Lead scoring criteria examples that survive contact with real pipeline.
The hard part of lead scoring is not writing the model. It is picking the criteria that correlate with closed-won revenue in your book and ignoring the ones that only correlate with form fills. This page walks through fifty-plus concrete criteria across firmographic fit, behavior, intent, engagement, and negative signals, each with a suggested weight, a decay rule, and the exact field in Strkr that captures it. Build the model on the first scroll, tune it on real conversions after thirty days.
The ICP shape, scored on the fields that actually mean something.
Firmographic fit is the first filter a lead has to pass. The criteria are not "big companies are good." The criteria are "accounts that match the shape of our best closed-won customers over the trailing twelve months are good." That shape has four or five anchor fields and a dozen secondary ones. Strkr enriches the fields on inbound capture and exposes each one as a scored criterion in the model builder. Suggested weights below are starting points. After thirty days of real conversions, the auto-tune suggestion panel re-weights each criterion against actual closed-won conversion rate.
Industry match
Target vertical: +25 points.
Companies in your top three verticals carry the strongest firmographic signal. SaaS, financial services, and healthcare are typical anchors for a horizontal product. A clean industry mapping (NAICS or your taxonomy) matters more than the number of verticals you cover. One tight industry list beats three fuzzy ones.
Employee count band
Sweet-spot headcount: +20 points.
For most B2B products, the sweet spot is a specific headcount range. 50 to 500 employees, or 500 to 5000, or 5000 and above. Score the band that matches your closed-won shape. Companies at twice the sweet spot often score lower, not higher, because the deal cycle becomes a different animal than the one your motion is built for.
Revenue band
Annual revenue in range: +15 points.
Revenue is a stronger buy signal than headcount in some verticals. Enrichment data from Clearbit, ZoomInfo, or Apollo lands in the revenue field. Score the band that matches the price tier the account is likely to buy, not an aspirational band three tiers above the actual conversion rate.
Geography
Country or region in GTM scope: +10 points.
If the GTM motion only sells in North America, EMEA, or specific countries, the geography field weighs heavily. International leads that fall outside the scope get negative scores, not neutral ones, so the sorted queue does not surface them for the SDR layer.
Technology stack
Compatible stack signal: +15 points.
For integration-heavy products, the current tech stack is a hard filter. Companies using the platforms your product plugs into score higher. Companies using a direct competitor score with a different weight, often positive, because they are in-market and already paying for the category.
Growth signal
Hiring in target roles: +10 points.
An account hiring sales ops, revenue ops, or sales enablement is likely in a buying motion for your category. Job posting feeds from Clearbit, Crunchbase, or LinkedIn hiring data land on the account record and raise the fit score when they match the roles you sell into.
Funding stage
Recently funded: +12 points.
For products priced for scaling teams, a recent Series B or Series C funding round signals budget availability. Crunchbase funding data lands on the account and surfaces as a scorable field. The signal decays after six months, which the model handles with a decay rule rather than a hard cutoff.
Public vs private
Entity type: +5 points for match.
Some products convert better on public companies, others on private. Score the entity type that matches the actual conversion data in your book. The weight is smaller than other firmographic signals but tightens the ranking when two leads otherwise score similarly.
Negative firmographic
Out-of-ICP: minus 30 points.
Leads from companies in excluded industries (if any), from banned geographies, or at an obvious off-ICP headcount band carry a strong negative score. The sorted queue surfaces them at the bottom with the rejection reason visible. Nothing goes to a rep without being seen, nothing goes to a rep as a priority when the firmographic data disqualifies it.
Behavioral criteria
The signals that actually track buying intent.
Behavioral criteria are what the lead did, not who the lead is. These signals have shorter half-lives, higher weights, and better conversion correlation than firmographic signals. A lead who visited the pricing page three times this week signals buying intent in a way that no demographic field matches. Strkr tracks the behaviors natively through the web tracking SDK, the forms layer, the messaging platform, and the email engine. Each behavior becomes a scorable criterion with a configurable weight and decay.
Pricing page visit
Hit /pricing: +15 points.
The pricing page is the highest-intent URL on most B2B sites. One visit is a signal, three visits in a week is a buying motion, three visits in a day is a shortlist evaluation. Score escalates by frequency and recency. Decays over fourteen days if no follow-on behavior fires.
Demo request
Submitted the demo form: +50 points.
