magnitude growth signal, time-stamped, with zero cookies involved.
Why can’t most intent vendors offer this? Because it requires longitudinal snapshot infrastructure across the whole company universe, refreshed continuously. A tracking pixel can’t see a hiring freeze. A publisher co-op can’t see a leadership change. Macro hiring data is public every month in the government’s JOLTS releases, but detecting it company by company, daily, is the hard part. For disclosure: this is the corner CUFinder builds in. Its buying signals engine tracks 99 signal types across 10 categories, refreshed daily against 1B+ contact profiles and 85M+ company profiles, and each signal is queryable by name and magnitude through the Company Signals API. The company’s own docs frame the gap this type fills better than I can:
“A static lead list tells you who a company is. It never tells you when to call them.”
CUFinder buying-signals documentation
The family runs wider than most people expect. Company-side events cover growth, decline, funding, identity shifts, and location moves. People-side events cover joins, departures, promotions, and executive hires, seventeen distinct signal types on the people side alone. And since every quantitative event also carries a magnitude bucket, signal types multiplied by magnitude levels give 1,000+ distinct signal variations to filter on. That’s why “we track hiring” undersells what this type can do.
And the honest limit, because every type gets one: a funding round doesn’t prove anyone wants YOUR product this week. Change signals time and contextualize the conversation; topic data confirms category research. They answer different questions, and I compare them head to head in buying signals vs intent data.
The 4 Types of Intent Data Side by Side
Here’s the whole map on one screen. Read it as a stack, not a menu.
| Type | What it captures | Where it comes from | Example | Blind spot |
|---|---|---|---|---|
| Search intent | What the market types into search | Search platforms, SEO tools | Spike in “email verification tool” queries | Anonymous; no account attached |
| Topic (research) intent | Accounts reading category content | Publisher co-ops, bidstream, review sites | Account surging on “deliverability” articles | Person unknown; topic is not product |
| Engagement intent | Behavior on your own properties | Your site, product, and email stack | Three pricing-page visits this week | Only sees accounts that found you |
| Firmographic-change signals | Events in the company’s reality | Snapshot diffs of public profiles | Headcount +13%, new sales leader | Change is not category research |
Notice how the blind spots interlock. Search sees demand without names. Topic sees names without people. Engagement sees people without reach. Change signals see reality without research. That’s why mature teams usually run engagement data plus ONE research type plus change signals, and route each to a different play.
One more reading of the table before you choose. The four types of intent data also differ in WHO should own them. Search belongs to marketing. Topic belongs to marketing and SDRs together. Engagement belongs to whoever can respond fastest. Change signals belong to whoever owns the account relationship. Ownership confusion, not data quality, kills most intent programs I’ve watched.
Which Type of Intent Data Should You Start With?
Start with the engagement data you already own, then add the type of intent data that covers your biggest blind spot. That’s the whole decision, and it usually plays out like this:
- Inbound-heavy team: you’re rich in engagement intent. Your gap is timing on accounts that never visit, so change signals add the most.
- Outbound-heavy team: your list is big and cold. Topic intent narrows WHERE to look; change signals decide WHEN to reach out.
- ABM program: named accounts make topic data sharper (you know who to watch), and change events give reps a reason to write this week.
- Tiny budget: honestly, skip contracts. Mine your own funnel and work the free intent data sources until the motion proves itself.
Whatever you pick, measure it by bucket. After our Hamburg whiteboard session split that one “intent” column into four, we re-ran the same plays with matched openers. Reply rates on the change-signal bucket alone went from 2% to 7% in a quarter. Same data. Different labels. And buyers won’t wait for you to figure it out: Harvard Business Review’s research on B2B buying shows committees run most of the journey before sales ever hears from them.
Keep collecting intent data examples from your own pipeline as you go, because your best training material is a closed-won deal with the evidence trail still attached. Which type flagged it first? That answer, repeated across ten deals, is your real vendor shortlist.
The formula I tape to my monitor:
→ Own data first → one paid blind-spot fix → route by type → measure reply rate per bucket → renew only what closes.
💡 Pro Tip: Measure reply rate PER TYPE for one full quarter before renewing any intent contract. One label column in your CRM is enough, and it will settle every vendor debate with numbers.
FAQ
What is intent data?
Intent data is evidence that an account may be preparing to buy. It comes in four types: search intent, topic (research) intent, engagement intent, and firmographic-change signals, and each type justifies a different outreach play.
What are the four types of intent?
It depends on which field is asking. In SEO, search intent splits into informational, navigational, commercial, and transactional. In B2B sales data, intent splits into the four types this page maps: search, topic, engagement, and change signals.
How is intent data collected?
Each type collects differently. Search data comes from query logs, topic data from publisher networks and ad-auction exhaust, engagement data from your own analytics, and change signals from snapshot diffs of public company profiles. The collection method is exactly what determines each type’s blind spot.
Who offers the best intent data?
No single vendor leads all four types. Topic co-ops lead research intent, review platforms lead declared category interest, your own stack owns engagement, and change-signal trackers own firmographic events. Decide the type first; the vendor shortlist writes itself afterward.
What are buyer intent signals?
Buyer intent signals are the individual behaviors and events inside intent data. Research-side examples: repeat pricing-page visits, a content-topic surge. Change-side examples: a new decision-maker hired, a funding announcement. Strong programs read both sides together.
What does high buyer intent mean?
High buyer intent means strong, recent evidence from more than one source at once. A pricing-page visit is interesting. A pricing-page visit at an account that just hired a new VP of Sales and popped a topic surge? That’s a call, today.
It’s Time to Stop Calling Everything “Intent”
Four buckets. Four kinds of evidence. Four different openers. That’s everything I know about the types of intent data after seven years of buying, mislabeling, and re-labeling them. And the trick costs nothing but a labeling habit.
Picture your dashboard with four labeled columns instead of one mystery feed, and every rep opening with the evidence they actually have. That team sounds prepared instead of creepy. And that team wins the reply.
Start small this week: relabel last month’s “intent” alerts into the four buckets, and count how many were sitting in the wrong one. My first count was 61 out of 90. Yours will surprise you too.
So tell me in the comments: which bucket does your team actually act on today, and which one have you been ignoring? I’ll trade you my routing sheet for your answer.
Bad
Okay
Good
Amazing
