Join the waitlist

Let us know how we should get in touch with you.

Thank you for your interest! We’re excited to show you what we’re building very soon.

Close
Oops! Something went wrong while submitting the form.

How to Use Intent Data to Find Warm Accounts (Workflow)

Austin Hughes
·
Updated on: November 18, 2026
TL;DR: Build a warm-account shortlist by applying fit, signal meaning, reachability, and recency filters after checking ownership and outreach eligibility. Keep the evidence behind each decision. The result is a prioritized set of accounts worth investigating or contacting, not proof that a named person is ready to buy.

Intent data gives you candidates. Your workflow must decide which candidates merit a sales action, which need research, and which should remain out of outreach. Start with the difference between first-party and third-party signals so a website event and an external topic signal are not treated as interchangeable evidence.

Subtraction is useful when it removes irrelevant or ineligible work. It becomes counterproductive when the team optimizes for a smaller list without checking what was lost. Keep a reason for each exclusion and inspect the rejected population alongside the shortlist.

Warm-account filtering decisions
Gate or filterWhat it removesPractical ruleRisk if skipped
Eligibility gateAccounts the team should not contact nowCheck ownership, suppression, open work, and channel permissionConflicting or prohibited outreach
FitAccounts outside the offer’s plausible use caseRequire a relevant business problem and supported account criteriaA strong signal distracts from poor fit
Signal meaningActivity with no defensible connection to the buying questionPreserve source, entity level, and observed eventActivity becomes an invented buying claim
ReachabilityAccounts without an appropriate contact pathIdentify the role and validate the chosen channelThe right company receives outreach through the wrong person
RecencyEvidence that no longer supports the planned actionEvaluate the underlying event and current account contextOld interest is presented as current urgency

What must happen before the four filters?

Check whether the team is allowed and expected to work the account before scoring its attractiveness. Preserve existing customer relationships, active opportunities, opt-outs, promised callbacks, and another rep’s ownership. An account may be interesting while the correct action is to notify its current owner.

In Unify, engagement policies control claiming and contacting separately and require an appropriately configured CRM connection with bidirectional sync. For assigned users, the policy’s exclusions replace sequence-ruleset exclusions; other policy restrictions still apply, with the most restrictive result governing the action. Review that configuration before assuming a sequence’s old exclusions still protect the team.

Also distinguish an account that fails a policy from one with missing evidence. “Do not contact” is a decision; “ownership unknown” is an unresolved condition. Keep the latter in a review queue with an owner rather than silently assigning it to the first rep who sees the signal.

How should the fit filter remove unsuitable accounts?

Remove accounts where the offer lacks a plausible use case, even when their activity looks interesting. Use criteria that describe the problem you solve, the organization that experiences it, and any genuine service constraints. A large company or familiar industry label alone is not a sufficient fit argument.

  • What it removes: unsupported geographies, incompatible business models, unsuitable company types, and accounts outside an agreed market definition
  • Rule of thumb: require an account-level reason the offer could be relevant before spending time on a contact-level message
  • Implementation: save the criteria, inspect both accepted and rejected records, and preserve unresolved attributes instead of turning missing values into negative answers
  • Failure mode: apparent activity persuades the team to contact companies it would otherwise reject

Use company segments to reuse market criteria and audiences to combine those criteria with changing activity. A company can belong to every segment it matches, so overlapping segments do not establish exclusive rep ownership. Changing a shared segment also changes its uses, including engagement policies; review those dependencies before editing it.

When a business-model question needs research, a research agent in a Play can answer defined questions using selected response types and route qualifying and non-qualifying results. Test accepted and rejected records and inspect the supporting research. Do not treat an unsupported answer as a reason to widen the market automatically.

How do you find accounts that are in market right now?

Use current signals to identify a buying hypothesis, then test that hypothesis against the account’s situation. An event can show activity without establishing budget, an active project, or a particular employee’s involvement. The question is what the evidence supports saying and doing next.

  • What it removes: irrelevant topics, unexplained score changes, repeated copies of the same event, and activity with no connection to your offer
  • Rule of thumb: retain the underlying observation and distinguish it from your interpretation
  • Implementation: record source, event type, observed entity, event time, collection time, and the business question the event raises
  • Failure mode: the message claims the buyer is evaluating a solution when the evidence only shows account-level activity

Inspect whether combined signals add independent information. Several fields derived from the same event should not become several independent reasons for confidence. Use the compound-trigger framework to design combinations, then review actual matching records rather than assuming more conditions always improve the selection.

Account identity is also different from visitor identity. A company identified from website activity does not identify every person who browsed. Likewise, interpret G2 intent in its own source context rather than rewriting it as a direct conversation with a named employee.

For product-led teams, inspect which account and user generated a product event, what the event means, and whether a sales intervention fits the relationship. Do not assume any product action outranks all external signals. A specific request for help and routine background usage require different responses.

How do you find warm leads you can actually reach?

Identify an appropriate person and a permitted contact path before calling the account workable. The person’s role must connect to the business question, and the selected address or phone number must belong to the intended person. Contact enrichment can support reachability without proving that the contact generated the signal.

  • What it removes: wrong-role contacts, unresolved identity matches, unusable channels, and records that cannot pass current outreach controls
  • Rule of thumb: separate “this account deserves attention” from “this person should receive this message”
  • Implementation: inspect current role, company association, channel status, ownership, and prior conversation before choosing the next action
  • Failure mode: a plausible account signal turns into a highly personalized but incorrectly attributed message

In Unify, a company-based Play trigger needs a Prospect or Get People action before enrolling people. Prospect can prioritize existing records or find new people, while Get People returns people already present in Unify. Set persona priorities deliberately and review the resulting contacts before increasing the workflow’s scope.

