By Siddharth Gangal What is a signal scoring framework?
Direct answer. A signal scoring framework is a structured method that ranks every buying signal — job changes, pricing-page visits, competitor reviews, funding events — by the probability that working it now produces pipeline. It assigns weighted points across factors like source quality, intent strength, account fit, recency, activity volume, and the role of the person, then triggers a defined outreach action and SLA based on the composite. Reps work the highest-probability signals first instead of the loudest ones.
Most outbound teams have the opposite problem they had three years ago. The shortage is not signal. The shortage is judgment. A single AE inside a mid-market software company can see 200 to 600 surfaced signals a week across signal detection tooling, intent platforms, CRM enrichment, and inbound forms. Without a scoring framework, reps either chase the loudest one or freeze and work nothing. The pipeline cost is the same either way.
A signal scoring framework fixes the judgment problem by encoding it. Instead of asking a rep to decide whether the third pricing-page visit by a director at a 200-person SaaS company in Texas matters more than a job change at a 4,000-person fintech, the framework scores both, ranks them, and tells the rep which one to open first. The number on the screen replaces the gut feel. Gut feel still matters at the edges. The score is what scales.
According to the 2023 Gartner Technology Marketing Benchmarks Survey, 93 percent of technology marketers with 100 million dollars or more in annual revenue used third-party intent data for at least one use case (Gartner, 2024). The teams that report exceptional ROI from that data are the 24 percent who layer multiple signal sources and weight them by probability of conversion. Everyone else is collecting signals and not scoring them. That is the gap this framework closes.
Lead scoring vs. signal scoring: the real difference
Lead scoring and signal scoring solve different problems. Lead scoring evaluates a person across firmographic fit and historical engagement, usually inside the marketing automation platform. The output is a number that tells marketing whether the lead is ready for sales. Signal scoring evaluates a single moment of intent and tells sales whether the next 24 hours are worth a phone call. The two answers do not always agree.
A lead can be a perfect-fit VP at a target account, score 95 out of 100 in HubSpot or Marketo, and still be cold this week. A non-MQL contact at the same account can drop a 30-point signal — they searched a competitor on G2, then visited the integrations page twice in two days — and that signal is worth a rep call inside the hour. Lead scoring would route the first contact. Signal scoring routes the second. Teams that run both score the same account twice, with two different operating questions, and let the higher-priority signal win when the two collide.
| Dimension | Lead scoring | Signal scoring |
|---|---|---|
| Unit of analysis | A person, evaluated over time | A single intent moment, evaluated right now |
| Primary inputs | Firmographics, demographics, historical engagement | Source, intent type, recency, activity, role |
| Output | An MQL flag with a static score | A composite score plus an outreach SLA |
| Owner | Marketing operations | Sales operations or RevOps |
| Decay model | Slow — quarters | Fast — hours to weeks |
| Failure mode | Stale scores that never trigger action | Loud-but-shallow signals that waste rep time |
| Best practice | Run both — let signal scoring break ties on routing speed | Run both — let signal scoring break ties on routing speed |
The Salespanel team frames the trade-off the same way: lead scoring is simple, deterministic, and works for small teams that distrust opaque models. Signal-based selling delivers more pipeline per rep hour when the data is reliable, but it requires a governance layer the lead scoring world never needed (Salespanel, 2025). The SIGNAL Score in the next section is that governance layer.
Pro tip. If your CRM still routes leads off a marketing automation MQL flag alone, expect 60 to 80 percent of your signal-based revenue to leak. The signal arrives, no one is paged, and the moment is gone by the time the lead syncs into Salesforce overnight.
The SIGNAL Score: a 6-factor model that ranks every buying signal
The SIGNAL Score is a transparent, six-factor composite that scores every signal on a 0 to 100 scale and maps the result to a defined outreach SLA. Each factor scores 0 to 10. The factors carry different weights based on how much they shift conversion probability in practice. The composite is the weighted sum, multiplied by a normalization constant so the maximum lands at exactly 100.
