Lead scoring that rewards fit, not just size, with points assigned by revenue band.
Classic lead scoring handed companies a single revenue field and asked them to invent a weighting curve. The result is a model that over-rewards F100 names and under-rewards the 10 to 50 million mid-market segment where the actual deals close. Strkr ships a revenue-range scoring methodology that lets the RevOps lead assign points by band, tune the ICP curve to the actual close data, and audit every scoring change.
Every RevOps lead who has configured HubSpot predictive scoring or Salesforce Einstein Lead Scoring has run into the same problem. The vendor sold a model that treats revenue as a linear input, when in reality the ICP curve is almost never linear. A 500 million dollar company is not five times better than a 100 million dollar company for a mid-market SaaS seller; it is actually worse because the sales cycle doubles and procurement makes the deal smaller than the mid-market version. Teams that lean on vendor defaults end up overweighting the biggest names and routing their best reps at accounts that will never close. The fix is a scoring methodology that assigns points by revenue band, not by raw revenue number, and that lets the RevOps lead calibrate the band weights against actual close-rate data. Three pain patterns show up on every scoring-model conversation.
Linear revenue weighting
The vendor default treats revenue as a straight line.
HubSpot predictive scoring and Salesforce Einstein both treat revenue as a continuous input with a monotonic curve: more revenue, more points. The curve ignores the reality that most B2B sellers have a sweet-spot band and both the lower and upper tails close less often. Strkr lets the RevOps lead define bands (under 1M, 1 to 10M, 10 to 50M, 50 to 250M, over 250M) and assign points per band, so the curve matches the shape of the actual close data.
No calibration against close data
Nobody checks whether the scoring model actually predicts closes.
A scoring model that nobody validates against outcomes is a prejudice with a dashboard. Teams spin up models in month one, forget them in month two, and the sales team starts routing on gut feel because the scoring stopped matching reality. Strkr runs a weekly calibration job that compares score bands against actual close rate over the trailing 90 days and flags bands that drifted. The RevOps lead gets a prompt to retune when the model stops predicting.
ICP sweet-spot invisible
The 10 to 50M band is where the money is and nowhere in the scoring config.
Most mid-market B2B sellers close most of their revenue in a specific band (often 10 to 50M in company revenue). The scoring config that treats every revenue number linearly dilutes this signal. Strkr lets the admin set a sweet-spot band with elevated weight so the model directs rep attention at the segment that actually converts, rather than at the F100 names that look impressive but close 2 percent of the time.
Missing revenue data
A third of your accounts have no revenue field populated.
Vendor enrichment coverage on revenue data is thin for private companies, non-US accounts, and new entities. A naive scoring model treats missing revenue as zero and buries legitimately qualified leads. Strkr supports a null-safe band ("revenue unknown") with its own weight, so a RevOps lead can either neutral-weight unknowns or elevate them based on secondary signals (employee count, funding stage) that are more consistently populated.
Compounding bias across bands
The team keeps rewarding what the model keeps rewarding.
When reps work the leads the model scores highest, the close data loops back into the model and the top band gets reinforced regardless of whether it was ever the right band. Strkr's audit log captures every band weight change with timestamp and reviewer so the RevOps lead can see when a weight shifted and why, and the calibration job surfaces when the loop is reinforcing bias rather than tracking reality.
Model opacity to the sales team
Reps do not know why a lead scored 72.
An opaque score (just the number) tells a rep nothing about why the lead qualified. Strkr renders the score with a per-signal breakdown on the lead card: 25 points from revenue band, 15 points from employee count band, 12 points from industry, 20 points from behavior. The rep sees the composition, which also lets managers coach reps toward recognizing good-fit leads rather than chasing high scores on autopilot.
How the methodology works
Points assigned per revenue band, tunable per segment.
The scoring methodology is built on five configurable primitives: revenue band definitions, points per band, band-level overrides per segment (vertical, geography, product line), a null-safe unknown-revenue band, and a calibration overlay that compares band weights against close-rate reality. Everything is admin-tunable and everything writes to an audit log.
