TL;DR: Intent data produces a list, not a shortlist. Turn it into one by running every account through four filters in order: fit, signal strength, reachability, and recency. A feed of 2,000 accounts typically narrows to 30-40 a single rep can realistically work in a week. This is for Sales, RevOps, and Growth teams sitting on more intent volume than they can act on, and the payoff shows up as reply rate, not list size.

Key Facts at a Glance

‍

Methodology and limitations. The four-filter order reflects cost of being wrong: fit errors waste the most rep time, so fit is screened first. The pass-through rates in the worked example below are illustrative figures chosen to show the arithmetic, not measured benchmarks. Your own rates depend on ICP breadth, traffic volume, and data coverage, and the reasoning behind each figure is stated inline so you can substitute your own numbers. We did not run a controlled study comparing filter orderings. Vendor coverage claims are attributed to the source that published them. Every Unify statistic in this article is attributed to the specific product page or customer case study it came from; there is no aggregated "Unify benchmark" dataset, and none is implied here.

Intent Data Does Not Find Warm Accounts. It Produces a List.

Intent data tells you something happened at an account. It does not tell you whether that account is worth a rep's time this week. Those are different questions, and conflating them is why so many teams sit on a feed of thousands of "in-market" accounts and still can't tell a rep who to call.

The fix is subtraction, not addition. A feed that surfaces 2,000 accounts and a rep with capacity for 40 means the real job of the filtering workflow is discarding roughly 98% of it defensibly, not finding more signals. Before going further, it's worth being clear on what kind of intent you're filtering: first-party signals (behavior on your own site or product) and third-party signals (behavior on other sites, like G2 or industry publications) decay and verify differently, and first-party vs. third-party intent signals is the prerequisite read if you haven't drawn that line yet.

What follows is a four-filter funnel: fit, signal strength, reachability, and recency, applied in that order. Each filter has a job, a rule of thumb, a way to implement it, and a specific failure mode if you skip it.

Why Should Fit Be the First Filter You Apply?

Fit comes first because a strong signal at a bad-fit account is still a bad account. No amount of buying intent turns a 12-person company into a viable deal if your product needs 200+ seats to make sense, and no amount of filtering downstream fixes an account that should never have entered the funnel.

• What it removes: Accounts outside your ICP on firmographics (size, industry, geography) or technographics, regardless of how strong the intent signal looks.
• Rule of thumb: Define fit as a hard gate, not a weighted score input. An account either clears your ICP thresholds or it doesn't; don't let a strong signal compensate for a fit failure.
• How to implement it: Screen against your closed-won firmographic profile first, using company data and technographics, before any signal data enters the evaluation. This is the same principle behind a composite account scoring model that multiplies fit by intent, recency, and reachability rather than averaging them, since a zero on fit should produce a zero overall.
• Failure mode if skipped: Reps spend their limited weekly capacity chasing accounts with real intent but no realistic path to close, which is a worse outcome than a smaller list of accounts that actually fit.

Which Signals Are Strong Enough to Justify Outreach Alone?

Not every signal earns a rep's time by itself. A single pricing-page visit and a job change into a decision-making role are not equivalent, and the signal-strength filter is where that distinction gets made explicit instead of left to instinct.

• What it removes: Accounts where the only activity is a single weak touch, such as one page view with no compounding behavior.
• Rule of thumb: Some signals justify outreach alone (a new hire in a target title, a G2 comparison-page view); most only count in combination. Reply rates roughly double when four or more signals stack on the same account, so treat compounding as the default bar rather than the exception.
• How to implement it: Tier signals into "strong alone" and "only counts combined" categories before you run the filter, so the rule is defined once rather than re-litigated per account. This is exactly what compound signal triggers covers in depth, including the AND-logic pairing (for example, a new hire plus a website visit) that consistently outperforms either signal alone; don't rebuild that logic here, just apply it.
• Failure mode if skipped: Every weak, single-touch signal gets treated as equally warm, which is how a 2,000-account feed turns into 2,000 mediocre outreach attempts instead of 40 good ones.

