Feature · B2B Scoring Methodology

From qualification framework to scoring model, step by step.

Named B2B qualification frameworks (BANT, MEDDPICC, CHAMP, SPICED) are mental models. A scoring methodology turns the mental model into a math model: specific signals, specific weights, specific calibration rituals. Below is the build guide.

Start free trial See pricing
The frameworks and their scoring translation

Four qualification frameworks mapped to math.

Named B2B qualification frameworks give sales teams a shared vocabulary. BANT, MEDDPICC, CHAMP, and SPICED each name the dimensions that matter. The methodology below translates each framework into specific scoring components. Teams running a framework in rep training should run the matching scoring model in CRM.

BANT scoring

Budget, Authority, Need, Timing.

BANT translates to four score components: Budget (fit score from firmographics plus confirmed budget signal), Authority (committee role identification plus economic buyer confirmation), Need (pain signal from discovery notes plus intent signal), Timing (projected close date plus urgency signal). Each dimension carries a configurable weight. BANT works best for high-velocity transactional B2B motions.

MEDDPICC scoring

The complex sale framework.

MEDDPICC adds Metrics, Economic buyer, Decision criteria, Decision process, Paper process, Identify pain, Champion, Competition. The scoring model expands to eight components with higher weights on Champion identification and Economic buyer confirmation. MEDDPICC works for enterprise sales with 90-plus-day cycles and multi-stakeholder committees.

CHAMP scoring

Challenges first, then Authority, Money, Prioritization.

CHAMP leads with Challenges (pain signal) as the primary qualifier. Scoring weights CHAMP challenges higher than BANT budget as the leading signal. The approach fits consultative B2B sales where discovery produces the deal rather than the lead inherently qualifying. Scoring Model Design: challenges 40 percent, authority 25 percent, money 20 percent, prioritization 15 percent.

SPICED scoring

Situation, Pain, Impact, Critical event, Decision.

SPICED (from Winning by Design) emphasizes Impact and Critical Event as the forcing-function components. Scoring weights the critical event signal heavily; a deal with a named critical event (CFO board review in 60 days, system end-of-life in 90 days) scores dramatically higher than one without. SPICED fits enterprise SaaS sales with champion-led buying journeys.

Hybrid approach

Framework for training, model for scoring.

Most mid-market B2B teams run a hybrid: the rep training uses one framework (often MEDDPICC) for discovery discipline; the scoring model implements the signals that the framework names. The two layers reinforce each other. The rep thinks in MEDDPICC during discovery; the CRM scores on the MEDDPICC-derived signals. The team speaks one language.

Framework-agnostic signals

Three signal types every framework uses.

Across all four frameworks, three signal types recur: firmographic fit (ICP match), behavioral intent (what the buyer is doing), and qualified-by-human (what the rep has learned). The scoring model always has these three components; the frameworks just differ in which specific sub-signals matter inside each component. Starting from the three gives a solid baseline before layering a framework.

The six-step build process

From blank CRM to live scoring in two weeks.

Building a scoring model from scratch takes two to three weeks when done methodically. The six-step process below is the sequence we see work across the 23 mid-market B2B teams Strkr has worked with on scoring methodology deployment. Skipping any step produces drift within the first quarter.

Step one

Audit the won deals.

Pull the trailing 24 months of closed-won deals. For each, document: industry, employee count, geography, deal size, cycle length, lead source, and the dominant qualification narrative (what signal led to the win). The pattern across 40-200 won deals reveals the actual ICP, which is often different from the stated ICP. Two to three days of ops lead time.

Step two

Audit the lost deals too.

Pull the same trailing 24 months of closed-lost deals. For each, document the same fields plus the lost reason. The pattern reveals which signals produced false positives in the pre-model world. Classic insight: deals with Industry X and Size Y looked promising but lost at 70-percent rate. Those deals need lower scoring weights going forward.

Step three

Pick the framework.

Based on sales cycle length and complexity, pick BANT, MEDDPICC, CHAMP, or SPICED. The pick drives training vocabulary alignment. Transactional sub-30-day cycles: BANT. Enterprise 90-plus-day cycles with named champion: MEDDPICC. Consultative with discovery-led deals: CHAMP. Enterprise SaaS with critical-event motions: SPICED. One day of exec alignment.

