Intent Data vs Contact Data: What Each Proves, Where Each Fails, and How to Build a Record That Uses Both

Intent data and contact data answer different questions. Intent data answers why now and which account: it is behavioural evidence that a company, or occasionally a named person, is researching a problem you solve. Contact data answers who and how to reach them: a name, a role, a work email, a phone number. Neither substitutes for the other, and a sales record that is useful on the day it is used needs both, plus a verification date and enough context to write a first line. This page sets out what each proves, where each fails alone, and how to combine them.

Intent Data vs Contact Data: The Short Answer

• Different questions, different units. Intent data is mostly account-level and time-bound; contact data is person-level and fails one person at a time.
• Neither is self-verifying. An intent score is an inference from behaviour you did not observe directly; a contact record is a claim about a person that may have been true when it was collected. Both need a verification step before money or a rep's day is spent on them.
• Buy for your constraint. If reps reach the right people but at the wrong moment, the gap is intent. If reps know which accounts are moving but cannot reach anyone there, the gap is contact data. If both are missing, fix contact coverage first, because timing evidence you cannot act on decays while you look for a name.
• The useful record has four columns: account fit, reason to act now, reachable person, and proof (how and when each field was checked). Most data purchases fill one column and leave the record incomplete.

Start with the question you are asking, not the data type you were offered

Most "intent data vs contact data" comparisons are written by companies that sell one of the two, so the comparison ends where the product begins. A buyer's version starts with the question a rep is asking at the moment of use:

• Which accounts should we work this week? That is a prioritisation question. Intent data, firmographic fit and first-party engagement answer it; contact data does not.
• Who at this account should hear from us, and on what channel? That is a reach question. Contact data answers it; intent data can narrow the department or, in the contact-level variants, suggest a person, but it does not give you a deliverable address.
• What do we say first? That is a context question. Neither dataset answers it on its own. The reason to act (from the signal) plus the person's role and public footprint (from the contact record and whatever sits around it) produce the first line.
• Can we rely on this? That is a verification question, and it applies to both columns. A score without a stated method and a contact without a checked date are both unverified claims.

Sorting the question first prevents the two commonest purchasing mistakes: buying more contact records to fix a timing problem, and buying an intent feed to fix a reach problem.

What intent data proves, and what it only suggests

Intent data is a family of datasets, not one product. The field guide to B2B intent signals covers the families in detail; for this comparison the important split is how the behaviour reaches the vendor and at what level it is resolved.

• Third-party topic consumption. A cooperative or ad-exchange feed observes content reads across publisher sites, resolves them to a company (typically through IP address or cookie matching) and reports a surge in a topic relative to that company's baseline. What it proves: some people at some location associated with that company read about the topic more than usual. What it does not prove: who they were, whether they are the buyer, or whether the reading is a project or a student.
• Review-site and marketplace research. A software marketplace records category research, competitor comparisons, pricing-page visits and engagement with a vendor profile, and reports them at the account level; G2's own description of its Buyer Intent product lists exactly those activity types and frames the output as which accounts are likely to convert. What it proves: activity at the account on that site, of the kinds the marketplace tracks, within the dates it reports. What it does not prove: the person, the budget, or the timing beyond those dates.
• Public event signals. Executive hires, job postings, funding events, technology changes, procurement notices and public statements are observable directly from the source. What they prove depends on the event; a job posting is a dated statement of need by the buyer itself, a funding announcement is a dated public fact, and a procurement notice proves different things at each stage of the buying process. What they do not prove: that the need is for your category specifically, unless the wording says so.
• First-party engagement. Visits to your own site, form fills, product usage, event attendance. What it proves: a specific action by a specific visitor or, when resolved, a specific company. This is the only intent family that is not an inference about someone else's observation.
• Contact-level intent. Some vendors resolve signals to a named person, either because the person identified themselves (a webinar registration, a download behind a form, a review written under a profile) or through identity matching. When the person self-identified, contact-level intent is strong evidence about that person. When it is matched, it inherits the match's error rate, which most vendors do not publish; treat it as a lead to verify, not a fact.