Band definitions
Up to 12 bands with inclusive-exclusive boundaries.
The admin defines bands by lower bound and upper bound (inclusive on the lower, exclusive on the upper). Default bands ship as under 1M, 1 to 5M, 5 to 10M, 10 to 50M, 50 to 100M, 100 to 500M, 500M to 1B, over 1B. The admin edits the bands to match the business. Teams selling to the public sector add a public-sector-budget-size dimension; teams selling to startups add a stage-based cut.
Points per band
Point weight assigned independently per band.
Each band carries a point weight (default range 0 to 50, admin configurable). The admin can create a sweet-spot curve (0, 5, 15, 40, 25, 10, 5, 0) that elevates the 10 to 50M band without rewarding the F100 tail. The band weights apply to the account-level score and compound with other scoring dimensions (behavior, industry, source) to produce the final lead score.
Segment overrides
Different revenue curves per vertical or product line.
A horizontal CRM product might weight bands flat across verticals. A finserv-specific product might elevate the 100 to 500M band because that is where the compliance-driven deal size lives. Strkr lets the admin define per-segment overrides on the band weights, with the base model as the default and segment-specific curves applied when the account carries the matching segment tag.
Null-safe unknown band
Accounts with no revenue data get a configurable weight.
Missing revenue is a common data condition and the model should not punish accounts that happen to be privately held or outside the enrichment vendor coverage zone. Strkr supports a "revenue unknown" band with a configurable weight, often set at the median band weight so unknowns do not get buried. The admin can also route unknowns to an enrichment job that pulls secondary signals before scoring finalizes.
Calibration overlay
Weekly job compares band weights vs actual close rate.
The calibration job runs weekly: it joins the scoring band for each account against actual close rate over the trailing 90 days, and renders a chart showing where the model agrees with reality and where it does not. The RevOps lead reviews the overlay and retunes bands when drift exceeds a threshold. The job also writes a calibration score (0 to 1) that leadership can watch over time as a model-health metric.
Audit log
Every band weight change captured with reviewer and timestamp.
Scoring models drift because somebody tweaks a weight in month 3 and nobody remembers why by month 8. Strkr logs every band weight change with the user who made it, the before and after value, and the free-text rationale the user entered. The audit log is reviewable by compliance and by the next RevOps lead when the current one leaves, which is the thing a scoring methodology needs to survive turnover.
Score composition breakdown
The lead card renders the per-signal contribution.
Every lead card in Strkr shows the final score and the breakdown: how many points came from revenue band, how many from employee count, from industry, from behavior, from source. The rep sees the composition. The manager uses the composition in coaching: reps who chase high scores without reading the composition miss that a 72-score lead with 60 points from behavior and 12 from fit is actually a bad fit that is just engaging a lot.
Export to warehouse
Band assignments stream to the data warehouse hourly.
For teams that run attribution modeling or lifetime-value analysis outside the CRM, Strkr streams the band assignments and the component scores to a data warehouse hourly via a built-in CDC export. The downstream analyst can join scoring data against product usage, support tickets, revenue outcomes, and close the loop between the model and the business reality.
Backward compatibility with legacy fields
Import existing scoring data from HubSpot or Salesforce.
Teams migrating from HubSpot predictive scoring or Salesforce Einstein can import existing lead scores as a legacy field alongside the Strkr methodology. The RevOps lead runs both in parallel for 30 to 60 days, compares outcomes, and switches cutover when the Strkr model demonstrates equal or better predictive accuracy on the calibration overlay. The dual-run period keeps the sales team confident during the switch.
Tuning the model
What to adjust when the calibration drifts.
Scoring models are not set-and-forget. They drift as the business shifts, as the ICP narrows, as the comp plan changes. The cards below are the common tuning actions a RevOps lead takes when the weekly calibration job flags a drift.
Sweet-spot moved
Close rate shifted to a different band.
When the close-rate overlay shows that the 50 to 100M band now outperforms the 10 to 50M band that was previously the sweet spot, the model needs retune. Common cause: product moved upmarket in a maturity shift. The RevOps lead rebalances the weights (shift 10 points from the old sweet spot to the new) and lets the model run for 30 days before re-reviewing. The audit log captures the rationale.