G2 intent is a useful case study in signal-strength tiering on its own: a visit to your own G2 profile, a competitor comparison view, and a category-page browse are not the same strength of signal, and how to use G2 intent data for outbound breaks down which of those moments justify solo outreach.

How Do You Know an Account Is Actually Reachable?

This is the filter most teams forget, and the one that most often turns a "warm" account back into an unworkable one. Account-level intent never becomes person-level certainty on its own: a company can show real buying signals while nobody there is reachable at the seniority or role you need.

• What it removes: Accounts with no verified, seniority-appropriate contact, even when the account-level signal is strong.
• Rule of thumb: Require a verified email or phone number at the right title before an account counts as workable, not just a company name and a signal. Realistic website-visitor match rates run 30-65% of U.S. B2B traffic at the company level; vendor claims north of 80% are outliers worth treating skeptically, per how website visitor identification actually works.
• How to implement it: Waterfall multiple contact-data vendors rather than relying on one; Unify's own contact layer chains 11+ email and phone vendors specifically because no single source clears every account, and bounce rate is the metric that tells you whether this filter is working. CandorIQ's founding SDR cut bounce rate from 15% to under 2% (an 87% reduction) after tightening exactly this filter, per the CandorIQ case study.
• Failure mode if skipped: Reps burn time on accounts that look warm on paper and produce zero conversations, because "the company visited our pricing page" was never the same claim as "we can reach the person who matters there."

How Old Is Too Old? The Recency Filter

Intent has a shelf life, and it's not the same shelf life for every signal type. A pricing-page visit and a job change decay at completely different rates, and applying one blanket "signals from the last 30 days" rule to both wastes the fast-decaying ones and prematurely drops the slow-decaying ones.

• What it removes: Accounts whose signal fell outside its category's useful window, even if fit, signal strength, and reachability all cleared.
• Rule of thumb: Match your outreach window to the signal's half-life rather than using a single cutoff for every signal type; a pricing-page visit decays in roughly 24 hours, while a job change stays relevant for closer to 30 days.
• How to implement it: Tag each signal type with its own decay window at ingestion, so recency filtering is a timestamp comparison against that signal's specific half-life rather than a manual judgment call per account. Full half-life tables by signal type live in signal decay: the half-life of buying signals, which reports 2-5x higher reply rates for teams that match cadence to decay versus those that don't; that math isn't rebuilt here.
• Failure mode if skipped: Reps reach out to accounts whose "in-market" moment already closed, which reads as generic or late rather than timely, and it quietly erodes trust in the whole signal-based motion.

What's the Arithmetic From Raw Feed to Working Shortlist?

Here is one illustrative pass through the funnel. Every rate below is a labeled assumption to show the shape of the math, not a benchmark to hit; your own pass-through rates depend on ICP breadth, traffic volume, and data coverage.

• Start: 2,000 accounts in the raw intent feed for the week.
• After the fit filter (illustrative 40% pass-through): 800 accounts remain. Reasoning: a moderately disciplined ICP typically excludes more than half of raw traffic on size, industry, or geography alone.
• After the signal-strength filter (illustrative 35% pass-through of the remainder): ~280 accounts remain. Reasoning: most accounts clear fit on a single weak touch; requiring a strong-alone signal or a compound pairing is a meaningfully higher bar.
• After the reachability filter (illustrative 55% pass-through of the remainder): ~155 accounts remain. Reasoning: even with a multi-vendor waterfall, not every account yields a verified, seniority-appropriate contact.
• After the recency filter (illustrative 25% pass-through of the remainder): ~39 accounts remain. Reasoning: signal decay is the harshest cut of the four because it's the one most teams skip entirely, so a disciplined pass here removes a large share of what "felt" current.

That's a working shortlist of roughly 35-40 accounts from a starting feed of 2,000, which is a realistic weekly capacity for one rep. If your own funnel produces a very different number, the filters are the same; only your pass-through rates change.

What Should You Do With the Accounts You Filtered Out?