If no suitable contact is available, keep the account for research or an owner-led introduction. Do not substitute an unrelated employee just to fill a list. For sparse regional coverage, retain the same identity standard and investigate another permitted data source or channel rather than treating missing data as proof of poor fit.

When is an intent signal too old to use?

A signal is too old when it no longer supports the intended next action. Evaluate what happened, when it happened, what changed since, and whether the account still fits. An ingestion timestamp should not make an old event look newly relevant.

  • What it removes: superseded events, delayed imports presented as recent activity, and signals whose business context has ended
  • Rule of thumb: choose a review window for the specific signal and action, then inspect the records near its boundary
  • Implementation: retain event and collection times separately, recheck account context, and define what new evidence would return an account to review
  • Failure mode: stale information becomes a false statement about the prospect’s current priorities

Use the signal-recency guide as a prompt to examine timing, not as a universal expiration rule. A company change and a website session need different interpretation. Set windows from the workflow’s purpose and review results before treating those windows as reliable defaults.

When automating in Unify, audience and criteria triggers can include records already matching at publication. They do not mean future matches only. Add the appropriate time condition when that is the intended boundary, and review re-entry settings so a repeated signal does not restart outreach unexpectedly.

How can you measure the filtering funnel without inventing benchmarks?

Use one deduplicated starting cohort and record the actual count that survives each gate. Keep the cohort definition and evaluation time fixed for the comparison; otherwise a dynamic audience can change while you are counting. A record may have several reasons for exclusion, so retain secondary reasons without counting it as several removed accounts.

Blank shortlist funnel worksheet
StageAccounts enteringAccounts retainedPrimary exclusion reasonsReview owner
Eligibility
Fit
Signal meaning
Reachability
Recency

Stage retention equals accounts retained divided by accounts entering that stage. Overall retention equals the final retained accounts divided by the original starting accounts. Leave the rate undefined when its denominator is zero, and keep technical failures or unresolved research separate from definitive rejections.

Review enough individual records to understand why the counts changed. A higher retention rate can reflect broader criteria, better data, or a broken exclusion. It does not establish better targeting by itself. Compare downstream conversations and accepted opportunities for equivalent cohorts before deciding that a filter improved commercial results.

What should you do when the shortlist is too long or too short?

Adjust the bottleneck you can explain, not whichever filter is easiest to loosen. A long list may need clearer signal relevance or a narrower market; a short list may expose limited data, an incorrect exclusion, or a genuinely small opportunity set. Neither length alone establishes success.

  • Too many suitable accounts: use account scoring to order work after the eligibility and evidence checks
  • Too few contacts: investigate reachability and identity coverage before broadening fit
  • Too little first-party activity: add permitted research sources while keeping their evidence levels distinct
  • A broad ICP: divide it into meaningful use cases so one generic message is not forced across different needs
  • An overloaded rep: use daily prioritization to protect commitments instead of declaring every surviving account urgent

In Unify, audience previews let you inspect companies and people separately, and audience exclusions affect that audience’s membership. Outreach still has its own eligibility checks. Before attaching automation, review how a change to audience criteria affects every Play or alert that uses it.

Check plan availability before designing an automated rollout: website intent, product signals, and signal-triggered automations are included in the Business plan. A manual review process and a continuously scheduled workflow have different prerequisites.

What happens to accounts that do not make the shortlist?

Assign a reason and a next state. Preserve eligible accounts that need more information, monitor those awaiting a meaningful new event, and exclude accounts that should not receive outreach. Nurture is appropriate only when the relationship, permissions, and channel make it appropriate.

A signal sent to Slack is still an alert, not a completed handoff. Use the signal-alert triage workflow to attach an owner and decision, then record whether the account was contacted, researched, deferred, or rejected. Explore Unify to connect account selection, qualification, and the next action in one workflow.

Frequently asked questions

How do I use intent data to find warm accounts?

Check eligibility first, then filter for fit, signal meaning, reachability, and recency. Preserve the evidence and a reason for each decision instead of treating the raw feed as a ready-to-contact list.

How do I find accounts that are in market right now?

Look for current evidence relevant to your offer and validate the account context. Intent data can identify a hypothesis worth investigating, but it does not prove an active purchase or budget.

How do I find warm leads for outbound?

Start with an appropriate account, identify a relevant current contact, and verify the contact path and outreach eligibility. Do not attribute an account-level event to that person without evidence.

Should I reject every account with an old signal?

No. Recheck what the event means and whether it still supports the planned action. Keep event time separate from import time, and use a context-specific review window.

How many accounts should survive the filters?

There is no universal target. Measure actual stage retention for a defined cohort and inspect the rejected records. Capacity, market size, evidence quality, and available contact paths all affect the workable list.

Can Unify automate this workflow?

Audiences can combine company and person criteria, and Plays can apply qualification and prospecting actions. Configure trigger timing, re-entry, ownership, exclusions, and enrollment behavior before activating the workflow.

Glossary

  • Intent data: Observed activity used to investigate possible commercial relevance
  • Warm account: An account with a defensible reason for attention, not guaranteed purchase intent
  • In market: Actively evaluating a purchase, which a signal alone may not establish
  • Account-level resolution: An activity association with a company rather than a verified individual
  • Reachability: Availability of an appropriate person and usable, permitted contact path
  • Signal meaning: What the underlying observation supports, distinct from inferred motivation
  • Signal recency: The age of the actual event in its current business context
  • ICP fit: Alignment between the account and the business problem your offer addresses

Sources