Step four

Define the component weights.

Map the framework dimensions to the three Strkr scoring components (ICP fit, intent, engagement) with specific weights. Example MEDDPICC mapping: ICP fit 50 percent (Metrics plus Decision criteria plus Competition fit), intent 30 percent (Pain plus Paper process signals), engagement 20 percent (Champion identification plus Economic buyer confirmation). Two to three days of model design.

Step five

Backfill and calibrate.

Run the proposed model against the trailing 12 months as a dry run (no record writes). Review the score distribution. Target: 10-15 percent Hot, 25-35 percent Warm, 30-40 percent Cool, 15-30 percent Cold. If the distribution is skewed (every lead scores Hot, or nothing scores Cold), the weights are wrong. Tune and re-run. Three to five days.

Step six

Deploy with monitoring.

Enable the model on new leads first; the backfill brings existing leads to the new model. Monitor the win-rate-by-band metric for the first full quarter. The band gap (Hot win rate minus Warm win rate) should be 15-25 percentage points if the model is calibrated. Below 10, recalibrate. The first quarter is tuning; by quarter two the model is stable.

The common calibration mistakes

Six scoring errors that recur predictably.

Across the 23 mid-market B2B teams we have observed on scoring deployments, six specific mistakes recur. Each has a defense. Teams that know the mistakes catch them in week one; teams that do not spend a quarter working through the symptoms before finding the cause.

Mistake one

Weighting what you can measure.

Teams weight firmographic signals heavily because they are easy to capture (enrichment provides them) and under-weight the human-qualified signals because capturing them requires rep discipline. Result: the model scores well-fit uninterested leads above interested but less-well-fit ones. Defense: force equal attention to the rep-qualified components. The MEDDPICC Champion identification is more predictive than any firmographic.

Mistake two

No decay on engagement.

Teams build the ICP and intent components but skip engagement decay. Result: a stale high-fit lead from eight months ago clogs the Hot queue indefinitely because nothing ages it out. Defense: engagement decay is mandatory. The half-life depends on cycle length; the presence of decay is non-negotiable.

Mistake three

Overfitting on last year.

The audit of won deals produces a tight ICP that fits the trailing 12 months perfectly. Then the market shifts (upmarket motion, new vertical, new product tier) and the model is wrong. Defense: track model performance quarter by quarter; recalibrate on 15-point band-gap drops. The engine's calibration dashboard surfaces this.

Mistake four

No committee aggregation.

Teams score contacts individually without rolling up to the account. B2B deals close at the account level with a committee. The symptom: five Hot contacts at one account produce no account-level signal because the model never aggregated. Defense: account-level aggregation is mandatory for B2B. The committee shape is itself a scoring input.

Mistake five

No per-segment calibration.

Teams run one model across SMB and enterprise. The signals that matter in SMB (fast pricing-page activity) differ from the signals that matter in enterprise (champion identification over months). The symptom: the model is OK for one segment and terrible for the other. Defense: segment the model when the win-rate variance across segments exceeds 10 points.

Mistake six

No AE feedback loop.

Teams build the model in isolation from the AEs who will use it. The AEs see the Hot leads, disagree with the ranking, and quietly stop trusting the model. Defense: AE feedback is a monthly ritual. The engine captures AE override events (AE marked a Hot lead as "not actually hot"); the pattern surfaces miscalibration.

The signals that matter most

Ten signals across the three components.

Not every signal is created equal. Across the win-rate measurements from 23 mid-market B2B teams, ten specific signals account for roughly 80 percent of the predictive value. The breakdown below assigns each signal to one of the three scoring components with the typical lift and the right weight range.

ICP: industry match

Baseline 2.3x win lift.

Industry match to the ICP produces the baseline 2.3x win-rate lift versus industry mismatch. The signal is foundational. Weight range: 10-15 points of a 100-point ICP component. Teams with narrow verticals weight higher; horizontal plays weight lower. The signal fires at lead creation from enrichment.

ICP: size match

Baseline 1.9x win lift.

