Intent based targeting: how to turn a buying signal into a targeting rule, a play, and an honest measurement.

What intent based targeting actually is

Intent based targeting is the practice of deciding who to reach next from what people did, rather than from what their company looks like on paper. Firmographics answer who could buy. Behavior answers who is looking right now, or at least who is looking at something.

The word "targeting" is the important half. Plenty of teams buy signals, stare at a dashboard of surging accounts, and change nothing about who gets an ad or an email. That is data collection, not targeting.

Targeting starts when a signal becomes a rule with three fixed parts: what qualifies, how fresh it has to be, and who you contact because of it. A rule can be run, audited and switched off. A hunch cannot.

This page is about writing those rules. If you want the data side first, read our entry on B2B intent data, which covers where signals come from and what they cost you in accuracy.

Signal, rule, play: the three things people mix up

A signal is an observation. A rule is a written decision about which observations count. An audience is the list the rule produces. A play is the single thing you do to that audience.

Most programs fail between the signal and the rule. Someone shares an interesting behavior in a meeting, everybody agrees it means something, and nobody writes down the threshold. Six weeks later nobody can say whether it worked.

First party signals you already own

The best signals are the ones you observed yourself, on your own properties, about a named account. They are free, they are current, and you know exactly what the behavior was. Start here before buying anything.

Notice the pattern. Signals get stronger as the behavior gets more specific, more repeated, and harder to do by accident. A rule built on "visited the site" catches everyone. A rule built on "opened the security page twice" catches a real evaluation.

Your CRM holds first party signals too: a closed lost deal with "no budget this year" written twelve months ago is a dated but real signal. So is a champion who changed jobs, which is one of the cleanest triggers in B2B.

Intent based targeting, intent based marketing and lead scoring

Three terms get used for the same intent data, and the difference is scope. Intent based marketing is the whole strategy: the content, the campaigns, the budget and the sales motion built around buying signals. Intent based targeting is the narrow decision inside it about who each campaign reaches.

The practical difference between targeting and lead scoring is direction. Scoring accumulates over the life of a lead and answers a ranking question. A targeting rule is a gate with an expiry date that either fires today or does not.

They coexist well. Scoring decides which leads sales works through in order; intent based targeting decides which accounts jump the queue this week because their buying behavior changed. Both feed the same ABM plays and the same marketing calendar.

One term is worth retiring from internal conversation: "high intent lead". It hides all three clauses of a rule. Say which signal, how fresh, and which person, and the disagreements in the room turn out to be about definitions rather than about strategy.

Third party intent, and what you are really buying

Third party intent is research behavior observed somewhere other than your site, then attributed to a company and scored against a topic. Providers build it from publisher networks, review and comparison sites, co-op arrangements between vendors, and advertising traffic.

Two things follow from how it is built, and both matter for your rules. First, it is usually modeled at the account level, not the person level, so it tells you a company is active without telling you who. Second, the match from activity to company can be wrong.

• Topic, not product: you are buying activity against a topic taxonomy that the provider defined, which may not map cleanly to what you sell.
• Baseline, not volume: most scores are a surge against that account's own normal level, so a quiet account can surge on very little activity.
• Attribution guesswork: matching traffic to a company relies on network and device signals, and remote work has made that harder, not easier.
• Latency: ask how old the underlying activity is by the time it reaches your CRM, because a weekly refresh changes what freshness window is possible.
• Overlap: the same account often surges for every vendor in the category at once, so the signal is rarely yours alone.

None of this makes third party intent useless. It makes it a prioritization input rather than a reason to act. Our entry on bidstream data covers one common source in more detail.

Ad platform intent options: what LinkedIn and Google document

Ad platforms sell intent targeting under their own names, and the documentation is more precise than the sales language around it. Here is what the official help pages actually say.

Two platform limits shape your rules directly. LinkedIn states that an ad set needs at least 300 member accounts, and suggests a minimum of 50,000 to drive results, with larger suggested minimums for Sponsored Content and Sponsored Messaging.

Google states that a data segment needs at least 100 active visitors or users within the last 30 days to serve on Display, Search or YouTube. A tight rule that qualifies twelve accounts a month cannot be an ad audience at all, so it has to be a sales play instead.

LinkedIn also documents retargeting lookback windows running from 30 to 365 days, and notes that the size of Matched Audiences may be limited within the European Economic Area and Switzerland. Both facts belong in your freshness and coverage math before you promise anyone a number.

Why intent targeting b2b needs two rules, not one

Consumer intent targeting points at a person who will decide alone. Intent targeting b2b almost never does. The research is split across colleagues who never mention it to each other, and the person who reads your documentation is often not the person who signs.

So a B2B rule has two halves that have to be written separately:

1. The account qualifier: what behavior, from how many people, in what window, makes this company worth attention.
2. The contact rule: which roles at that account you approach, in what order, with which message.

Collapsing the two is the most common mistake in the whole discipline. An account surges, somebody exports every contact, and eleven people get the same email on the same morning. The account now associates your brand with spam, and the signal is burned.

Keeping them separate also lets fit do its job. A signal from an account that does not match your ideal customer profile is noise you can drop before it costs anyone time.