Two properties hold across the family. Intent is time-bound: a research pattern is reported for a window and a job posting closes. And most intent is relative: a surge is measured against the account's own baseline, so a large company reading a little more can outrank a small company reading a lot. Both properties push the same way: intent data tells you when to look, and it needs a reachable person before the window closes.

What contact data proves, and how it goes wrong

Contact data is a set of claims about a person: name, title, employer, work email, direct dial, mobile, location, seniority, department, sometimes a LinkedIn URL. Its value is that it converts a decision to approach an account into a deliverable message. Its weaknesses are structural, not vendor-specific.

• It describes the past. Every field was true at collection time. People change jobs, titles and employers continuously: the US Bureau of Labor Statistics' Job Openings and Labor Turnover release for July 2026 (published September 1, 2026) reports 5.1 million total separations in the month, a rate of 3.2 percent, and 3.1 million quits, a rate of 1.9 percent. Every one of those separations changes at least one field on someone's record, so a contact list is a snapshot with a date, not a standing fact. The mechanics of that decay, and what it costs, are covered in how fast B2B contact data decays.
• It is only as good as its provenance. Records assembled from public professional profiles, email-pattern inference, crowd-sourced address books and partner feeds have different error profiles. A vendor that cannot tell you where a field came from cannot tell you how to verify it.
• Deliverability failures are punished by the receiving side. Google's Email sender guidelines instruct senders to keep spam rates reported in Postmaster Tools below 0.10 percent and to avoid ever reaching a spam rate of 0.30 percent or higher, require SPF or DKIM authentication for all senders, and require SPF, DKIM and DMARC for senders of more than 5,000 messages a day. Bad contact data does not just miss; it raises bounce and complaint rates on the domain every good message also uses.
• It is personal data, with the obligations that follow. Contact records identify people. Account-level intent avoids those obligations only to the extent that what you receive and process does not identify a person, which depends on the data and the processing rather than on the product's label. Under the GDPR, processing can rest on Article 6(1)(f), which permits processing that "is necessary for the purposes of the legitimate interests pursued by the controller or by a third party, except where such interests are overridden by the interests or fundamental rights and freedoms of the data subject", and Article 14 requires a controller that did not obtain the data from the person to provide the information listed in that article "within a reasonable period after obtaining the personal data, but at the latest within one month". In the United States, the FTC's compliance guide states that the CAN-SPAM Act "makes no exception for business-to-business email" and that a recipient's opt-out request must be honoured within 10 business days. California's Attorney General states that the CCPA exemptions for employment-related and business-to-business personal information "expired on December 31, 2022". None of this makes purchased contact data unusable; it makes provenance, notice and suppression handling part of the purchase, and your counsel's reading of your own situation controls.

Contact data does not tell you which of ten thousand valid records to use this week. That is the reach column doing the timing column's job, and it is why lists bought without a reason to act produce sequences that are technically deliverable and practically ignored.

Side by side: intent data vs contact data on the properties that change a decision

The table's last row is the point. A useful record is not one dataset with the other bolted on; it is a record with columns that different sources fill.

The four-column record: fit, reason, person, proof

The decision device on this page is a record structure rather than a scoring model. Before a rep spends time on an account, the record should hold four columns, and each column should name its source and its date.

1. Fit. Does the account match the profile you win with? Firmographic and technographic filters fill this column; the guide to combining intent data with ICP, firmographic and technographic filters covers how to set them without excluding the accounts you actually close.
2. Reason. Why this account, this week? An intent surge, a public event, a first-party action or a procurement record fills this column. The column holds the evidence, its date and its alternative explanations, not just a score.
3. Person. Who should hear from you, on what channel, and is the address deliverable? Contact data fills this column. Where a contact-level intent signal exists, it goes here as a lead to confirm, not as a confirmed person.
4. Proof. For every field in the other three columns: how was it checked, by whom or what, and when? A role confirmed against the employer's own site on a stated date is a different asset from a title copied from a two-year-old profile export.

Three rules follow from the structure:

• A record with an empty person column is a research task, not a lead. Route it to whoever builds contacts, with the reason attached so the person search is scoped to the right department.
• A record with an empty reason column is a list entry. It can be sequenced, but it should not consume a rep's personalisation time until a reason exists.
• A record with an empty proof column is a hypothesis. Do not report pipeline from it.