A demo request is the strongest single behavioral signal in most inbound funnels. Fifty points on first fire, plus the firmographic enrichment that fires on the same submission. Combined with a mid-market firmographic match, the lead often lands in the hot-ready-for-SDR tier on the first scroll.
Content download
Downloaded a gated asset: +10 points.
Downloading a buyer's guide, a comparison report, or an ROI calculator is a stage-two signal. Not as strong as a demo request but stronger than a blog post view. Different assets carry different weights depending on where in the journey the content sits: top of funnel gets smaller scores, bottom of funnel gets larger ones.
Webinar attendance
Showed up live to a webinar: +20 points.
A live webinar attendance is a stronger intent signal than a replay download. The weight reflects the effort gap. Attended plus asked a question is a tier higher still. Attended plus asked a question plus visited the pricing page within forty-eight hours is a lead the SDR should call that day.
Multiple session visits
Three sessions in seven days: +15 points.
A lead that returns to the site three times in a week signals active evaluation. The web tracker counts unique sessions, not pageviews, so a lead who bounces three times does not accidentally rank as a buying motion. The score fires on the third session and stays live for seven days.
Deep docs visit
Read the docs for 3+ minutes: +12 points.
Technical evaluators read the documentation before scheduling a call. A session on the docs site with real time-on-page is a signal that a technical buyer is in the room. The score is scoped to engineering-adjacent docs sections, not the marketing-side help center.
Comparison page visit
Hit /vs or /compare: +18 points.
Comparison pages are late-funnel traffic. A lead visiting the "vs competitor" page is actively shortlisting. The score reflects the stage. Score escalates on repeat comparison-page visits.
Form abandonment
Started a form, did not submit: +5 points.
A partial form fill is a quieter signal than a submission but still indicates intent. If the form has field-level tracking enabled, the first-name and email fields often fire before the rest. A re-engagement email can bring the lead back to complete the submission, and the score lets the system flag the behavior.
Chat engagement
Chat conversation started: +10 points.
A chat conversation on the site, whether it reached a human or not, is a captured intent signal. The transcript lands on the lead record if the chat identifies the user. The score adds on first interaction and stays live for seven days.
Email and content engagement
What the lead did with the outbound you sent.
Email and content engagement are the second-order signals a scoring model should weigh. Opens are weak but non-zero. Clicks are stronger. Replies are strongest. The model has to distinguish between a lead who opens a newsletter in Gmail's promotions tab (background noise) and a lead who clicks a specific CTA on a sequenced email (real intent). Strkr tracks the full engagement surface on every email the platform sends, so the scorer can read the signal with enough precision to avoid false positives.
Email open
Opened an outbound: +2 points.
A single open is low signal, especially since Apple Mail Privacy Protection fires opens on preload. The score is small, intentionally, so the model does not inflate on noise. Multiple opens of the same email within a short window raise the weight because the behavior is more deliberate.
Email click
Clicked a link in an outbound: +8 points.
A click is a real engagement signal. The destination matters: a click on a demo CTA scores higher than a click on a blog link. The scoring engine respects link-level categorization set in the campaign builder.
Email reply
Replied to an outbound: +25 points.
A reply is the single strongest engagement signal on an email channel. Even a short reply ("not now, maybe Q2") converts at a higher rate than most other mid-funnel behaviors. The model weights replies accordingly, and the reply content triggers an assistant classification that can escalate the score further for positive sentiment.
Multiple opens
Opened the same email 3+ times: +10 points.
Multiple opens of a single email within a seven-day window signal deliberate re-reading. The score accounts for the behavior change from passive open to active revisit. This often fires when the lead is sharing the email internally with other stakeholders.
Sequence completion
Finished a nurture track: +15 points.
A lead who stays engaged through a six-step nurture sequence without unsubscribing has signaled sustained interest. Completion fires at the end of the sequence and carries forward as a persistent score for sixty days.
Negative email signal
Unsubscribed: minus 25 points.
An unsubscribe is a clear intent to stop the conversation. The score reflects it. The lead stays in the system for context but drops out of the hot tier. A re-engagement path only triggers on a new high-intent behavior (demo request, pricing page revisit) after the unsubscribe.
Spam report
Marked outbound as spam: minus 50 points.
A spam report is a strong negative signal both for scoring and for deliverability. The lead drops to the bottom of the queue, the sending address pauses to that recipient, and the account record flags the behavior for the account team's awareness on future outreach.
