Buyer intent data isn't working: how AI turns enriched data into real insights

Your enriched data doesn’t help you personalise at scale because buyer intent data gives you surface-level signals without the context to act on them. Most intent tools tell you a company is researching a topic. They don’t tell you which person to contact, what they care about, or what to say. That gap between signal and action is where personalisation breaks down.

AI-native enrichment closes that gap. Instead of appending fields (job title, company size, industry), it surfaces research-grade intelligence per contact: tenure, recent activity, certifications, strategic priorities, and tailored conversation starters. The difference between “this company is in-market” and “this specific person, in this specific role, cares about this specific problem, and here’s how to open the conversation.” That’s what makes AI data enrichment work as a complete approach rather than just another data source.

Half of all teams using intent data report too many false positives (Forrester ). And those are the teams that actually try to use the signals. Most buy intent data and never operationalise it at all.

The B2B intent data problem: 52% false positives and account-level guesswork

Most B2B intent data is built on three sources, and each one has accuracy problems. Content consumption tracking monitors publisher networks for topic-related reading. Bidstream advertising data captures ad interactions. IP-based web tracking identifies companies visiting web pages. All three tell you a company might be interested in something. None of them tell you who at that company is interested, or whether the interest is genuine.

The false positive problem is well documented. 52% of sales professionals report frequent false positives from intent data signals (Salesforce ). More than half the time, the “buying signal” isn’t real. Employees browse for professional development. Students research for coursework. Journalists investigate for articles. A content consumption spike at Acme Corp might mean their VP of Operations is evaluating tools, or it might mean an intern is writing a report. Your intent tool can’t tell the difference.

Then there’s the account-level problem. Most intent data resolves to companies, not individuals. “Acme Corp is researching CRM software” is useful directional information, but it doesn’t tell you whether to contact the CTO, the Head of Sales Ops, or the marketing coordinator who triggered the signal. You still need to figure out the who, the why, and the what-to-say.

And 70% of the buyer journey now happens in what’s called the dark funnel (Forrester ). Buyers research through AI tools like ChatGPT and Perplexity, through peer conversations, through private Slack channels and closed LinkedIn groups. None of that activity shows up in your intent data. By the time a prospect surfaces through a traditional intent signal, they may already have a shortlist that doesn’t include you.

I know what you’re thinking: “We’ve already invested in intent data, it should be working.” But having intent signals and being able to act on them are two completely different problems. Most teams buy signals that end up stranded in spreadsheets or disconnected tools. By the time a signal is exported, researched, and turned into an outreach sequence, the buying window has narrowed or closed.

Surface intent signals vs research-grade insight: where the gap really is

Intent signals aren’t useless. They’re incomplete. A bidstream signal tells you a company interacted with an ad. An IP tracking signal tells you someone at a company visited a web page. Neither tells you who, why, or what to do about it. The signal is a starting point, not an answer.

Traditional enrichment tools fill in fields on top of those signals: job title, company size, industry, direct dial, maybe some technographic data. Tools like ZoomInfo and Clay will give you these fields reliably enough (I’ve written about the best data enrichment tools in detail). But they frame the output as “intelligence” when it’s really structured data. Knowing someone is “VP of Marketing at a 500-person SaaS company” is a field, not an insight. It doesn’t tell you what to say to them.

AI-native enrichment surfaces what fields can’t capture:

• What a contact has been doing recently: Professional activity, publications, speaking engagements, job changes, new certifications. Real signals about what someone is focused on right now, not what their LinkedIn bio said when it was last updated.
• What their strategic priorities might be: Based on company announcements, hiring patterns, competitive moves, and sector trends. Not a guess, but a researched inference.
• What a relevant opening conversation looks like: Tailored to their specific role, demonstrated activity, and organisational context. Not “Hi {first_name}, I noticed you work at {company}.” Actual context.

The difference is between data enrichment and data intelligence. One fills in blanks on a record. The other generates understanding of a person. Just because an intent signal looks coherent doesn’t mean it’s good. A coherent “in-market” flag with a 52% false positive rate isn’t intelligence. It’s noise dressed up as a signal.

From fields to lead intelligence: what AI-powered enrichment actually surfaces

Lead intelligence built by AI looks different from a database export with extra columns. I’ve seen the difference first-hand building enrichment pipelines, and the gap between “more fields” and “actual intelligence” is larger than most people expect.

For a Fortune 500 enterprise software company with hundreds of technology partners, the AI enrichment layer we built produced the following per contact:

• LinkedIn profiles and tenure: How long someone has been in their role matters. New hires are often evaluating new tools and relationships. Long-tenured contacts have entrenched workflows and need different conversation angles.
• Recent professional activity: Certifications earned, content published, events attended. Not stale profile data, but signals about what someone is actively engaged with right now.
• AI-generated intent signals: Not bidstream data or IP tracking. Analysis that connects a person’s role, their company’s recent announcements, and sector trends to suggest what their priorities and pain points might be. Research-grade, not algorithmic guesswork.
• Tailored conversation starters: Summaries aligned to each person’s demonstrated activity and organisational context. The kind of opening that makes a recipient think “this person actually understands what I’m dealing with” rather than “this is another templated email.”