Company size match (employee count band or revenue band) produces 1.9x win lift. Weight range: 10-15 points of the ICP component. The band depends on the ICP: for mid-market SaaS, 50-500 FTE is a typical band. The signal fires at lead creation and refreshes with enrichment updates.

ICP: geography

Baseline 1.4x win lift (coverage-dependent).

Geography match (in-coverage vs out-of-coverage) produces 1.4x lift for most B2B teams; the lift is higher for teams with strict territory coverage. Weight range: 5-10 points of the ICP component. The lift is lower than industry or size but still material. The signal fires at lead creation.

Intent: pricing page

Baseline 3.1x win lift.

A pricing page view in the trailing 7 days produces 3.1x win lift. The signal is the single highest-ROI intent signal in B2B. Weight range: 10-15 points of the intent component. Repeat views within the window compound the signal. The surface should be a dedicated pricing page, not a buried subpage.

Intent: demo page

Baseline 4.2x win lift.

A demo-request page view (without actually requesting) produces 4.2x win lift. The signal captures the pre-request browse pattern that precedes most inbound requests. Weight range: 10-15 points. Combined with pricing page views, the two together produce a compound signal that is nearly deterministic for the Hot band.

Intent: content download

Baseline 1.6x win lift per download.

Content downloads (whitepapers, guides, calculators) produce 1.6x win lift per download, with diminishing returns past three downloads. The signal is lower than pricing or demo views but captures the research-mode buyer. Weight range: 3-5 points per download, max 15 points.

Qualified: champion identified

Baseline 2.8x win lift.

A rep-tagged Champion on the account produces 2.8x win lift. The signal is only available after discovery, so it only applies to leads past the initial qualification. Weight range: 15-20 points in the committee/engagement component. The signal is one of the strongest in enterprise B2B.

Qualified: economic buyer confirmed

Baseline 3.4x win lift.

A rep-tagged Economic Buyer on the account (confirmed, not inferred) produces 3.4x win lift. The signal is the single strongest in enterprise B2B (MEDDPICC). Weight range: 20-25 points in the committee component. The signal fires after a discovery step.

Qualified: critical event named

Baseline 2.1x win lift.

A named critical event (board review, product end-of-life, compliance deadline) produces 2.1x win lift. The signal reflects urgency, which is the force multiplier on all other signals. Weight range: 10-15 points in the engagement component. SPICED methodology weights this signal heavily.

Negative: ghosting

No response 14 days drops score.

A lead that stops responding for 14 days triggers a score decay step. The signal reflects buyer disengagement. The engine does not reset the score to zero; it decays by a configured step. Teams typically configure a 15-point drop for 14-day no-response, another 15 points for 30-day no-response.

The ongoing calibration ritual

Four reviews that keep the model honest.

A scoring model deployed and forgotten drifts into noise. The ritual below is the cadence that keeps the model healthy. Teams that run the rituals catch drift in the quarter it happens; teams that skip them wake up to a model that lost rep trust six months before anyone noticed.

Weekly

Open the health dashboard.

Each week, the scoring admin opens the model health dashboard for 10 minutes. Review: win-rate-by-band, exception queue, AE override count, drift signals. Fifteen minutes. Catches drift at the week level, not the quarter level. The habit is small; the compounding is large.

Monthly

AE feedback session.

Each month, the scoring admin runs a 30-minute feedback session with 3-4 AEs. Review: specific scored leads the AEs disagreed with, specific signals the AEs want weighted differently, patterns in the override events. The session surfaces trust issues before they become quiet-stops-using-the-score issues.

Quarterly

Full model recalibration.

Each quarter, the scoring admin runs a full recalibration against the trailing quarter data. Review: did the band gap hold at 15-25 points, did any signal lose predictive value, did the committee mapping discipline hold up. Adjust weights or add signals as needed. The recalibration takes a day; the output is a stable model for the following quarter.

Annually

Framework and ICP review.

Once a year, the scoring admin plus the CRO plus marketing leads review the ICP itself. Did the win zone shift materially? Did the team move upmarket or downmarket? Does the framework still fit? The annual review is strategic; it catches the market shift that quarterly calibration smooths over.

On market events

Pricing changes, product launches, new segments.