The acronym is the model. Source quality. Intent strength. Geography and account fit. Newness. Activity volume. Lead role. Six factors, six letters, one number that lands in the rep queue.
| Factor | What it measures | Score range | Weight | Max contribution |
|---|---|---|---|---|
| Source quality | Trust in the data provider | 0–10 | 1.5× | 15 pts |
| Intent strength | How close the action is to a buying decision | 0–10 | 2.5× | 25 pts |
| Geo + account fit | ICP match plus territory eligibility | 0–10 | 2.0× | 20 pts |
| Newness | Time since the signal fired | 0–10 | 1.5× | 15 pts |
| Activity volume | Number of correlated signals in the past 14 days | 0–10 | 1.0× | 10 pts |
| Lead role | Decision authority of the person on the signal | 0–10 | 1.5× | 15 pts |
| Composite SIGNAL Score | 0–100 | |||
The weights are not arbitrary. Intent strength carries the heaviest multiplier because no other factor predicts a meeting as reliably — a pricing-page visit plus a G2 comparison view inside the same week is closer to revenue than any firmographic profile. Source quality and newness are tied for the third weight because a perfectly-timed signal from a shaky source still warrants a call, and a flawless signal from last quarter does not. Activity volume gets the lowest weight because it is the easiest factor to game with noisy automation traffic.
Each of the next six sections defines one factor, gives the 0 to 10 rubric, and shows how it behaves in practice. The model is designed to be auditable. A rep should be able to open a 78-score signal, read the six component scores, and either trust the number or override it with a one-line note. Black boxes lose rep trust inside a quarter. Transparent rubrics survive year after year.
Factor 1 — Source quality: rank the platform before you rank the signal
Source quality scores the data provider, not the signal itself. A first-party event on the company website is structurally more trustworthy than a third-party topic surge inferred from cookie pools. The factor exists because every signal scoring framework eventually meets a noisy provider, and operators need a way to discount that provider without throwing out the rest of the data.
Score the source on a 0 to 10 scale using the rubric below. Build the table once with RevOps, lock it inside source control, and review every quarter.
| Source tier | Examples | Score |
|---|---|---|
| Tier 1 — Verified first-party | Form fill, demo request, product trial, sales-routed inbound | 9–10 |
| Tier 2 — Identified first-party | De-anonymized web visit, identified email click, in-product event | 7–8 |
| Tier 3 — High-trust third-party | G2 / TrustRadius review activity, LinkedIn job change, funding round | 5–7 |
| Tier 4 — Aggregated third-party intent | Bombora Company Surge, 6sense topic spike, Demandbase keyword cluster | 4–6 |
| Tier 5 — Inferred or modeled | Lookalike account signals, predictive fit scores without intent input | 1–3 |
Bombora itself frames its Company Surge score as the ratio of recent topic consumption to a 12-week baseline (Bombora, 2025). That is useful, but it is still inferred from cookie traffic and topic taxonomies, not a buyer raising a hand. In the SIGNAL Score, a Bombora signal that lands at Surge 75 enters the pipeline with a source score of 5 or 6, not a 10, even though Bombora calls it strong. The other five factors will lift the composite if the signal is real.
Factor 2 — Intent strength: separating curiosity from active evaluation
Intent strength is the heaviest factor in the SIGNAL Score because it correlates the most tightly with closed-won deals. Score the signal type on the action ladder: a pricing-page visit ranks above a blog read. A demo request ranks above a pricing-page visit. A competitor review on G2 ranks above either because it implies the buyer has already evaluated alternatives.
The intent strength rubric collapses dozens of raw signal types into five behaviour clusters. Calibrate the rubric against the last 90 days of closed-won deals — pull every won deal, tag the earliest detectable signal, and use the historical conversion rate as the score floor.
| Intent cluster | Examples | Score |
|---|---|---|
| Direct buying action | Demo request, pricing-page visit ≥ 2, RFP download, contract-page visit | 9–10 |
| Active evaluation | G2 / TrustRadius comparison view, vendor shortlist behavior, integration page visit | 7–9 |
| Solution research | Category page reads, multiple blog posts on the same problem, webinar registration | 5–7 |
| Trigger event | Job change into a buyer role, funding round, leadership announcement, hiring spike | 4–6 |
| Passive awareness | Single blog read, one newsletter open, social engagement | 1–3 |
The score should jump when two intent clusters overlap inside a short window. A Tier 9 G2 comparison plus a Tier 9 pricing visit within seven days is not 18 — it is a Tier 10 with a manual annotation flagging the multi-signal pattern. Common Room and other modern dark funnel signal tools surface these overlaps automatically. The SIGNAL Score captures them as a single elevated intent score so the composite stays clean.
