A revops lead at a mid-market B2B SaaS pulled the lead scoring criteria from Salesforce and found 22 signals weighted by the previous VP of Marketing in 2022. Three signals had been retired from the forms years ago. The score did not correlate with closed-won anymore. Strkr runs priority score criteria calibrated against your actual closed-won data with transparent weights, retirement rules for drifted signals, and ongoing model refresh.
Lead priority scoring works when the criteria match what actually predicts revenue in your motion. Most teams run scoring criteria that drifted from reality over multiple VP transitions. The six pains below are what the drift looks like.
Lead score weights set by the founding marketer rarely get recalibrated. The business grew, the ICP shifted, the sales motion changed, but the scoring model stayed static. Reps lose trust in the score because the top-scored leads do not actually close. Strkr recalibrates against closed-won data monthly so the model stays current.
Scoring criteria often reference fields that forms stopped capturing. A rule says "add 15 points if timeline is this_quarter" but the timeline field retired 18 months ago. The rule fires zero times and no one notices. Strkr audits scoring criteria against live form fields monthly so dead rules retire.
Weights get set by intuition rather than correlation. A pricing-page view might carry 10 points when the data shows it correlates with closed-won strongly enough to warrant 40 points. Strkr weights calibrate against actual closed-won correlation so the score reflects reality.
A score without visible drivers is a black box. Reps distrust the number because they cannot see what drove it. Strkr exposes the top 3 drivers of every score inline on the record so reps and managers understand the reasoning.
Collapsing firmographic fit (is this an ICP company) and behavioral intent (did they engage) into one number loses nuance. A high-fit low-intent lead needs nurture; a low-fit high-intent lead needs disqualification. Strkr separates fit score and intent score so routing and sequencing can act on both.
Scoring often adds points for positive signals without deducting for negative signals (personal email domain for B2B, free-email submissions, invalid phone numbers, obvious spam patterns). Positive-only scoring inflates scores on bad leads. Strkr includes negative weights for signals that predict non-conversion.
The nine cards below describe the signal categories that make up a well-calibrated priority score. Each category combines through weighted additive scoring.
Firmographic signals score ICP fit: employee count, revenue tier, industry (NAICS), geography, business type. Weights calibrate against which firmographic patterns correlate with closed-won. A software company in the US with 100 to 500 employees might score high; a sub-10-person retail business might score low or disqualify.
Behavioral signals score intent: pages viewed in the past 30 days, time on site, repeat visits, pricing page view, demo page view, case study view, content downloads. Pricing-page views typically carry high weight because they correlate with closed-won.
Email engagement scores nurture responsiveness: opens, clicks, replies, forwards, unsubscribes. High engagement on recent emails signals active evaluation. Unsubscribes or spam-reports flag negative signals for score deduction.
Form submission history scores the depth and recency of engagement: content downloads, webinar registrations, demo requests. Multiple form submissions in a short window suggest active evaluation. A demo-request submission carries higher weight than a content download.
Third-party intent data (G2 Buyer Intent, 6sense, Bombora) scores category research. A prospect researching the category on G2 or showing in-market signals on 6sense carries elevated score. Intent data complements first-party behavioral data.
Qualification data (budget range, timeline, decision-maker identification) scores readiness. Identified decision maker plus near-term timeline typically correlates strongly with closed-won. Qualification data captures through sales activity rather than forms.
Negative weights deduct for signals that predict non-conversion: personal email domain (gmail, yahoo, outlook) for B2B motion, invalid or missing phone, free-email submissions, obvious spam patterns in company name field, matched to no-serve geographies.
Signals from 90 days ago carry less weight than signals from the past 7 days. Pricing-page view 3 months ago is informative but less predictive than one from this week. Recency decay prevents stale signals from inflating current scores.
Account-level signals score the broader buying context: multiple contacts from the same account showing engagement, named-account match, account tier (strategic vs standard), account history (prior customer, closed-lost reopen). Account signals inform individual lead priority within account context.
