Buyer intent data: what actually predicts a deal

A 180-person API monitoring company sees Northstar Payments visit its pricing page, read three articles, and download a migration guide. The answer to buyer intent data: what actually predicts a deal is not “more activity.” It’s a credible business trigger, the right account, the right people, and behavior that moves toward a buying conversation.

That sounds obvious until you look at how most intent programs work. Someone visits a page, an account score jumps, and an SDR gets told to “work the signal.” The result is usually a bad email sent to a person who was researching a problem, not buying a product.

The useful answer is narrower: intent data predicts a deal when it shows fit, timing, multiple stakeholders, and movement from learning to evaluation.

Buyer intent data: what actually predicts a deal?

Start with the reason the account might buy. Then check whether the people involved are doing work that supports that reason.

For Northstar, a new VP of Platform Engineering would be interesting but not enough. Reliability hiring adds pressure. A public audit document that mentions gaps in service monitoring adds a deadline. Visits from an engineering manager, a platform engineer, and someone in security suggest the issue is spreading across the buying group.

Then the behavior gets more serious. The team looks at OpenTelemetry documentation, deployment options, data retention, and access controls. Those aren't random content clicks. They’re questions a team asks when it’s checking whether a product can survive contact with its systems, security review, and procurement process.

That combination is much more useful than a score of 80.

The sales message should reflect what you actually know. A short note about monitoring evidence for payment APIs is reasonable. Asking whether the audit work has changed Northstar’s requirements is reasonable. Saying “I saw you looking at our pricing page” is not. It makes the company sound watched, not understood.

The signals teams overrate

Third-party topic activity is good for finding accounts to investigate. It’s poor evidence that an opportunity exists.

A company researching “API monitoring” might be comparing vendors. It might also have an engineer solving a one-off problem, a consultant preparing a report, or a new hire trying to understand the category. The data can tell you that a topic is present. It usually can’t tell you who cares, why now, or whether there’s money attached to the problem.

First-party activity gets overtrusted too. A webinar registration can be casual. A whitepaper download can come from a junior employee. A run of page views can be one person doing research for an interview. Pricing-page visits are especially noisy on sites with poor navigation.

None of that makes the activity worthless. It makes it incomplete.

My view is that teams get this wrong by treating observation as evidence. They build a lead score that rewards every click, then act surprised when sales rejects the resulting accounts. Intent data should help decide where to investigate. It should not manufacture urgency where none exists.

For Northstar, the page views only become meaningful after the external trigger and stakeholder pattern appear. Before that, the account belongs in a watch list or a light education motion. Not an SDR sequence.

Four things that make intent more credible

Fit comes first

The account needs to operate in conditions where the product solves an expensive problem. For the API monitoring company, that might mean 200 to 2,000 employees, a substantial API surface, a platform or SRE team, and compliance work that makes incident evidence painful to produce.

This is where a clear ideal customer profile earns its keep. A small software company can show intense interest and still be a poor commercial fit. Northstar has the scale, technical complexity, and regulatory pressure to make a monitoring change plausible.

Without fit, intent is just activity wearing a suit.

A trigger explains the timing

A trigger changes priorities or exposes a cost. It could be an audit finding, a new platform leader, a processor migration, a reliability incident, or hiring tied to infrastructure pressure.

The closer the trigger is to the product’s problem, the better. A funding announcement is weak on its own. Funding followed by platform hiring is more useful. A new CTO is a clue. A new CTO who has spoken publicly about reliability targets is a much better reason to look closer.

You don’t need a dramatic event. You need a believable reason for the account to spend time and money now.

More than one relevant person gets involved

One person’s research is a clue. Related activity from engineering, security, operations, and finance suggests the problem may be moving through the business.

Those people will not ask the same question. Engineering wants to know whether the product works. Security wants to know what data it touches and how access is controlled. Finance or procurement wants to know what the contract and implementation burden look like.

That’s where account-based marketing becomes practical. The job isn’t to contact everyone at Northstar. It’s to coordinate a few relevant conversations around the same business issue, with different proof for each person.

The behavior progresses

Early research is not the same as evaluation.

A category article shows that someone is learning. Integration documentation, deployment comparisons, implementation questions, security reviews, pricing conversations, and procurement requests show that the account is testing whether the product can fit.

Track the direction of the activity, not just the volume. Five beginner articles can be weaker than one technical validation request. A pricing visit followed by a security document download tells you more than ten unrelated page views.

Also watch for negative signals. A high-fit account that just signed with a competitor shouldn’t receive the same treatment as one asking for a technical validation call. Silence after repeated, relevant engagement may mean the project stopped, not that the account is secretly “warming up.”

How to test buyer intent data without fooling yourself

Use the last four quarters of closed-won, closed-lost, and stalled opportunities. Look at what was visible before each opportunity was created. Not what became obvious afterward.

Compare accounts with only topic activity against accounts with a verified trigger, several engaged stakeholders, and evaluation behavior. Measure opportunity creation, pipeline per 100 target accounts, time from signal to opportunity, win rate, and sales rejection. Replies can be included, but don’t make them the headline metric. A campaign can get plenty of replies from people with no authority, no budget, and no active project.

Then run a controlled test. Put Northstar-like accounts with fit, trigger, and progression into a sales motion. Keep similar accounts with only third-party topic activity in a lighter education track. Give both groups a defined period and the same basic account coverage. Otherwise, you won’t know whether the signal worked or a rep simply worked one list harder.

The test should also record why sales rejected an account. “No project,” “wrong person,” “bad fit,” and “already committed elsewhere” are more useful than a vague quality score. After a few cycles, you’ll know which signals deserve attention and which ones are just producing busywork.

Intent platforms will keep improving at spotting anonymous research. That’s useful. It helps teams decide where to look first.

But a deal still needs a business reason, people who care about it, and evidence that the account is willing to do the work of buying. Use the data to choose the next account to investigate. Don’t use it as permission to pretend a deal already exists.