Lead Seeker's own approach maps onto the structure rather than replacing it. Its Trigger Signals monitor companies for public events such as executive hires, specific job postings, technology changes, funding events and public statements, which fill the reason column with a dated, citable event rather than a modelled surge; the prospect dossier fills the person and proof columns with the matching people, their verified work email and the date the role was verified. It does not sell third-party topic-surge feeds, so a team that needs publisher-cooperative intent adds that column from a specialist provider.

Which should you buy first? A decision by constraint

The right order depends on which failure your team has, and the failures leave different evidence.

Symptom: reps reach real people, but the conversations are early or irrelevant. Reply rates are acceptable, meeting rates are poor, and discovery calls end with "not a priority this year". The reach column works; the reason column is empty. Add intent or public-event signals and use them to sequence the list you already have. Do not buy more contacts.

Symptom: the team knows which accounts are moving and cannot get to anyone there. Signal dashboards are full, target accounts are named in pipeline reviews, and outreach bounces or lands on gatekeepers. The reason column works; the person column is empty or stale. Buy or build verified contacts for the accounts already flagged, in the departments the signal implicates, and record the verification date. Do not buy a second intent feed.

Symptom: both. No agreed account list, a contact database nobody trusts, and reps prospecting from memory. Fix contact coverage for a small, well-defined segment first, because timing evidence you cannot act on within its window is wasted spend, then add reason signals scoped to that same segment. Widen only when the record is complete for the segment.

Symptom: neither, but forecasts are wrong. Records look complete and deals still slip. The proof column is the problem: signals that were never corroborated and contacts that were never re-verified are being reported as pipeline. Add verification dates and a re-check cadence before adding any data.

A cost note that does not depend on vendor prices: whatever each costs you, a contact record is only valuable on the day a reason exists, and an intent signal is only valuable while a person can be reached. Budget for the pair, and for the verification step that makes either one reportable.

Contact-level intent: the overlap, and the questions to ask before paying for it

The phrase "contact-level intent data" sits exactly on the boundary between the two datasets, which is why it is marketed hard. It is worth separating what it can be:

• Self-identified behaviour. The person filled a form, registered for an event, wrote a review under a profile or engaged with content in a logged-in state. This is both intent and contact data at once, and it is the strongest object in the category, subject to the consent and notice terms under which it was collected.
• Matched behaviour. A signal observed at the account or device level was attributed to a named person through identity resolution. The attribution has an error rate. Ask the vendor for its method and for the rate; if neither is available, treat the name as a starting point for verification.
• Inferred relevance. No behaviour by the person was observed at all; the account surged and the vendor listed the people whose titles usually buy. This is contact data with a topic label attached. It is useful for scoping a person search and should not be described internally as intent from that person.

Whatever the variant, the record structure holds: the behaviour goes in the reason column with its source, the person goes in the person column, and the attribution method and date go in the proof column.

The workflow: from signal to first message with both columns filled

A workflow that keeps the columns separate until they are verified avoids the two classic failures, and it runs in the same order whether the team is three people or thirty.

1. Fix the segment. Write the fit filters down, including the exclusions. Every later step is scoped to this segment.
2. Log reasons as they arrive. Each intent surge, public event or first-party action becomes a dated entry against an account with the source, the evidence, the alternative explanations and a decay date. A weekly review retires expired reasons instead of letting them accumulate.
3. Build the person list from the reason, not the whole account. Use the signal's wording to pick the department and seniority: a posting for a data-platform engineer implicates the data leader, not the CFO. The article on building a prospecting list that combines ICP fit with verified intent signals covers the list mechanics.
4. Verify before sequencing. Confirm the role against the employer's own site or a recent primary source, validate the mailbox, and record the date and method in the proof column. Suppress anyone on your opt-out list. Records that fail verification go back to step 3 rather than into a sequence.
5. Write the context line from the record. The first sentence names the reason (the dated event or research pattern) and the person's role in it, without claiming knowledge of the person's private behaviour. "You posted for a data-platform engineer last week" is evidence the buyer published; "you have been reading about data platforms" is a claim about surveillance the recipient did not consent to hear about, even where the data was lawfully collected.
6. Act inside the window. A research pattern is reported for a period; a job posting closes; a funding announcement stops being news. Set a service level from reason logged to first verified message and measure it.
7. Feed outcomes back into the proof column. Bounces, wrong-person replies and "no longer here" messages update the contact record; "not relevant" replies update the reason's weighting for that signal type. Over a quarter this produces the only evidence that matters for the next purchase decision: which signal types and which contact sources produced conversations in your segment.