Unknown-band bias
The unknown-revenue band converts higher than you weighted it.
If unknown-revenue accounts are closing at 18 percent and your scoring weighted them at the 2 percent band weight, you are burying good leads. Rebalance the unknown-band weight upward and consider launching an enrichment job to resolve the unknowns rather than leaving them bucket-scored. Teams that enrich the unknown band in week one often discover the enrichment payback alone covers the vendor cost.
Segment override needed
One vertical closes a totally different band than the rest.
A horizontal model that works for most verticals might underperform in one specific vertical where the deal shape is unusual. Finserv teams buying infrastructure, for example, close in a different band than mid-market SaaS. Add a segment override for that vertical with a vertical-specific curve rather than retune the base model and dilute the signal for everyone else.
Weight compression
The top-band weight is too high and nothing else matters.
If the top revenue band carries 50 points and the second band carries 10, the model is effectively just "big accounts win." That is not scoring, that is sorting. Compress the gap: bring the top band down and the middle bands up. The scoring curve should produce a meaningful distribution across the whole lead population, not a bimodal split.
Behavior vs fit rebalance
Reps are chasing engagement over fit.
If reps are closing a lot of 72-score leads where 60 points came from behavior and 12 from fit, the model is rewarding engagement over fit. The composition breakdown on the lead card helps the manager coach around this, and the admin can shift the component weights to elevate fit relative to behavior at the aggregate level. The dial is one slider in the admin console.
Decay on behavior signals
Last-touch recency weight is too flat.
A behavior signal that fired 90 days ago should count less than one that fired yesterday. If the model treats all engagement equally the score inflates on dormant leads. Strkr supports a half-life decay on behavior signals (default 30 days) with admin tunability. Teams that implement decay typically see score compression on dormant leads and clearer routing to the live ones.
What the methodology will not solve
The honest limits of revenue-range scoring.
Revenue-range scoring is a strong primitive but it is one input in a lead-scoring stack, not the whole thing. The limits below show up at every org that leans too hard on this methodology without pairing it with the other signals that matter.
Not predictive of actual intent
Revenue tells you fit, not readiness.
A perfect-band account that is not currently shopping will not close this quarter. Revenue-range scoring measures fit, which is a necessary condition for a close but not a sufficient one. Pair the revenue bands with intent signals (G2 category traffic, Bombora topic clusters, in-product usage for product-led motions) to get a composite score that predicts both fit and readiness.
Data quality dependency
The model is only as good as the revenue field.
If the revenue field on 40 percent of accounts is stale, scraped from a decade-old filing, or wrong, the model is scoring on fiction. Strkr tracks per-field recency and surfaces stale revenue data in a scoring-health dashboard, so the RevOps lead can prioritize enrichment pulls on the accounts that most affect the model. Garbage in, garbage out applies here more than almost anywhere.
Lagging for high-growth accounts
Last year's revenue misses this year's rocket.
A startup that was 2M a year ago is 25M now. Last year's revenue field has the account in a low band, and the scoring model buries a legitimately qualified lead. Teams selling to high-growth segments should layer in a growth-rate signal (year-over-year revenue delta, headcount growth, funding events) that captures trajectory, not just current state.
Private-company blindspot
Revenue data on private companies is weaker.
Public companies disclose revenue; private companies estimate it. Vendor enrichment estimates vary by plus or minus 40 percent for private firms. If the business sells primarily to private companies, pair revenue bands with employee count bands and funding-stage bands as parallel sizing proxies. The composite usually beats revenue alone for private-dominated TAMs.
Overfitting on close history
The model learns what you have won, not what you could win.
A model calibrated against closed-won history is biased toward the deal shapes the team has already proven. New-segment expansion needs a different signal set, because the historical data cannot calibrate on segments the team has not sold to yet. Run the new-segment motion with a separate scoring profile that leans on fit heuristics rather than close history.
Not a routing decision alone
High score does not mean "give to the top AE."
