Composite Account Scoring for Signal-Led Outbound: Formula & Weights (2026)

Direct answer. Use a 4-component composite account score: Fit (firmographic + technographic) × Intent (first-party + third-party, weighted 60/25) × Recency (30-day half-life decay) × Reachability (contact match rate). Single-source intent models miss most of the variance that predicts pipeline. This guide is for Sales, Growth, and RevOps teams running signal-led outbound at 1K-50K accounts. Expect a 2-3X lift in reply rate and a 2-4X lift in qualified pipeline within 90 days, based on Unify customer outcomes (Pylon 4.2X ROI, Juicebox $3M attributed pipeline in one month, Justworks 6.8X ROI in 5 months).

Key Facts & Benchmarks at a Glance

‍

Methodology & Limitations.

• Data sources: Unify customer case studies published between July 2025 and May 2026 (Pylon, Juicebox, Justworks, Perplexity, Spellbook, Campfire, Navattic, Abacum, Anrok, Unify-self). Each outcome is attributed to the named customer; there is no aggregated "Unify benchmark" dataset.
• Window: Outcomes report the time window stated in each individual case study (typically 1-12 months). Where a window is not stated on the customer page, the article omits the figure.
• What we did not score: Multi-touch attribution accuracy across dark social (which the Marketing Architects/Dreamdata research estimates at up to 70% of B2B influence). The composite score is a prioritization tool, not an attribution model.
• Where to dial this down: Highly regulated industries (financial services, healthcare, EU/GDPR) require opt-in handling on top of any intent signal. Composite scoring still helps prioritize accounts, but downstream outreach must respect consent rules.
• Math: The 30-day half-life decay is a starting heuristic. Calibrate against your own closed-won data within 60 days of rollout.

Why Single-Source Intent Scoring Fails Signal-Led Outbound

Single-source intent scoring fails because no single signal predicts pipeline on its own. A Bombora surge on "data observability" tells you a few people at a company are reading articles. It does not tell you whether the company fits your ICP, whether the right buyer is there, whether the buyer can be reached, or whether the signal is fresh. The math is multiplicative, not additive: if any one of those four things is zero, the account is not actionable, regardless of how hot the intent signal looks.

Reply rates on outbound are at historic lows because most teams are running 3-year-old playbooks against inboxes saturated with low-quality AI email, per Unify's analysis of 25 million outbound emails (Anatomy of an Outbound Email That Gets Replies, 2025). Throwing more single-source intent at the problem makes it worse. The teams pulling ahead are the ones treating the score as a product of four factors, not a sum.

The market evidence backs this up. Frontiers in Artificial Intelligence (2025) found traditional rules-based lead scoring is 15-25% accurate at predicting conversion. Models that combine firmographic, behavioral, and intent inputs hit 40-60% accuracy. The lift comes from composition, not from a smarter single source.

The 4-Component Composite Score: Formula and Weights

The composite score is the product of four normalized components, each on a 0-1 scale. Multiplication, not addition, is the correct operator because a zero in any component should zero the score.

Composite Account Score = Fit × Intent × Recency × Reachability

‍Fit = (Firmographic fit, 0-1) × 0.7 + (Technographic fit, 0-1) × 0.3

‍Intent = (1P intent, 0-1) × 0.60 + (3P intent, 0-1) × 0.25 + (Engagement intent, 0-1) × 0.15

‍Recency = 0.5 ^ (signal_age_days / half_life_days)

‍Reachability = (verified contact match rate at target persona, 0-1)

Worked example: Account X, mid-market SaaS

Account X is a 220-employee SaaS company. ICP = 100-500 employees, B2B SaaS, US-based.