In plain English that means you stop sending the same message to 500 people and start having 500 different conversations. At scale. We explore how this kind of intelligence powers personalised email outreach in a separate guide.

The TCS ABM work followed the same principle. Deep account research surfaced strategic priorities for each target account, then aligned those priorities to specific propositions TCS could take to the table. The output wasn’t a generic capability deck. It was a tailored approach per account that demonstrated understanding of their specific business context. That’s what lead intelligence enables: conversations that start from a position of relevance, not cold outreach hoping something sticks.

Personalization at scale without an army of SDRs doing manual research

The personalisation bottleneck has always been research time. Writing a genuinely personalised message to one contact requires understanding their role, their company, their priorities, and what would be relevant to them. For one account, that’s a morning’s work. For fifty accounts with multiple contacts each, it’s weeks. For two hundred, you need a team. Or you need AI.

Two years ago, researching one account in proper detail took me a week. Tailoring messages to individual contacts within that account took another day on top. Now I can build a process that does the research in minutes and generates personalisation hooks per contact at scale. The quality of the output is genuinely comparable to (and in some cases better than) what I’d produce manually, because the AI can cross-reference more sources, faster, than any human.

Personalisation at scale requires three things working together:

1. Clean data underneath. If your records are full of duplicates, outdated contacts, and naming inconsistencies, personalisation just amplifies the errors. We’ve written about fixing data quality issues as the essential first step.
2. Enrichment that goes beyond fields to intelligence. Job titles and company size aren’t enough to personalise meaningfully. You need the context layer: what someone is doing, what they might care about, and what a relevant opening looks like.
3. A process that connects insight to outreach. If the intelligence lives in one system and the outreach happens in another, the gap between insight and action erodes the value. The process needs to be connected.

For SAP , we’re building AI-powered tools that help their partner marketing team engage partners at scale across a large ecosystem. The principle is the same: use AI to do the research and prep that humans can’t do manually at volume, so the human conversation that follows is informed and relevant. Not replacement of the human. Augmentation of the human’s ability to operate at a scale that wasn’t previously possible.

One caveat: if you haven’t cleaned your data first, AI-powered personalisation makes things worse, not better. You’re personalising with wrong information, which is worse than sending something generic. Clean first, enrich second, personalise third. That order matters.

What B2B personalization looks like when it’s built on real intelligence

Most “personalised” B2B outreach is templating with a veneer. {first_name}, {company}, maybe a line about their industry pulled from a firmographic field. That’s not B2B personalisation. That’s mail merge.

Real B2B personalisation, built on genuine lead intelligence, looks and performs differently. The EscherCloud AI ABM campaign is the clearest proof I can point to. 110 target prospects. Each one received tailored messages built on deep account research and enriched contact data. Not “Hi, I see you’re in the GPU space” but messages that referenced specific strategic priorities, demonstrated understanding of their business context, and opened with something genuinely relevant to the individual.

What made EscherCloud work wasn’t a better email tool or a cleverer subject line. It was the intelligence underneath. The enrichment gave us something specific and relevant to say to each person. The AI did the deep account research at speed. The human brought the strategic framing. The result was outreach that didn’t look or feel like mass email, because it wasn’t.

The market data backs up what practitioners already know. Personalised outreach achieves 15-25% response rates compared to 3-5% for generic approaches (Outreach ). And 75% of B2B buyers now expect tailored experiences (Salesforce ). The bar for “personalised” keeps rising. {first_name} and {company} don’t clear it anymore. Context, relevance, and demonstrated understanding do.

Not bad, hey? AI surfaced everything short of their star sign and favourite pizza topping, humans brought the strategic framing, and the result was conversations that converted. But those results only happen when the data underneath is right. Intelligence built on broken data is just confident-sounding noise.

Why buyer intent data needs AI-powered intelligence to deliver results

So why doesn’t your buyer intent data help you personalise at scale? Because intent data gives you signals without context. Account-level flags without individual relevance. A 52% false positive rate. And a 70% dark funnel that no intent tool can see. The problem isn’t that you don’t have data. The problem is that you don’t have intelligence.

AI-native enrichment turns surface-level signals into real lead intelligence: who to contact, what they care about, what they’ve been doing, and what to say. The companies getting results (39% open rates, 200% ROI from a single opportunity) aren’t using better email tools. They’re using better data. Clean data, enriched with genuine intelligence, activated through personalised outreach that demonstrates real understanding of the person on the other end.

For the full picture of how AI data enrichment works across the funnel, from cleaning to intelligence to activation, read our complete guide to AI data enrichment .

Owen Steer is Senior Marketing Executive at Fifty Five and Five, where he builds AI-driven tools and processes for B2B sales and marketing teams. If your intent data isn’t translating into personalised outreach that converts, get in touch .

Frequently asked questions