The six cards below describe how to calibrate scoring weights against actual closed-won data. The pattern replaces intuition with correlation.
Pull the last 12 months of closed-won deals with the full signal data captured at various stages (lead creation, MQL promotion, SQL promotion, opportunity creation). The sample should include 50 to 300 deals depending on volume. Smaller samples work with narrower weight bands.
Pull a matched sample of closed-lost deals (same volume as closed-won). The comparison between closed-won and closed-lost reveals which signals discriminate winners from losers. Signals that appear equally in both samples do not warrant weight.
Correlation analysis measures how often each signal appears in won deals versus lost deals. A signal that appears in 60 percent of won deals and 20 percent of lost deals discriminates strongly. A signal that appears equally in both does not.
Assign weights proportional to correlation strength. Strong discriminators (correlation ratio over 3) get higher weights; weak discriminators (ratio under 1.5) get low or zero weight; negative discriminators (more frequent in lost than won) get negative weights.
Define score thresholds mapped to action: 0 to 25 (cold, nurture); 25 to 50 (warming, continue nurture); 50 to 75 (MQL, route to sales); 75-plus (high priority, SLA-tightened). Thresholds calibrate against conversion rate at each range so the actions make sense.
Monthly recalibration pulls new closed-won and closed-lost data, re-runs correlation analysis, and updates weights. Signal drift (a signal that was predictive 2 years ago but is not now) catches on refresh. The model stays current rather than static.
The six cards below describe how score tiers translate to specific sales and marketing actions. The score drives behavior, not just prioritization order.
Scores above 75 route to the inbound AE pod with 5-minute first-touch SLA. The rep gets the lead notification immediately. Follow-up includes a same-day meeting booking attempt. High-priority leads represent the top of the inbound queue.
Scores 50 to 75 route to BDR for qualification call within 2 hours, then handoff to AE if qualified. The BDR qualifies on BANT or MEDDIC criteria. Qualified leads promote to SQL; unqualified continue in nurture or disqualify.
Scores 25 to 50 enter nurture cadence without sales touch. The nurture presents content matched to interest signals, aiming to raise the score through engagement. If engagement rises the score past 50, the lead promotes to MQL and gets sales touch.
Scores 0 to 25 enter minimal cadence (monthly newsletter, major announcements). The volume is high but the economics do not warrant active investment. Re-engagement campaigns periodically test whether cold leads have warmed.
Negative scores (from negative signal accumulation) flag for disqualification review or auto-filter as spam. Personal-email B2B submissions with invalid phone and spam-pattern company name typically land below zero and never reach a rep. The spam filter saves rep attention.
A sudden score jump (e.g., prospect at 32 views pricing page and jumps to 68) triggers an alert to the account owner or inbound rep. The jump signals active evaluation and the real-time alert captures the window of attention. Standard batch-scored systems miss this.
Most CRM scoring runs on templates set up once and rarely recalibrated. Here is the head-to-head when the question is dynamic, calibrated scoring.
| Feature | Strkr | Static vendor-template scoring |
|---|---|---|
| Calibration data | Your closed-won and closed-lost | Vendor template averages |
| Recalibration cadence | Monthly refresh | Set once, rarely updated |
| Signal retirement | Dead-signal audit monthly | Dead signals linger |
| Weight assignment | Correlation-driven | Intuition-driven |
| Negative signals | Native with deductions | Often positive-only |
| Fit vs intent separation | Two scores, both visible | Collapsed into one |
| Driver transparency | Top 3 drivers visible on record | Black box |
| Recency decay | Native | Varies |
| Account-level signals | Native | Varies |
| Score jump alerts | Native real-time | Batch-scored, misses the window |
Five rebuilds of lead priority scoring on Strkr. Each one replaced static intuition weights with calibrated criteria.
A mid-market B2B SaaS rebuilt their 22-signal scoring model that had not been recalibrated in 3 years. The analysis found 7 signals with zero predictive power (retired), 4 signals with reversed polarity (previously positive, actually negative), and 11 signals needing weight adjustment. The rebuilt model raised MQL-to-SQL conversion by 23 percent in the following quarter.