A filtering workflow that only discards is a lossy one. Accounts that don't clear the bar this week aren't necessarily dead; most belong in one of three buckets rather than the trash.

• Nurture: Good fit, weak or absent signal. Route to a lighter-touch marketing sequence rather than rep-led outbound, and let a future signal promote the account back into the workflow.
• Monitor: Good fit, single weak signal that didn't clear the signal-strength bar alone. Hold and watch for a second signal to compound with the first, then re-run the filter.
• Discard: Poor fit regardless of signal, or a signal type that's structurally unreliable for your ICP (for example, job-seeker traffic on a careers page). These shouldn't re-enter the workflow without a fit change.

Real-time routing matters here too: a signal that clears all four filters loses value if it sits in a queue for two days before a rep sees it. Teams that pair this filtering workflow with tiered, real-time alerting (rather than a daily digest) convert a larger share of what survives filtering; see Slack-native signal triage for a four-tier routing structure that pairs well with the funnel above.

Which Filter Should You Tighten First? A Decision Framework

Use this when your filtered list is the wrong size or the wrong quality, rather than adjusting all four filters at once.

• If your working list has 200+ accounts per rep after all four filters, tighten fit first. A bloated ICP is almost always the root cause, not weak signal-strength or reachability rules.
• If your working list has fewer than 15 accounts per rep, loosen recency first. You're likely discarding signals still inside a usable window because the cutoff is too conservative.
• If reps book meetings but the accounts turn out to be a poor fit, tighten the fit filter's firmographic thresholds; don't touch signal strength, since the problem isn't signal quality.
• If reps call an account "warm" but can't reach anyone, tighten reachability by requiring a verified contact before an account counts as workable, not just a company-level match.
• If reply rates are healthy but total volume is too low for pipeline targets, loosen signal strength to accept single strong signals (a target-title job change, a demo request) instead of requiring compound signals for every account.
• If you run product-led growth with high free-tier signup volume, weight product-usage signals above website traffic in the signal-strength filter; a repeated paywall hit outranks a single site visit.
• If you run sales-led motion against a named-account list, keep fit narrow by definition (the named list is the fit filter) and use the other three filters purely for timing and reachability, not qualification.

How Do You Evaluate a Filtering Workflow, Regardless of Tooling?

These four criteria apply whether you're filtering manually in a spreadsheet, inside your intent vendor's platform, or through an outbound tool: does the workflow enforce fit as a hard gate rather than a weighted nudge; does it distinguish solo-strength signals from combine-only signals; does it verify reachability with a real contact rather than assuming one exists; and does it match recency windows to each signal type rather than using one blanket cutoff. A workflow that does all four, continuously, will consistently produce a smaller, higher-quality list than one that does even three of the four well.

How Unify covers this. Unify is outbound AI for sellers, where agents and reps work side by side. Its agents apply fit, signal-strength, and reachability screening continuously across a feed pulling from 40+ data sources, and research what survives before a rep ever opens the account, so the rep gets a short worked list and decides what to send rather than building the filter by hand every week. The honest limitation still applies here: filtering cannot manufacture warmth that isn't there, and account-level intent never becomes person-level certainty just because a platform automated the screening. Unify is built as AI for sellers, not an AI SDR that replaces one; a person still owns the message and the send.

Try Unify free to see the four-filter workflow run continuously against your own intent feed instead of rebuilding it in a spreadsheet each week.

Worked Example: One Account Through the Full Funnel

A mid-market fintech account visits a pricing page (weak alone), then three days later a VP of RevOps joins the company (a strong solo signal). Fit clears: the account matches ICP on size, industry, and existing tech stack. Signal strength clears on the new-hire signal alone, and the compounding pricing-page visit only strengthens the case. Reachability clears: a verified email and direct-dial are found for the new VP through a contact waterfall. Recency clears: the hire is 6 days old, well inside a roughly 30-day window for that signal type, and the pricing visit is corroborating rather than load-bearing at this point. The account enters the working list; a rep gets a short brief citing the hire and the pricing visit, and sends a personalized first touch inside the week rather than a generic sequence. This is the same shape of qualification, compounding, and quick action reflected in Perplexity's PQL and MQL plays, where the more qualified path produced a 20% reply rate against 5% for the thinner one, per the Perplexity case study.