How to write a targeting rule from a signal

Write rules as one sentence with three clauses. If you cannot fit it into one sentence, the rule is doing two jobs and should be split.

The fit gate deserves special attention. Putting fit first turns a noisy signal into a workable one, because most false positives come from accounts that were never going to buy anyway. Your target account list is the cheapest filter you own.

Write the rule where other people can read it. A rule that lives only inside an automation tool gets forgotten, duplicated and contradicted. A one page table of active rules, with owners and dates, prevents most of that.

How fresh is fresh enough

Freshness is the clause teams skip, and it is the one that decides whether the play lands. A signal has a half life, and the half life depends on the kind of behavior, not on how excited you are about it.

Decide what expiry means. Some rules should drop the account entirely. Others should demote it to a lower tier that still receives ads but no sales touch. Write the answer down, because otherwise expired signals quietly accumulate in the same queue as fresh ones.

Who to contact when the signal belongs to an account

Account level signals do not name anyone. Your contact rule has to, and guessing badly is worse than not acting.

• Start with the known person. If a named contact triggered the signal, they are first, and the message continues what they were doing.
• Add the problem owner. The role that lives with the problem daily, whether or not they control budget.
• Add the economic buyer later. Usually a level up, contacted with a different message about cost or risk, not about features.
• Cap the account. Two or three people in the first week is plenty; more reads as a list blast to everyone who receives it.
• Vary the message per role. If two people forward your emails to each other and see the same paragraph, you lose both.

This is ordinary multithreading, just triggered by a signal instead of by a calendar. The signal changes the timing and the topic. It does not license a bigger blast.

How to build an intent based targeting rule

1. Pick one problem, not one product

   Name the business problem you want to catch people researching, because signals cluster around problems and your product name catches almost nobody.

2. List the signals you can actually see

   Write down every signal you own or can buy for that problem, and mark which are observed behavior and which are inferred or modeled.

3. Set the qualifying bar

   Decide what counts: which signal, how many times, by how many people, and whether company fit has to be met before the signal counts at all.

4. Set the freshness window

   Give the rule an expiry, short for a page visit and longer for a hiring or funding signal, and write down what happens when it expires.

5. Name the audience the rule produces

   Decide whether the rule targets the person who acted, the account, or a role at the account, and route it to a named owner.

6. Attach one play and one message

   Bind the rule to a single play with a first touch that stands on its own without mentioning the tracking.

7. Run a holdout and review monthly

   Hold back a random share of qualifying accounts, compare them with the targeted group, and retire rules that do not beat the holdout.

Scoring how much evidence a signal carries

Not every qualifying signal deserves the same play. Score the evidence, then match the effort to the score. The scale below was written for this page as a starting point, not as a standard.

Read the table in one direction: observed and repeated behavior earns a person, inference earns an impression. Teams get into trouble by sending a personal message on the strength of a modeled score, which is where "how did you know that?" comes from.

The plays a rule can trigger

One rule, one play. Rules that fan out into three plays cannot be measured, because you never learn which part did the work. If you want a different play, write a different rule with a different bar.

Content and campaigns the rule points at

A targeting rule is only as good as the thing waiting at the other end. Intent based marketing campaigns fail most often because the signal was precise and the landing content was a generic product page written for everyone.

Match the content to what the buying behavior implied, not to the stage of the customer journey you wish they were in.

Two rules of thumb keep this manageable. Write the content before you switch the targeting rule on, and reuse one strong piece across several rules rather than commissioning a new asset for every signal you discover.

The same content also carries your ad campaigns. Because ad platforms need audience volume, the broad campaigns run on inferred segments while the sharp rules drive sales outreach, and both point at the same small library of genuinely useful pages.

Sequencing after a signal

The sequence after a signal is shorter and narrower than an ordinary cadence, because the signal already told you the topic. What it did not tell you is whether there is a project, so the first touch has to leave room for "no".

If the signal expires mid sequence and nothing new arrives, stop. Continuing turns an intent play into ordinary cold outreach, which is a legitimate thing to do, but it should be a separate decision made on purpose. Our sales cadence guide covers that case.

Behavior during the sequence outranks the schedule. A reply, a second pricing visit or a new colleague appearing should change the next step immediately. That is the same logic as trigger marketing, applied to a signal you are already acting on.

False positives you will definitely hit

Every intent program spends part of its budget on people who were never buying. You cannot remove that cost, but you can size it and stop paying for the obvious cases.

• Competitors researching your pricing and features, often more thoroughly than customers do.
• Job applicants reading everything before an interview, which looks identical to an evaluation.
• Analysts, consultants and students writing about the category rather than buying in it.
• Existing customers who show up as new accounts when their identifiers do not match.
• Your own staff and agencies, whose traffic is trivially easy to exclude and frequently is not.
• Shared networks, where a coworking space or a university maps to one company.
• Bots and previews, including link scanners and mail clients prefetching images and URLs.

Two cheap habits cut a lot of this. Exclude your own and your partners' traffic at the source, and require a second independent signal before any rule triggers a human. One anonymous visit should never reach a sales rep.