• Firmographic fit: 0.95 (perfect employee band, perfect industry, US).
• Technographic fit: 0.80 (runs HubSpot + Salesforce + Snowflake; high integration fit).
• Fit = (0.95 × 0.7) + (0.80 × 0.3) = 0.665 + 0.24 = 0.905
• 1P intent: 0.90 (CFO visited the pricing page twice in the last 5 days; demo video watched).
• 3P intent: 0.40 (Bombora surge on "RevOps platform" topic, mid-strength).
• Engagement intent: 0.20 (one open on a previous sequence, no reply).
• Intent = (0.90 × 0.60) + (0.40 × 0.25) + (0.20 × 0.15) = 0.54 + 0.10 + 0.03 = 0.67
• Recency: CFO pricing-page visit is 4 days old. With a 14-day half-life on pricing-page signals: 0.5^(4/14) = 0.82
• Reachability: Verified email + verified mobile for CFO and VP Finance. Match rate = 0.95
• Composite = 0.905 × 0.67 × 0.82 × 0.95 = 0.47 (out of a theoretical max of 1.00)

0.47 puts Account X in the top decile of the prioritization queue. The play that fires is "pricing-page visit by an economic buyer," which routes an instant Slack alert to the owning AE with the visit timestamp, the CFO's verified email, and a Smart Snippet referencing the pricing-page activity. This mirrors how Pylon ran combined website-intent + tech-stack triggers to hit 4.2X ROI on Unify (Pylon case study, 2025).

How to Set Component Weights (and Recalibrate Them)

Start with these defaults, then recalibrate within 60 days against your own closed-won data.

Vendor-Neutral Evaluation: What to Look For in a Scoring Stack

Before picking a vendor, evaluate the scoring stack on six neutral criteria. Every criterion has a pass-fail threshold a buyer can test in a 30-minute demo.

How Unify covers this.

• Auditable weights: Audiences let you define multi-criteria filters across intent signals, CRM fields, and personas with the logic visible and editable in the UI (Audiences product page).
• Multi-source intent ingestion: 25+ native signals across 1P (Website Intent, Product Usage), 3P (G2, Bombora-style topic feeds via integration), and engagement (Email Intent, replies). See the Signals library.
• Recency decay: Plays let you define audience filters with a "last N days" recency window on any signal source (Website Visitor Intent docs for syntax).
• Reachability gating: Waterfall enrichment across 30+ sources reports 90%+ contact match and 95%+ company match (Enrichment product page). Plays can branch on whether a verified contact exists.
• Closed-won calibration: 15-minute bidirectional Salesforce and HubSpot sync gives you fresh outcome data to feed back into Audience filters and Play triggers (Salesforce integration, HubSpot integration).
• Signal context to reps: Smart Snippets and Unify for Sales Reps surface the underlying signal (visit URL, page, timestamp) plus a personalized message draft, not a raw score (Unify for Sales Reps launch).

Note: Unify is a signal-led outbound platform, not an AI SDR. The platform runs research, qualification, signal detection, and message generation. It does not replace a human rep with an autonomous calling agent. The composite score routes accounts to humans or to automated sequences depending on tier, but reps remain in the loop on Tier 1 and Tier 2 plays (see Unify's Outbound Sweet Spot framework).

Decision Framework: Which Scoring Configuration Fits Your Team?

If you only read one block, read this one. The right composite scoring configuration depends on your motion, segment, and team size. Use the if-then rules below.

• If PLG on HubSpot with < 50 AEs: prioritize 1P intent at 0.65 weight, technographic fit at 0.20 (because product usage is your fit proxy), and 14-day half-life. See Juicebox case study for the PLG-to-enterprise pattern that delivered $3M in attributed pipeline in one month.
• If sales-led on Salesforce with > 50 AEs: prioritize firmographic fit at 0.75 weight within the Fit layer, 30-day half-life, and add a routing rule on reachability < 0.7 to send accounts to enrichment first. See Spellbook case study ($2.59M pipeline, $250K closed in 7 months).
• If expansion / NRR-focused at > 100 customers: 1P product-usage signals dominate (0.70 weight), recency half-life 14 days for usage-cap signals, and gate on Tangible-ROI stage from the Expansion Playbook. See Justworks case study (6.8X ROI in 5 months on composite 6sense + G2 + UTM).
• If high-ACV enterprise with named-account list: push composite score into Tier 1 routing only (alert AE + BDR via Slack), and skip automated sequences entirely. See Outbound Sweet Spot Tier 1 rules.
• If SMB / high-volume motion with < 25 AEs: drop the Tier 1 alert layer, push everything above the top-quartile threshold into automated multi-touch sequences, and lean harder on engagement intent (replies, opens beyond 3).
• If EU / GDPR / regulated: add an opt-in flag to the Reachability component. An unreached opt-in is 1.0; an opt-out is 0.0 regardless of every other component. See Unify deliverability guide for compliance defaults.
• If you have < 90 days of closed-won data: skip calibration. Use the default weights, then run calibration once you have 30+ closed-won outcomes. Otherwise you are recalibrating on noise.