Role and Segment Variants

• Sales (BDR/AE): Fit is often pre-decided by an assigned named-account list, so the reachability filter (finding the right person, not just the right company) tends to matter most day to day.
• Growth/Marketing: Fit runs at the audience or segment level rather than per account, and signal strength should weight top-of-funnel signals (content downloads, pricing-page views) more heavily than a sales-led motion would.
• RevOps: Owns the exclusion rules across sales and marketing tools so the same account isn't filtered twice, contacted twice, or dropped by one team's filter while still active in another's.
• PLG motion: Product-usage signals (paywall hits, seat growth, feature adoption) usually outrank third-party intent in the signal-strength filter, since they're first-party and behaviorally specific.
• Sales-led/enterprise motion: The fit filter is closer to a named-account allowlist, and the recency filter matters somewhat less, since ongoing human relationship-building offsets some signal decay.

Edge Cases and Disambiguation

• Low-traffic sites with thin first-party signal: First-party volume may be too sparse to be statistically meaningful; weight third-party intent more heavily in the signal-strength filter until first-party volume grows.
• Very broad ICPs: The fit filter alone won't shrink a broad ICP enough on its own; narrow to a sub-segment (vertical, tech stack, company stage) before adjusting the other three filters.
• PLG companies where product usage outranks third-party intent: A free-tier account hitting a usage cap repeatedly is a stronger signal than an anonymous website visit; don't let a generic signal-strength rule undervalue it.
• Non-U.S. accounts with sparse data coverage: Waterfall vendors typically have thinner non-U.S. coverage, so a strict reachability threshold will over-reject internationally; loosen it slightly and lean on manual verification for these accounts.
• Job-seeker traffic vs. buyer interest: A visit to a careers or jobs page is not a buying signal and should be excluded at the fit or signal-strength stage by URL pattern, not treated as intent.

When Is an Account Not Actually Warm? Stop Rules and Red Flags

Top 5 Mistakes to Avoid

• Treating feed volume as pipeline. A 2,000-account feed is a list that hasn't been filtered yet, not 2,000 opportunities.
• Skipping the fit filter to chase every signal. A strong signal at a bad-fit account still produces a bad account.
• Counting a signal as warm without checking reachability. Real intent with no verified contact isn't workable this week regardless of how the account looks on paper.
• Using signals past their half-life. A 45-day-old pricing-page visit behaves like noise, not intent, once it's well past its decay window.
• Filtering once instead of continuously. Intent decays; a list filtered on Monday is already partly stale by Thursday.

Frequently Asked Questions

How do I use intent data to find warm accounts?

Run every account through four filters in order: fit, signal strength, reachability, and recency. Fit removes accounts outside your ICP regardless of signal. Signal strength removes accounts where the only activity is a single weak touch. Reachability removes accounts with no verified, seniority-appropriate contact. Recency removes accounts whose signal is past its useful window. What survives all four is your working list for the week, usually a small fraction of the original feed.

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

In-market accounts are the subset of your intent feed where the signal is both strong and recent, meaning it fell inside its category's decay window. A pricing-page visit from three weeks ago is not in-market anymore; a job change or a compound signal from the last week usually still is. Apply the recency filter last, after fit and reachability have already narrowed the list, so you're checking timing on accounts worth timing.

How do I find warm leads for outbound?

A warm lead is a person, not just an account, so add a reachability check on top of account-level warmth: a verified email or phone number, at a title and seniority that matches your persona. An account can show real intent and still produce zero warm leads if nobody there is reachable, which is why reachability is its own filter rather than an assumption baked into the others.

What is the difference between an intent signal and a warm account?

An intent signal is a single data point, such as a website visit, a job change, or a G2 comparison view. A warm account is what's left after that signal has survived fit, signal-strength, reachability, and recency filtering. Most raw signals never become warm accounts; treating every signal as a warm account is one of the most common mistakes in signal-based selling.