Routing decisions should combine the score with capacity (which AE has bandwidth), ownership rules (named accounts), and specialization (vertical expertise). A pure score-based routing rule sends the top leads to overloaded reps while under-utilized reps get the dregs. Strkr's round-robin and weighted-random routing primitives let the admin build routing logic that respects score, capacity, and ownership together.
Head-to-head
Strkr vs Pipedrive on revenue-range scoring methodology.
Pipedrive's native scoring treats revenue as a monotonic input with simple weight assignment. Buyers migrating from Pipedrive who want revenue-band scoring usually hit the limits of the native model within six months. Here is the direct comparison.
Feature
Strkr
Pipedrive (Professional with Lead Booster)
Revenue bands
Up to 12 bands, inclusive-exclusive boundaries
Linear weight, no band concept
Per-band point weight
Admin assigns points per band
Single multiplier on revenue field
Segment overrides
Per-vertical or per-product band curves
Not supported in native scoring
Unknown-revenue handling
Null-safe band with configurable weight
Treats null as zero
Weekly calibration overlay
Automated job vs close-rate reality
Manual report, not scheduled
Audit log on weight changes
Full change history with reviewer and rationale
Changes not logged
Score composition on lead card
Per-signal breakdown rendered inline
Score only, no composition
Behavior decay
Admin-tunable half-life on behavior signals
Static behavior scoring
Export to warehouse
Hourly CDC export with band assignments
Report export, no streaming
Model dual-run
Parallel legacy and new model for 30-60 days
Not supported
How teams use Strkr
How scoring-literate teams run the methodology today.
The teams that get the most out of revenue-range scoring treat the model as a living artifact: calibrated weekly, audited quarterly, retuned when drift exceeds a threshold. The playbooks below show how customers run this motion.
Mid-market SaaS, 38 reps
Sweet-spot band moved from 10 to 50M to 25 to 75M in month six.
A 38-rep mid-market SaaS team launched Strkr scoring with a sweet-spot band at 10 to 50M, matching historical close data. Six months in, the calibration overlay flagged drift: close rate in the 25 to 75M band had risen past the 10 to 50M band. The RevOps lead investigated, found the product had added enterprise features that moved the ICP up, and rebalanced bands over a 30-day transition. The audit log captured the rationale. Pipeline velocity rose 11 percent over the following quarter.
FinServ B2B, 22 AEs
Segment override for compliance-driven deals in the 100 to 500M band.
A 22-AE finserv team found that compliance-regulated accounts closed in a different band than non-regulated ones. The RevOps lead added a segment override on accounts tagged "compliance-regulated" that elevated the 100 to 500M band from 20 points to 35. Non-regulated accounts kept the base curve. The segmented approach lifted score-to-close conversion on the compliance segment by 24 percent without diluting signal on the base segment.
Growth-stage startup, 14 AEs
Growth-rate overlay caught 40 breakout accounts the revenue band had buried.
A 14-AE startup selling to other startups was burying high-growth accounts in low revenue bands because the enrichment vendor was 12 months stale. The RevOps lead added a growth-rate overlay (year-over-year headcount delta from LinkedIn data) that elevated accounts growing more than 50 percent year-over-year by 15 points regardless of band. The overlay surfaced 40 accounts in month one that reps had deprioritized. Six closed inside two quarters.
Private-dominated TAM, 19 AEs
Employee-count band paired with revenue band to cover private-firm blindspot.
A 19-AE team selling to a mostly-private TAM found that revenue data was missing or stale on 46 percent of accounts. They layered an employee-count band (under 50, 50 to 250, 250 to 1000, over 1000) as a parallel fit signal with its own weight curve. Composite score dropped the unknown-revenue bucket from 46 percent to 12 percent of the pipeline, and the sales team reported the composite "felt right" within three weeks.
Multi-product horizontal CRM, 60 reps
Product-line-specific curves let one SKU elevate while another stayed baseline.