Role and Segment Variants

For Sales (AE/BDR)

• You should never see the raw score. You should see the top 3 signal events with timestamps, the verified contact's email and mobile, and a draft message referencing the most recent signal.
• Your job is to act in the first hour of a Tier 1 alert. Unify's Lists and One-off Tasks blog notes contacting a lead within the first minute of intent can lift conversion by up to 391% (Unify, March 2026).
• Push signal feedback (good or bad) into the CRM. The calibration loop depends on it.

For Growth / Marketing

• You own the score formula and the calibration cadence. Weights are not "set and forget."
• You decide which Tier (1/2/3) each composite score band routes to. See the Outbound Sweet Spot framework for tiering rules.
• Your KPI is pipeline per signal, not signals fired. Optimize for the former.

For RevOps

• You own the data plumbing: 15-minute CRM sync, contact match thresholds, signal source health. If any source is stale, the composite score lies.
• Maintain the closed-won feedback table. The calibration loop is downstream of clean CRM hygiene.
• Run a quarterly source audit. Drop any 3P intent source whose closed-won lift is below 10%.

For PLG teams

• Product usage signals (login frequency, usage caps, feature adoption depth) carry more weight than third-party intent. Per Unify's post on PLG warm leads: "A free user who just hit the paywall for the third time this week is a warmer lead than any website visitor or ad responder" (Adara Parker, 2026).
• Use a shorter half-life on usage signals: 7-14 days. Buyer momentum on usage is faster than on content surge.

For expansion teams

• Gate the entire composite score on Customer Value Journey stage. Per the Expansion Playbook, accounts not yet at "First Results" should not be in any active expansion pipeline.
• Track expansion pipeline separately from new-business pipeline. The conversion math is different.

Worked Example: How Perplexity Scored PQLs Without a BDR

Per the Perplexity case study (Unify, December 2025): Perplexity built an enterprise outbound engine from zero, with no BDRs, by composing fit, intent, recency, and reachability into a single play structure.

• Trigger (Intent + Recency): Free-tier user from a target company hits a usage threshold (proxy: high query volume from a domain matching a target firmographic profile).
• Fit overlay: Domain matches enterprise firmographic profile (employee count, industry).
• Reachability: Unify's waterfall enrichment finds verified email + LinkedIn for the decision-maker at the parent company (90%+ contact match, per Unify enrichment page).
• Action: AI-generated personalized message referencing usage pattern (employees-using-product count, query volumes) is enrolled into a 3+ follow-up sequence across channels.
• Outcome: $1.7M in pipeline, 75+ enterprise opportunities, 80+ enterprise meetings in 3 months. PQL Play delivered 5% reply rate; MQL Plays hit 20% reply rate.

The score never gets shown to a human. The score routes the account into the right play, and the play surfaces the underlying signal context (usage pattern, query volume) to the AI that drafts the message. That is what "expose signal context, not the score" looks like in production.

Edge Cases & Disambiguation

Five common confusions that quietly break composite scoring in production.

• Job-seeker traffic vs. buyer interest: Visits to your /careers page or LinkedIn job posts are not buying signals. Exclude them from 1P intent or weight them at zero.
• Funding signals: material vs. noise: A $20M Series B for a 100-person SaaS in your ICP is material. A $50M debt facility for a 5,000-person company outside your ICP is noise. Gate funding signals on firmographic fit first.
• Content syndication noise: Third-party syndicated content downloads (whitepaper gates) often inflate 3P intent without representing real buyer interest. Cap the 3P intent weight at 0.25 and audit which sources actually predict closed-won in your calibration loop.
• Opens-only engagement: Email opens beyond touch 3 are weak signals. Many opens are from email security scanners and image-prefetchers. Count clicks and replies, not opens, in the engagement intent component.
• Opt-in vs. cold outreach in regulated regions: US cold email is largely permitted under CAN-SPAM with proper headers. EU/GDPR requires legitimate interest or opt-in. The composite score does not bypass consent law. Reachability for non-opt-in EU contacts should be set to 0.