Note. A single Bombora topic surge can be triggered by an analyst researching a conference talk, not a buyer evaluating vendors. When the same week shows G2 comparison page visits plus first-party engagement, the probability of genuine buying intent jumps from roughly 25 percent on a single source to 70 to 85 percent with multi-source confirmation (industry analyst commentary, 2025). Intent strength should reward the overlap, not the loudest single source.
Factor 3 — Geography and account fit: filter out signals that cannot close
Geography and account fit is the only factor that can zero out the composite. A perfect intent signal on a 5-person company outside your territory cannot convert no matter what the other five factors say. The SIGNAL Score treats this factor as a multiplier of the intent signal, not an additive bonus. If the fit score is below 3, the composite is capped at 30 regardless of the other inputs.
Score account fit on three sub-criteria, each scored 0 to 10, then average them.
- Firmographic ICP match. Company size band, industry, revenue range, geography, tech stack. A 9 means a perfect ICP. A 4 means adjacent but unproven. A 1 means outside the ICP entirely.
- Territory eligibility. Whether the assigned rep can legally and contractually work the account. Score is binary — 10 if eligible, 0 if not. A signal on an account owned by another region zeroes this sub-score.
- Strategic priority. Whether the account is on the named target list, the ABM tier-one list, or general territory. Score 10 for named accounts, 6 for tier-one ABM, 4 for general.
The strategic priority sub-score lets sales managers bias the queue toward strategic deals without rewriting the entire model. Bump every named account by two points on the strategic sub-score and the composite shifts by about three points overall — enough to move a borderline signal above the human-outreach threshold, not so much that ICP discipline collapses.
Factor 4 — Newness and recency: model signal decay correctly
Buying signals decay. The SIGNAL Score models that decay as an exponential decline anchored to a half-life that varies by signal type. A job change holds its value for 60 to 90 days. A pricing-page visit decays inside 7 to 14 days. A demo-tool installation can decay inside 72 hours. Working a stale signal is worse than skipping it because the rep burns time on a window that has already closed.
Use the half-life table below to set the Newness factor. The score at time zero is 10. The score at one half-life is 5. The score at two half-lives is 2.5. After three half-lives, the signal scores 1 — auto-archive it.
| Signal type | Half-life | Auto-archive after |
|---|---|---|
| Demo request, RFP download | 3 days | 9 days |
| Pricing page visit, contract page visit | 7 days | 21 days |
| G2 comparison view, competitor review | 10 days | 30 days |
| Integration page visit, multiple blog reads | 14 days | 42 days |
| Trigger event — funding round, leadership announcement | 30 days | 90 days |
| Job change into a buyer role | 45 days | 135 days |
The half-life numbers come from operator-side benchmarks across signal-based selling teams, not from the platforms that sell the signals. Platform vendors often present a six-month-old job change as a fresh trigger because their dashboards do not penalize age. The SIGNAL Score does. By day 90 of a job change, the new executive has already chosen their first three vendors. Working the signal on day 14 lands the rep inside that selection window. Working it on day 90 lands them in a renewal conversation 12 months out.
Factor 5 — Activity volume: when one signal becomes a pattern
Activity volume scores the number of correlated signals on the same account inside the past 14 days. The factor is intentionally the lowest weight in the composite — volume is the easiest factor to game with noisy traffic, scraper bots, or a single curious analyst — but it is a useful tiebreaker between two otherwise-similar signals.
Use the rubric below. Volume scores compound only when the underlying signals are independent. A G2 view, a pricing visit, and a job change inside the same 14-day window scores high. Three pricing-page visits from the same IP in the same hour does not — that is one signal with three logs.
| Independent signals in past 14 days | Score | Interpretation |
|---|---|---|
| 5 or more | 10 | Active evaluation in progress — page reps now |
| 3 to 4 | 7–8 | Pattern forming — open a sequence |
| 2 | 5 | Possible early evaluation — monitor and warm |
| 1 | 3 | Single signal — let other factors decide |
| 0 in past 14 days, signal is the first | 2 | Cold start — only act if intent strength is 9+ |
The pattern matters more than the count. Signal-based outreach teams that work at scale almost always discover the same thing in their data — accounts that close generate three to seven independent signals across a 14 to 21 day window before a deal cycle opens. That window is the single richest moment in the entire pipeline.