An enterprise B2B team separated the single collapsed score into fit score (firmographic ICP match) and intent score (behavioral and intent-data signals). The separation revealed that 40 percent of inbound was high-intent low-fit (SMB prospects of an enterprise product) requiring disqualification. Rep time redirected to high-fit high-intent produced 32 percent more SQLs at identical headcount.
A B2B services firm introduced negative weights for spam patterns (personal email domain, invalid phone, obvious junk in company name field). 24 percent of inbound leads landed below the auto-filter threshold and never reached a rep. Rep queue size dropped 24 percent and MQL-to-SQL conversion rose 8 percent because reps worked a cleaner queue.
A B2B SaaS enabled real-time score-jump alerts. Prospects whose score rose by 20-plus points in a session (typically from pricing-page view or demo-request) triggered immediate alerts to the account owner or inbound rep. First-touch time on these spikes dropped from 2 hours to 4 minutes median. Connect rate rose 42 percent on alerted spikes.
A product-led SaaS integrated product usage signals (features activated, teammates invited, records created) into the priority score alongside marketing signals. The usage signals typically carried stronger correlation to closed-won than marketing signals. The combined score prioritized sales-worthy trial users accurately and lifted trial-to-paid conversion by 28 percent.
Start a 14-day trial with calibrated priority scoring, signal correlation analysis, score-jump alerts, and action tiers enabled. Replace vendor-template scoring with criteria that match your actual pipeline.
A well-calibrated priority score includes nine signal categories: firmographic fit (company size, industry, geography), behavioral engagement (pages viewed, time on site, repeat visits), email engagement (opens, clicks, replies), form submission history (which forms, when, with what intent), intent data (G2 Buyer Intent, 6sense, Bombora), qualification data (BANT or MEDDIC), negative signals (personal email, invalid phone, spam patterns), recency decay (signals age out over time), and account-level signals (buying committee, named-account match).
Monthly recalibration is the standard cadence. Pull the last 12 months of closed-won and closed-lost deals, re-run correlation analysis, update weights, and audit for retired signals. Monthly cadence catches signal drift before it degrades score reliability. Static models (set once, never recalibrated) typically become unreliable within 12 to 18 months as the business, ICP, and sales motion evolve.
Two scores. Firmographic fit (is this an ICP company) and behavioral intent (is this prospect engaged) represent different dimensions and warrant different actions. A high-fit low-intent lead needs patient nurture; a low-fit high-intent lead needs disqualification; a high-fit high-intent lead needs immediate sales attention. Collapsing both into one number loses the nuance and leads to misrouted sales effort.
Negative signals deduct from the score rather than being ignored. Common negative signals: personal email domain (gmail, yahoo) for B2B motions, invalid or missing phone, free-email submissions on gated content, spam patterns in company name field, matched to no-serve geographies, consent-withdrawn status, prior closed-lost with no-interest disposition. Negative weights prevent score inflation on bad leads and feed a spam-filter threshold below zero.
Score-jump alerting fires real-time notifications when a lead's score rises by a defined threshold in a short window (e.g., plus 20 points in one session). The jump typically indicates active evaluation (pricing-page view, demo page view, form submission). The alert captures the window of attention and routes the hot lead to the account owner or inbound rep for immediate follow-up. Batch-scored systems miss this window because they re-score weekly or daily.
Score tiers map to specific actions. 75-plus routes to AE pod with 5-minute SLA. 50 to 75 routes to BDR for 2-hour qualification call. 25 to 50 enters nurture cadence without sales touch. 0 to 25 enters minimal cadence. Negative scores filter as spam. The score triggers routing, SLA, and cadence choices automatically rather than just ordering a queue. The whole motion runs on the score rather than the score being an advisory number.
No sales call, no migration consultant, no four-month implementation. Enter your card, get 14 days of the full Pro tier, cancel any time before day 14 with zero charge. Spin up a workspace, import your CSV, and have something useful before lunch.