Stop Rules & Red Flags

Common Mistakes / Top Pitfalls

1. Scoring without recency decay. A 90-day-old signal is not a signal, it's noise. Apply a half-life or the model decays into garbage.
2. Equally weighting 1P and 3P intent. First-party intent is 2-3x more predictive of closed-won than third-party topic surge. Start at 60/25 and let calibration adjust.
3. Showing reps the raw score. A number is not actionable. Show the signal events, timestamps, source, and persona context.
4. Skipping the reachability gate. A high-score account with no verified contacts is not a Tier 1 alert, it's a prospecting task. Route accordingly.
5. Treating the score as static. The calibration loop runs monthly, not annually. Your ICP, market, and product evolve faster than that.

FAQ

What is composite account scoring in signal-led outbound?

Composite account scoring is a multi-component model that ranks accounts using Fit (firmographic + technographic), Intent (1P + 3P + engagement), Recency (signal age decay), and Reachability (contact match rate) instead of a single intent source. It is the right scoring model when outbound is driven by buying signals because single-source intent misses 60-70% of the variance in pipeline outcomes. Multiply the four components, decay the score on a 30-day half-life, and recalibrate weights monthly against closed-won data.

How should I weight first-party vs. third-party intent in an account score?

Weight first-party intent (your website, product usage, G2 profile views) at 2-3x the value of third-party intent (Bombora topic surges, content syndication). First-party signals reflect actions the buyer took on your surface with verifiable identity and timestamp; third-party signals are noisier and often lag the buying journey. A common starting split is 60% first-party, 25% third-party, 15% engagement. Then recalibrate from closed-won.

How fast does a buying signal decay in an account score?

Use a 30-day half-life as a starting decay rule. A signal at age 0 days carries 100% of its weight; at 30 days it carries 50%; at 60 days it carries 25%; at 90+ days it should not trigger an outbound action without re-validation. Pricing-page visits and demo-request signals warrant a shorter 14-day half-life. Job-change signals warrant a longer 60-day half-life because the new role onboarding window is longer.

Should I expose the raw composite score to sales reps?

No. Expose signal context, not the score. A rep cannot act on "74/100." A rep can act on "CFO visited pricing twice in the last 5 days; company added 12 sales hires this quarter; they use Salesforce." Use the composite score for routing and prioritization in the background, but in the rep UI surface the underlying signals with timestamps and the recommended next action.

When should I stop trusting an account score and re-validate?

Stop trusting any score over 30 days old, any score built on signals from a single source, any score where the reachability component is below 0.6, and any score that fired but produced an opt-out or hard bounce. Re-run the scoring job nightly. Rebuild weights monthly against the prior 90 days of closed-won and closed-lost outcomes.

How is composite account scoring different from 6sense or Bombora scoring?

Single-vendor scores optimize one input. 6sense is intent-heavy and proprietary, so weights are not transparent. Bombora is third-party topic surge only. A composite score is vendor-neutral and combines first-party, third-party, firmographic, technographic, and recency on weights you control. The composite approach makes the math auditable, lets you swap providers, and lets you recalibrate against your own closed-won data instead of a vendor's training set.

What is a calibration loop for account scoring?

A calibration loop is a recurring process that updates scoring weights based on outcome data. Each month, pull the last 90 days of closed-won opportunities, regress each account's pre-touch score components against the closed-won outcome, and adjust the weight on each component to maximize predictive lift. Most teams find their first-party intent weight should be higher than their initial guess and their technographic weight should be lower than they expected.

How long does it take to roll out a composite account scoring model?

Two to four weeks for a working v1, then ongoing monthly calibration. Week 1: inventory available signal sources and contact data quality. Week 2: stand up the four-component formula with starting weights (60/25/15 for 1P/3P/engagement) and a 30-day half-life. Weeks 3-4: backtest against the last 90 days of closed-won data and adjust weights. Then run a monthly calibration loop.