For teams that outsource part of this, the division of labour matters more than the tool. Agencies that run signal-based programmes, including Percepture's B2B intent data service, typically own steps 2 and 5 and hand steps 3 and 4 to a data provider or the client's own operations team; ask any agency which columns of the record it will fill and which it expects you to fill. Disclosure: the author of this article is President of Percepture, so treat that link as context from a related party rather than an independent recommendation.

A worked example, with synthetic data

The example below is constructed to show the record, not drawn from a real account or person.

Account: a 400-person logistics software company. Fit: inside the segment (mid-market, North America, sells to shippers). Reason: on September 15 the company posted a "Head of Revenue Operations" role whose description names a CRM migration and a new outbound motion; on September 18 an account-level review-site feed reported comparison activity in the sales-intelligence category. Two independent public and third-party sources, three days apart, on the same theme. Alternative explanations logged: the posting could be a backfill; the comparison activity could be an analyst or a competitor. Decay: the posting will close, so its date is logged as an expiry to re-check; the review-site activity is dated and will not stay current.

Person: the current VP of Sales, whose role was confirmed on the company's leadership page on September 19 and whose work email validated the same day, and the hiring manager named in the posting. The Head of Revenue Operations role is empty by definition; it is noted as a follow-up, to be re-checked when the hire is announced. Proof: each field carries its source and date. Context line: the posting, quoted, and a question about what the outbound motion is meant to replace. The review-site activity is not mentioned in the message; it informed the priority, not the copy.

Outcome logged: the VP replied, the hiring manager did not, and the reply cited the migration timeline. The record's reason column is now corroborated by the buyer's own words, and the proof column shows every field's last check within two weeks of the message.

The buyer's checklist before any data purchase

Copy this into the evaluation for either dataset. A vendor that cannot answer a question is telling you which column of the record you will still have to fill yourself.

• Which of the four columns does this product fill, and which does it leave empty?
• For intent: what is observed, how is it resolved to an account or a person, over what window, against what baseline, and is the method documented in writing?
• For contact data: where does each field come from, when was it last checked, is the verification date exposed per field, and how are opt-outs and suppression lists honoured?
• What is the vendor's own measured error rate for the claim it is selling, and how was it measured?
• What happens to a record when it fails verification: credit, replacement, or nothing?
• Can the data be exported with its provenance fields intact, so the proof column survives a tool change?
• Which legal basis and notice practice does the vendor rely on for personal data, in which jurisdictions, and what does that require from you as the recipient?
• Does the contract allow you to combine the data with other sources in your own record, or does it restrict use to the vendor's interface?

Who wrote this and how it was checked

Bob Generale is President of Percepture and works across SEO, AI search, digital PR, sales intelligence and AI-powered revenue systems. The guide was researched on September 25, 2026 from the primary sources listed below; the regulatory sentences restate the fetched text of the GDPR articles, the FTC's CAN-SPAM compliance guide and the California Attorney General's CCPA page as shown on that date, and the labour-turnover figures are from the Bureau of Labor Statistics release named in the Sources. Vendor pages ranking for this query were reviewed for what they claim and were not used as sources for any statistic, definition or rule; their reply-rate, conversion and return-on-investment claims were deliberately not repeated. No vendor named or linked here was contacted in the course of writing, and the worked example is synthetic.

Frequently Asked Questions

Is intent data or contact data more important?

Neither is more important in general; each is decisive for a different failure. If reps reach the right people at the wrong time, intent is the missing column. If reps know which accounts are moving and cannot reach anyone there, contact data is the missing column. Because timing evidence expires while a person search runs, a team that lacks both should establish verified contact coverage for one narrow segment first, then add reason signals scoped to that segment.