Factor 6 — Lead role: economic buyer beats end user every time
Lead role scores the decision authority of the person on the signal. The factor exists because a pricing-page visit by a CFO is not the same event as a pricing-page visit by a junior engineer. The first warrants an outreach inside the hour. The second warrants a passive nurture.
| Role band | Examples | Score |
|---|---|---|
| Economic buyer | VP, SVP, C-level, founder, owner of the budget line | 9–10 |
| Champion | Director, Head of, manager who owns the problem the product solves | 7–9 |
| Influencer | Senior IC, technical evaluator, internal recommender | 5–7 |
| End user | IC who will use the product but does not own the budget | 3–5 |
| Unknown or anonymized | De-anonymized account, no contact identified | 4 (default) |
The default of 4 for unknown roles matters more than it looks. Many third-party intent signals land at the account level with no identified contact. The default keeps the signal in the queue without inflating it. BDRs can lift the score manually once they identify the right contact through enrichment — the framework rewards research, not guesswork.
Composite score and the outreach SLA decision tree
The composite score is the weighted sum of the six factors. The score then maps directly to an outreach SLA — a hard contract between RevOps and sales about how fast the queue gets worked. Without the SLA, the score is just a number. With the SLA, the score becomes an operating system.
| Composite score | Tier | Outreach action | SLA |
|---|---|---|---|
| 85–100 | Tier 1 — Page now | Personal call plus same-day email and LinkedIn touch | 60 minutes from score creation |
| 70–84 | Tier 2 — Personal outreach | Researched email plus follow-up call inside 48 hours | 4 business hours |
| 55–69 | Tier 3 — Sequenced outreach | 4-touch multi-channel sequence opened by the rep | 24 business hours |
| 40–54 | Tier 4 — Light nurture | Automated 2-touch sequence, no human time | 72 business hours |
| Below 40 | Tier 5 — Monitor | Add to a watchlist, re-score on next signal | None |
The threshold numbers above are starting points. Calibrate them against historical pipeline data inside the first 30 days. Pull the last 90 days of closed-won deals, score the earliest detectable signal on each, and find the 25th percentile of those scores. Set the Tier 1 threshold at that percentile — every signal at or above it should warrant the same outreach speed that produced past wins.
Verdict. The SIGNAL Score is the model to build when the team already has signal volume and needs governance. Skip it when total surfaced signals are under 50 per rep per week — at that volume, a manual triage call beats a scoring layer. Use it once the queue exceeds 100 signals per rep per week and the rep is freezing instead of working the top of the list.
How Gangly operationalizes the SIGNAL Score
Gangly runs the SIGNAL Score as the default ranking layer inside the sales workflow system. The platform ingests signals from first- and third-party detectors, scores them on all six factors automatically, and pushes the result into the rep queue with the recommended outreach action attached. The score is visible. The component breakdown is visible. The rep can override any factor with a one-line note that flows back into the model for the next calibration cycle.
The visible score matters because reps refuse to trust black-box rankings for long. The platforms that hide their scoring logic — and most of the third-party intent tools do — see rep adoption decay inside two quarters. The Common Room team makes the same observation in their own product positioning: visibility into why a signal scored the way it did is what keeps reps inside the workflow (Common Room product documentation, 2025).
When a signal lands in Tier 1, Gangly opens an outreach draft in the queue for the rep with the signal context, the account brief, and the recommended channel pre-filled. The rep edits, sends, and the system logs the outcome against the score so the calibration loop closes. Over 90 days, the historical win rate per tier becomes the audit trail that proves the score works — or proves it needs to be retuned.
Pro tip. Tie part of rep compensation to closed-won deals from scored signals rather than to total touches. If you measure activity, reps will inflate activity. If you measure outcomes per score tier, the framework stays honest. See the buying signal glossary entry for the underlying terminology and the buying signals in B2B guide for the broader playbook.
Seven mistakes that break a signal scoring framework
Most signal scoring frameworks fail the same way. The model is built once, never recalibrated, and reps either stop trusting it or learn to game it. The seven mistakes below show up in nearly every team that has run a scoring layer for more than two quarters.
- No calibration against closed-won data. The thresholds are set on intuition instead of history. Fix: pull the last 90 days of won deals, score the earliest detectable signal on each, and anchor the Tier 1 threshold at the 25th percentile.