Privacy, consent and what the rules actually require

Intent based targeting collects behavior about people, so the law follows it. Two separate things apply, and teams routinely confuse them: the rules about storing and reading data on someone's device, and the rules about processing personal data.

On the first, the Information Commissioner's Office is explicit that online advertising purposes are not exempt from the consent requirements under PECR and never have been. Its guidance lists exceptions such as strictly necessary and statistical purposes, and advertising is not one of them.

The ICO also describes what a valid request looks like: consent must be specific to the purpose, supported by clear and comprehensive information, and involve a positive action, since silence or inactivity does not qualify. Users must be able to withdraw with the same ease they gave it.

On the second, the GDPR defines personal data as any information relating to an identified or identifiable natural person. It defines profiling as automated processing used to evaluate personal aspects, including interests, behavior and location, which describes intent scoring precisely.

Recital 47 says processing for direct marketing purposes may be regarded as carried out for a legitimate interest. Article 21 then gives the data subject the right to object at any time, including to related profiling, and states that the data shall no longer be processed for those purposes.

Suppression lists and the exclusions nobody writes down

A targeting rule is half inclusion and half exclusion, and the exclusion half is usually improvised. Write it once and reuse it across every rule.

• Current customers, unless the rule is deliberately for expansion.
• Open opportunities, so marketing does not walk into a live negotiation.
• Anyone who objected or unsubscribed, across every channel, not just email.
• Accounts contacted in the last N weeks, to stop several rules hitting the same people.
• Your own domains, partners and agencies.
• Accounts a rep already disqualified, with the reason and a date when they become eligible again.

The fourth one matters more than it looks. Once you run several rules, accounts qualify for multiple rules at once, and the prospect experiences the sum. A global contact cap per account per month fixes it, and belongs in your lead routing logic.

Measuring whether intent targeting pays

This is the part most intent programs never do honestly, and the reason is uncomfortable: intent targeting selects accounts that were already more likely to buy, so it looks brilliant no matter how good your rules are.

Comparing targeted accounts with the rest of your database proves nothing. The accounts that qualified are different accounts. The only clean answer is a holdout.

1. Define the rule and let it qualify accounts as normal.
2. Randomly hold back a fixed share of qualifying accounts and do nothing to them, big enough to read, small enough to afford.
3. Run the play on the rest.
4. After one full sales cycle, compare meetings, opportunities and pipeline between the two groups.
5. Keep the rule if the targeted group beats the holdout by enough to cover the cost of the data and the effort.

Measure per rule, not per program. An "intent program" that pays for itself often contains two rules that work and five that quietly do not. Only per rule numbers tell you which ones to retire.

Track the operational numbers too: how many accounts a rule qualifies per week, how many were acted on inside the freshness window, and how many were later marked as false positives. A rule nobody has capacity to work is a broken rule. Our guide on how to measure ABM covers the account level math.

The tool categories involved

This page does not rank vendors or quote prices. These are the categories a working setup needs:

• Web analytics and visitor identification: turns anonymous sessions into account level signals on your own property.
• Intent data providers: supply third party topic activity, scored per account.
• Enrichment: fills in firmographics and technographics so the fit gate can run.
• Marketing automation and CRM: hold the rules, the suppression lists and the record of who was contacted.
• Ad platforms: activate audiences, within the documented minimum sizes.
• A shared rules document: unglamorous, and the thing that actually keeps the program coherent.

Common intent based targeting mistakes

• Buying a signal feed before writing a single rule for it.
• No freshness window, so a two month old page visit triggers an urgent call.
• Mailing everyone at an account because the account, not a person, surged.
• Quoting the signal back to the prospect in the first sentence.
• Skipping the fit gate, so the rule spends its budget on accounts that cannot buy.
• Treating an inferred interest category as evidence of an active project.
• Measuring without a holdout, then congratulating the program for picking good accounts.
• Letting rules accumulate, so one prospect is hit by four of them at once.
• Running advertising storage without consent and calling it strictly necessary.

The first touch after a rule fires

The single fastest way to ruin intent based targeting is to tell the prospect what you saw. "I noticed you visited our pricing page three times" is accurate, legal in many places, and still lands like a stranger describing your living room.

Write about the problem instead. The signal earns you the right to guess the topic and the timing. It does not earn you the right to narrate their browsing.

• Lead with the problem in the words their team uses, not your product category.
• Say why now in general terms: what changed in their market or their stage, not what changed in your dashboard.
• Offer one specific thing that is useful even if they never reply.
• Make "no" cheap with an explicit line inviting it, and then honor it.
• Keep it short enough that a wrong guess costs them ten seconds.

The template below is for the moment a fitting account crosses your qualifying bar and you have a named person worth writing to. It names the problem, gives them something specific, and makes refusing easy.

Subject: {{problem}} at {{companyName}}

Hi {{firstName}},

Teams in {{industry}} are rethinking {{problem}} this year, usually when {{trigger}} forces the question. I do not know whether that is on your desk.

If it is, here is the part most {{role}} teams get stuck on: {{specificDetail}}. We wrote up how three teams handled it: {{link}}

If it is not, tell me and I will stop.

{{senderName}}

Frequently asked questions