How many accounts should a rep work in a week?

There's no universal number, but most individual reps can realistically research, personalize for, and follow up on somewhere in the 30-to-50-account range per week without the quality of outreach collapsing. If your filtered list routinely produces several hundred accounts per rep, tighten the fit or signal-strength filter rather than asking reps to move faster.

What if my filtered list is too small?

Loosen the filters in a specific order rather than all at once. Start with recency (accept a slightly wider decay window), then signal strength (accept single strong signals instead of requiring compound signals), and treat fit as the last lever to touch, since loosening ICP criteria tends to produce the lowest-quality accounts of the four.

How often should I re-run the filtering workflow?

Continuously, not weekly. Intent decays on a per-signal basis, so a list that was accurate on Monday can be stale by Thursday, especially for fast-decaying signals like page visits. Teams filtering on a fixed weekly cadence are usually working a list that's already partly expired by the time reps touch it.

Can filtering ever be fully automated?

The fit, signal-strength, and reachability checks can run continuously without a human, and recency is a timestamp comparison that automates easily. What doesn't automate away is judgment on the accounts that survive: a rep still decides what to say and whether to send it. Filtering can produce a short list; it can't manufacture warmth that genuinely isn't there.

Glossary

• Intent data: Behavioral information, first-party or third-party, indicating a company or person may be researching or evaluating a purchase.
• Warm account: An account that has survived fit, signal-strength, reachability, and recency filtering, distinct from any account that merely shows a raw signal.
• In-market: An account whose signal is both strong enough and recent enough to fall inside that signal type's decay window right now.
• Account-level resolution: Identifying which company a piece of behavior (like a website visit) belongs to, as distinct from identifying the specific person.
• Reachability: Whether a verified, seniority-appropriate contact exists at an account, independent of whether the account shows intent.
• Signal strength: Whether a given signal justifies outreach on its own or only counts when combined with another signal.
• Signal recency: How much time has passed since a signal fired, measured against that signal type's specific half-life rather than a single blanket window.
• ICP fit: Whether an account matches your ideal customer profile on firmographics and technographics, evaluated independently of any intent signal.

Sources

• Unify, Signals & Intent product page: unifygtm.com/products/signals
• Unify, B2B Company & Contact Data product page: unifygtm.com/product/b2b-company-contact-data
• Unify, Plays product page: unifygtm.com/product/plays
• Unify customer story, Quo: unifygtm.com/customers/quo
• Unify customer story, Perplexity: unifygtm.com/customers/perplexity
• Unify customer story, CandorIQ: unifygtm.com/customers/candoriq
• Unify customer story, Navattic: unifygtm.com/customers/navattic
• Unify Explore, First-Party vs. Third-Party Intent Signals: unifygtm.com/explore/first-party-vs-third-party-intent-signals
• Unify Explore, Compound Signal Triggers: unifygtm.com/explore/compound-signal-triggers
• Unify Explore, Signal Decay: The Half-Life of Buying Signals: unifygtm.com/explore/signal-decay-half-life-of-buying-signals
• Unify Explore, Composite Account Scoring for Signal-Led Outbound: unifygtm.com/explore/composite-account-scoring-signal-led-outbound
• Unify Explore, How to Use G2 Intent Data for Outbound: unifygtm.com/explore/how-to-use-g2-intent-data-for-outbound
• Unify Explore, How Website Visitor Identification Actually Works: unifygtm.com/explore/website-visitor-identification-how-it-works
• Unify Explore, Slack-Native Signal Triage: unifygtm.com/explore/slack-alerts-buying-signals-4-tier-triage
• Bombora, Company Surge intent methodology

Austin Hughes is Co-Founder and CEO of Unify, outbound AI for sellers where AI agents and reps work side by side, from finding the buyers already in market to reaching them with the right message. Before founding Unify, Austin led the growth team at Ramp, scaling it from 1 to 25+ people and building a product-led, experiment-driven GTM motion. Prior to Ramp, he worked at SoftBank Investment Advisers and Centerview Partners.