- Weights that never change. Markets shift inside two quarters. Fix: run a quarterly weight review and adjust based on what predicted pipeline in the trailing 90 days.
- Signal sources counted equally. A Bombora surge gets the same score as a first-party demo request. Fix: lock the source tier table in §source-quality-tier and apply it as a multiplier, not an additive bonus.
- No decay model. Signals stay in the queue forever. Fix: apply the half-life table in §newness-recency and auto-archive on three half-lives.
- Black-box composite the reps cannot read. The platform shows a 78 with no breakdown. Fix: expose all six factor scores, plus a one-line annotation per signal.
- Compensation tied to activity instead of outcomes. Reps inflate touches to hit MBOs. Fix: tie a portion of variable comp to closed-won from scored signals, not raw activity counts.
- No SLA paired with the score. The number exists but no one acts on it. Fix: publish the tier-to-SLA table in §composite-score-sla as a contract between RevOps and sales, audited weekly.
Watch out. The single most common failure pattern is the third one — counting all sources equally. It looks harmless on a dashboard. It crushes rep trust the first week a Tier-5 inferred signal scores higher than a Tier-1 demo request. Fix the source tier table before anything else.
A 30-day rollout plan for the SIGNAL Score
The rollout below assumes a team that already has signal data flowing — first-party web events, a third-party intent provider, and CRM-level account data. Add two weeks to the timeline if any of those pipes need to be built first.
- Days 1–3 — Pull the calibration baseline. Export every closed-won deal from the last 90 days. For each, find the earliest detectable signal in the data and record its source, type, age, account fit, activity volume, and contact role. This is the calibration set for every weight in the model.
- Days 4–7 — Build the source tier table. Sit down with RevOps and tag every signal source feeding the CRM. Assign each one a tier from 1 to 5 using the rubric in §source-quality-tier. Commit the table to source control so it cannot drift.
- Days 8–14 — Wire the six factor calculators. Implement each factor as a separate function so it can be tested in isolation. Newness uses the half-life table. Geo and account fit uses the territory map and ICP rules. Lead role uses the contact enrichment field. The other three pull from the signal record directly.
- Days 15–21 — Run the composite in shadow mode. Score every incoming signal but do not route off the score yet. Compare the score against rep judgment on a sample of 100 signals per day. The disagreements are the calibration data — investigate the top 20 and adjust weights.
- Days 22–25 — Set the SLA contract. Publish the tier-to-SLA table to sales leadership. Confirm staffing levels can hit the Tier 1 SLA of 60 minutes during business hours. If they cannot, raise the Tier 1 threshold until they can — a missed SLA is worse than no SLA.
- Days 26–28 — Switch to live routing. Route signals off the composite score and the SLA tier. Track the time from signal creation to first rep touch as the leading indicator. Watch for queue freezing or signal cherry-picking — both indicate the threshold needs a small adjustment.
- Days 29–30 — Lock the audit cadence. Calendar a weekly review of the top 50 highest-scoring signals and a quarterly recalibration of weights. Tie a portion of rep variable compensation to closed-won deals from Tier 1 and Tier 2 signals. Start the next 90-day calibration cycle.
- Calibration baseline pulled from 90 days of closed-won deals
- Source tier table reviewed and locked in version control
- Six factor calculators implemented and unit-tested
- Shadow mode produced agreement above 70 percent with rep judgment
- SLA contract signed by sales leadership and staffing confirmed
- Weekly top-50 audit and quarterly weight recalibration on the calendar
Teams that follow this rollout typically see Tier 1 connect rates 2x to 3x the average inbound connect rate inside 60 days, based on operator reports from signal-based selling implementations (Forrester intent data research, 2025; Demand Gen Report intent data benchmarks, 2025). The lift comes from speed and ranking, not from any single magic signal source.
Want to see the SIGNAL Score running on real signals? Book a 20-minute demo or start a free trial and route your first scored signal into the rep queue this week. Related reading on the broader motion: buying signals in B2B, signal-based outreach, and the buying signal glossary entry. Sources cited inline: Gartner — How to Purchase Intent Data, Forrester — How to Evaluate Intent Data Providers, Bombora — Company Surge integration, Breadcrumbs — B2B Lead Scoring Framework, and Salespanel — Lead Scoring vs Signal-Based Selling.