When the business changes (new pricing, new product tier, new vertical push), the scoring model needs an explicit review. New product tier might shift the ICP size band; new vertical push might add a new segment-specific model. The event-triggered review catches the business change at the model level before it causes drift.

On rep turnover

Train the new reps on the model.

When new reps join the team, the scoring model orientation is part of onboarding. The hover-explanation feature makes self-service onboarding possible, but the admin should walk a new rep through five sample Hot leads and five Warm leads to anchor expectations. Trust in the model starts with understanding; investment in onboarding compounds trust.

Walk your scoring methodology with us in a 45-minute session.

Bring your qualification framework and your ICP document. We will map the components, pick the weights, and run a backfill against sample data.

Book a session See all scoring features
Common questions

What buyers ask about this feature.

Which framework should a mid-market B2B team pick?

The pick depends on cycle length and complexity. For sub-30-day cycles with transactional motions, BANT works. For 60-90-day mid-market cycles with 3-5 person committees, CHAMP or MEDDPICC with lighter weighting on the paper process components. For 90-plus-day enterprise cycles with 5-7 person committees and formal procurement, MEDDPICC with full weighting. For enterprise SaaS with champion-led buying journeys and critical events, SPICED. Most mid-market B2B teams land on MEDDPICC with slight adjustments. The pick matters less than the discipline; any of the four will work if the team sticks to the discipline.

How many signals is the right number to track?

Ten to fifteen signals cover 80 percent of the predictive value for most B2B teams. Beyond fifteen, the marginal signal adds complexity faster than predictive value. Below ten, the model is probably missing a strong signal (champion identification, pricing page views, industry match). The ten signals in the previous section are the baseline set; most teams add one or two vertical-specific signals (healthcare teams add BAA-signal, SaaS teams add stack-detection signals) and remove one or two that do not fit their motion.

How quickly should we see the model stabilize after deployment?

The model should stabilize by end of quarter two. Quarter one is tuning (watching the band gap, adjusting weights, catching the segment-specific drift). Quarter two is the first stable quarter. By quarter three, the AE feedback is converging and the model is producing confident predictions. If the model has not stabilized by end of quarter two, the issue is usually either a wrong framework pick (BANT when enterprise MEDDPICC was needed) or a missing component (no committee aggregation on a B2B motion). The stability is a tuning sign; the right model lands quickly if the inputs are right.

What happens when the business pivots and the ICP shifts dramatically?

A full ICP shift (upmarket from SMB to mid-market, new vertical push, acquisition that changed the product mix) requires a model refresh, not a tune. The refresh runs the full six-step build process again: new won/lost audit against the new segment, new framework pick if cycle changed, new component weights, backfill, deploy. The old model stays in version history for comparison. Expect six to eight weeks for a full refresh. Teams going through ICP pivots should plan the scoring refresh as part of the pivot, not after.

Can one team run multiple scoring models simultaneously?

Yes. The engine supports per-segment scoring models with explicit routing rules that assign leads to the right model at creation. Teams with distinct motions (SMB inbound plus mid-market outbound plus enterprise strategic) run three models in parallel. Each model has its own weights, bands, and calibration cadence. The admin surface shows the models side by side with per-model metrics. The complexity cost is real; teams should run multiple models only when the motions are genuinely different, not just for variety.

Is this HIPAA-compliant for healthcare lead scoring methodology?

Yes for teams on the BAA. The scoring methodology does not touch PHI directly; it reads firmographic data, behavioral intent (page views), and rep-qualified signals (champion identified). None of these are PHI in standard B2B scoring. For teams where the ICP itself is PHI-adjacent (healthcare provider marketing, telehealth lead generation), the standard tenant encryption plus access logging covers the compliance posture. SOC 2 Type II attestation is in progress for 2026-Q4. GDPR and CCPA data-subject requests cover scored leads on the standard SAR workflow.

Related features

Pair this with...

B2B Lead Scoring B2B Lead Scoring Criteria Examples Lead Scoring Models Lead Scoring Criteria Examples Lead Scoring Engine Criteria Examples Automated Lead Scoring Lead Scoring

Try it free. Bring your team next week.

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.

Start free trial See pricing