Can you use intent data without contact data?

Yes, for prioritisation and for channels that do not need a named person: advertising to an account, routing inbound leads, alerting an existing account owner, or deciding which accounts a rep researches first. You cannot use it for direct outreach without a reachable person, and account-level intent does not supply one. Contact-level intent supplies a name whose attribution method you should check before treating it as confirmed.

What is contact-level intent data?

It is intent data attributed to a named person rather than to an account. It comes in three forms with very different reliability: behaviour the person performed while identified, such as a form fill or a review under a profile; behaviour observed at the account or device level and matched to a person through identity resolution, which carries the match's error rate; and lists of likely buyers at an account that surged, where no behaviour by the individual was observed. Ask a vendor which form it sells before comparing prices.

How do you verify contact data before using it?

Confirm the role against the employer's own site or another recent primary source, validate that the mailbox accepts mail, check the record against your suppression and opt-out lists, and record the method and date on the record itself. A verification date per field is the test of whether the check happened; a vendor-level accuracy percentage is not. Records that fail go back to research rather than into a sequence.

Does buying contact data create legal obligations that intent data does not?

Contact records identify people, so personal-data obligations apply to them. Whether intent data carries the same obligations depends on whether what you receive and process identifies a person, not on the label "account-level": the fetched legal texts define obligations by reference to personal data and the data subject, not by data-product category. Under the GDPR, a lawful basis must apply; Article 6(1)(f) covers processing necessary for the legitimate interests of the controller or a third party, except where those interests are overridden by the interests or fundamental rights and freedoms of the data subject, and Article 14 requires a controller that did not obtain the data from the person to provide the listed information within a reasonable period and at the latest within one month. The FTC states that CAN-SPAM makes no exception for business-to-business email and that opt-outs must be honoured within 10 business days, and California's Attorney General states that the CCPA's business-to-business exemption expired on December 31, 2022. Contact-level intent inherits the same obligations. How they apply to your programme is a question for your counsel.

How quickly do intent data and contact data go stale?

Intent is reported for a window, and public events stop being current, so each logged reason needs an expiry date. Contact data has no single decay rate; it fails one person at a time as people change roles and employers, and those changes are frequent: the Bureau of Labor Statistics reported 5.1 million total separations in July 2026, a monthly rate of 3.2 percent. That is why a record's own verification date matters more than a vendor's headline accuracy figure. Set a re-verification cadence for contacts as an operating decision, and measure it against your own bounce and wrong-person rates.

Sources

• U.S. Bureau of Labor Statistics, Job Openings and Labor Turnover – July 2026, news release, September 1, 2026 (total separations 5.1 million, 3.2 percent; quits 3.1 million, 1.9 percent; hires 5.1 million, 3.2 percent).
• Google Workspace Admin Help, Email sender guidelines, as shown September 25, 2026 (spam-rate thresholds of 0.10 and 0.30 percent; SPF or DKIM for all senders; SPF, DKIM and DMARC for senders of more than 5,000 messages a day).
• Regulation (EU) 2016/679 (GDPR), Article 6 – Lawfulness of processing and Article 14 – Information to be provided where personal data have not been obtained from the data subject, text as shown on gdpr-info.eu, September 25, 2026.
• Federal Trade Commission, CAN-SPAM Act: A Compliance Guide for Business, as shown September 25, 2026.
• California Office of the Attorney General, California Consumer Privacy Act (CCPA), frequently asked questions, as shown September 25, 2026 (expiry of the employment and business-to-business exemptions, Civil Code § 1798.145(m)–(n)).
• G2, Buyer Intent, product page describing the tracked activities and account-level reporting, as shown September 25, 2026 (the page's former buyer-intent address now redirects to this one).
• Lead Seeker, Product: Trigger Signals and the Prospect Dossier, for the description of Lead Seeker's own signal coverage and verification fields.

Next Steps

Put the four columns on your current account list and count how many records have all four filled with a date. That number, not the size of either dataset, is the pipeline you can actually work this week. For the signal side of the record, start with how to read Trigger Signals; for the person side, the lead intelligence insights hub covers list building, verification and decay in depth.