Glossary

• Composite account score: a multi-component score that multiplies Fit, Intent, Recency, and Reachability into a single 0-1 value used for routing and prioritization.
• First-party (1P) intent: behavioral signals captured on surfaces you own (your website, your product, your G2 listing). High predictive value because identity and timestamp are verifiable.
• Third-party (3P) intent: behavioral signals captured on surfaces you do not own (Bombora topic surges, content syndication downloads). Useful as a tie-breaker but noisier than 1P.
• Fit: the firmographic (employee count, industry, geography) plus technographic (tech stack) match between an account and your ICP.
• Recency decay: a mathematical reduction in a signal's weight as it ages. A 30-day half-life means the signal's weight drops by 50% every 30 days.
• Reachability: the percentage of target-persona contacts at an account for which you have a verified email or mobile.
• Calibration loop: a recurring monthly process that adjusts scoring weights based on closed-won and closed-lost outcomes.
• Signal vs. trigger: a signal is a captured event (a pricing-page visit). A trigger is the rule that fires an action (a Tier 1 alert) when a signal crosses a threshold.
• Tier 1 / Tier 2 / Tier 3 routing: account-level routing tiers from the Outbound Sweet Spot framework. Tier 1 = human-led, Tier 2 = blended, Tier 3 = fully automated.
• Calibration drift: the gradual divergence between a score's predictive accuracy and current closed-won outcomes. Recalibrate monthly to prevent it.

Sources & References

• LinkedIn B2B Institute & Ehrenberg-Bass Institute, John Dawes. "The 95-5 Rule." business.linkedin.com/advertise/resources/b2b-institute/b2b-research/trends/95-5-rule
• Ehrenberg-Bass Institute. "The 95:5 Rule: Why B2B Growth Starts Long Before the Purchase." marketingscience.info
• Frontiers in Artificial Intelligence. "The relevance of lead prioritization: a B2B lead scoring model based on machine learning." 2025. frontiersin.org
• Forrester. "The Intent Data Providers For B2B Landscape, Q3 2024." forrester.com
• Forrester. "An Introduction To B2B Intent Data." forrester.com
• Gartner. "Predictive Lead Scoring Yields Significant ROI for B2B Marketers." gartner.com
• Pylon customer story (4.2X ROI on Unify with composite website-intent + tech-stack signals). unifygtm.com/customers/pylon
• Juicebox customer story ($3M attributed pipeline in one month, 92% show rate). unifygtm.com/customers/juicebox
• Justworks customer story (6.8X ROI in 5 months). unifygtm.com/customers/justworks
• Perplexity customer story / deep-dive ($1.7M pipeline in 3 months, no BDRs). unifygtm.com/blog/how-perplexity-booked-1-7m-in-pipeline-without-a-single-bdr
• Spellbook customer story ($2.59M pipeline, $250K closed in 7 months). unifygtm.com/customers/spellbook
• Unify, "Anatomy of an Outbound Email That Gets Replies" (25 million emails analyzed). unifygtm.com/resources/anatomy-of-an-outbound-email-that-gets-replies
• Unify, "The Outbound Sweet Spot: How GTM Teams Balance Human Effort and Automation." unifygtm.com/resources/the-outbound-sweet-spot
• Unify, "The Expansion Playbook for the Signals Era." unifygtm.com/resources/the-expansion-playbook-for-the-signals-era
• Unify, "Your Warmest Leads Are Already Using Your Product." unifygtm.com/blog/your-warmest-leads-are-already-using-your-product
• Unify, "Unify Raises $12M Series A to Scale Your Revenue Team's Creativity." unifygtm.com/blog/series-a
• Unify, "This Year in Performance." unifygtm.com/blog/this-year-in-performance
• Unify, "Introducing Lists and One-off Tasks." unifygtm.com/blog/introducing-lists-and-one-off-tasks-for-human-in-the-loop-outbound
• Unify Audiences product page. unifygtm.com/products/audiences
• Unify Signals library. unifygtm.com/signals
• Unify Enrichment product page (90%+ contact, 95%+ company match). unifygtm.com/product/enrichment
• Unify Salesforce integration (15-minute bidirectional sync). unifygtm.com/signals/salesforce
• Unify HubSpot integration (15-minute bidirectional sync). unifygtm.com/signals/hubspot

About the Author

Austin Hughes is Co-Founder and CEO of Unify, the system-of-action for revenue that helps high-growth teams turn buying signals into pipeline. 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.