A 60-rep horizontal CRM team sells four product lines. One line (compliance tooling) performs best in the 50 to 250M band; another line (sales enablement) performs best in the 10 to 50M band. The RevOps lead configured per-product-line segment overrides so leads scored differently depending on which product line was the primary interest. The product-aware scoring eliminated 70 percent of the misrouted leads reps had been quietly complaining about for six months.
Build a lead-scoring model that matches your sweet spot, not the vendor default.
Start a 14-day trial. Configure revenue bands, assign points per band, calibrate weekly against close-rate reality, and audit every weight change. The scoring model that stops drifting after month three.
What is revenue-range lead scoring methodology and why does it beat linear weighting?
Revenue-range scoring assigns point weights to revenue buckets rather than treating revenue as a continuous input. Linear weighting over-rewards the biggest numbers and under-rewards the sweet-spot band where most B2B teams actually close. Band-based scoring lets a RevOps lead define bucket boundaries (under 1M, 1 to 10M, 10 to 50M, and so on), assign points per band, and shape the curve to match historical close data. Teams that migrate from linear weighting typically see 10 to 25 percent lift in score-to-close conversion within the first calibration cycle, because the model stops routing attention at the F100 names that look impressive and rarely close.
How does Strkr calibrate the band weights against actual outcomes?
A weekly calibration job joins each scoring band against actual close rate over the trailing 90 days, renders an overlay chart that compares model prediction to reality, and writes a calibration score (0 to 1) that leadership tracks over time. When drift exceeds a configurable threshold the RevOps lead gets a prompt to retune. The job also surfaces band-level dispersion, so a RevOps lead can tell whether the drift is in one specific band or across the curve. The calibration is the piece that keeps a scoring model honest; without it, the model drifts and nobody notices for a quarter.
What happens when the revenue field on an account is missing or unknown?
Strkr supports a null-safe "revenue unknown" band with its own point weight. The admin assigns the weight based on how unknowns typically perform in the business. Teams selling to a mostly-private TAM often set the unknown-band weight at the median value so unknowns do not get buried. Teams selling to a mostly-public TAM might set it low, because missing revenue in a public-dominated segment is usually a signal of a sub-scale target. The admin can also configure a workflow that routes unknowns to an enrichment job before scoring finalizes, which resolves most of the ambiguity automatically.
Can we run segment-specific scoring curves without maintaining 10 separate models?
Yes, via segment overrides. The RevOps lead defines a base model with default band weights, then adds overrides per segment (vertical, product line, geography) that adjust specific band weights without duplicating the whole model. The override is a diff, not a copy, which keeps maintenance tractable even when the business has 8 to 12 segments. Changes to the base model propagate to segment-override models unless the override explicitly pinned the band. The admin surface shows the effective curve per segment side by side so the RevOps lead can audit the composite before activation.
How does this compare to HubSpot predictive scoring or Salesforce Einstein Lead Scoring?
HubSpot predictive scoring and Salesforce Einstein both ship opinionated ML models trained on behavior and firmographic inputs. They are strong for teams that want a managed black box and weak for teams that want to inspect, tune, or override the model. Strkr's revenue-range methodology is a white-box model where the RevOps lead sees every band, every weight, and every segment override. For teams that have run a managed ML model for a year and want direct control over the curve, the switch to Strkr's explicit methodology typically happens because the opaque models stopped matching the operating reality and the team had no lever to fix it. For teams that want managed ML, HubSpot and Salesforce stay strong.
What is the audit story for scoring model changes in regulated industries?
Strkr logs every band weight change with the user who made it, the before-and-after values, the free-text rationale, and the timestamp. The audit log is immutable, exportable, and reviewable by compliance. For regulated industries where lead scoring affects routing decisions that have fairness implications (finserv, healthcare, education), the audit log is also useful for demonstrating that scoring changes were deliberate and reviewed rather than ad-hoc. Compliance officers who have been through a scoring-model review at a legacy CRM tend to appreciate that the audit log actually captures what they need to see rather than making them reconstruct history from support tickets.
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.
We use cookies.
Essential cookies keep the site working. If you accept, we also enable Google Analytics
so we can see which pages help and which don't. Reject to opt out entirely. Details in our
Privacy Policy.