Shared the email
Forwarded to a colleague: +12 points.
If the email tracking detects a forward (via a unique link signature or a known forwarding pattern), the signal is strong. Forwarded emails typically mean an internal champion is building a case. The score captures the behavior and triggers a flag on the account for additional stakeholder mapping.
Social engagement
Engaged on LinkedIn: +8 points.
If the LinkedIn integration is enabled, the model can read engagement on company or employee posts. A like or comment on a product-related post is a lightweight intent signal. Multiple engagements over a week carry a higher composite score.
Intent and third-party signal
The outside data that confirms the lead is in-market.
Third-party intent data is noisy at the single-signal level but surprisingly accurate when it stacks with first-party behavior. The model should weight intent signals carefully: a Bombora or G2 intent spike alone is not a conversion trigger, but a G2 intent spike combined with a pricing page visit and a demo form submission is a lead the SDR should call within the hour. Strkr integrates with the common intent data providers and surfaces each signal as a scorable field in the model builder.
G2 category view
Viewed Strkr on G2: +15 points.
A G2 pageview on your product page is a shortlist signal. The G2 integration can push the signal into Strkr with the account id. The score fires on the signal arrival and decays over fourteen days.
G2 comparison view
Viewed a comparison page: +20 points.
A G2 comparison view is a late-stage signal. The lead is explicitly weighing options. Score escalates when multiple comparisons involving your product fire from the same account in a week.
Capterra or Trustpilot
Review-site activity: +10 points.
Capterra and Trustpilot pageviews mimic the G2 signal shape but are less common in B2B SaaS. The weight is slightly lower than G2 for comparable categories. The model treats them as additive, not duplicative, with other review-site signals.
Bombora intent spike
Topic spike on relevant keywords: +12 points.
A Bombora intent spike on your category keywords means the account is researching the topic across the open web. The signal is noisy at the individual-topic level but strong when stacked across multiple relevant topics in the same week.
6sense intent
In-market stage signal: +18 points.
6sense in-market tier (decision stage) is a high-weight signal because the vendor has already run multi-source intent fusion. The score reflects the composite confidence. Weight the vendor's own confidence tier directly.
TrustRadius activity
Review download or comparison: +10 points.
TrustRadius download of a review or comparison report fires through the integration. The signal is similar in shape to the G2 comparison view and weights accordingly.
Clearbit Reveal
Anonymous visit from a target account: +8 points.
Clearbit Reveal identifies anonymous web traffic by company IP. An anonymous visit from an account on your named list carries a non-trivial score even without a form fill. The score helps surface accounts that are quietly researching.
ZoomInfo WebSights
IP-matched anonymous visit: +8 points.
ZoomInfo WebSights produces a similar signal to Clearbit Reveal. The two can run in parallel with the model deduplicating against the account record so a single anonymous visit does not double-score.
Composite intent stack
Three intent signals in seven days: +25 bonus.
The strongest third-party intent signal is the stack. When three distinct intent providers fire on the same account in a seven-day window, the model adds a bonus on top of the individual signals. The composite catches accounts that are deliberately researching across channels.
Negative, decay, and tie-break criteria
The rules that keep the model honest over time.
A scoring model without negative signals and decay rules bloats into a top-heavy queue where every lead eventually scores well enough to look hot. The negatives are what distinguish a working model from a vanity dashboard. Decay rules prevent a sixty-day-old pricing page visit from dominating the queue against this week's active evaluation. Tie-break rules decide which of two 72-point leads gets the SDR's attention first. Strkr builds all three as first-class features in the model builder, not add-ons.
Competitor employee
Email domain is a competitor: minus 100 points.
A lead with a competitor email domain is a research visit, not a buyer. The negative score drops them out of the sales queue entirely. They stay in the system for competitive intelligence tracking but do not consume rep time.
Student or free email
Gmail, Yahoo, Outlook.com: minus 20 points.
A free email domain is a weaker signal than a corporate address. For most B2B motions, the negative weight is small and still allows a strong behavioral signal to lift the lead into the queue. For teams targeting SMB, the weight might be smaller or even positive depending on the ICP.
Previously disqualified
Marked DQ in last 90 days: minus 50 points.
A lead disqualified in the last ninety days should not re-surface at full score. The penalty decays over the period but keeps the lead out of the hot tier until it demonstrably re-engages with a strong new signal.
Behavior decay
A visit from thirty days ago is not a visit this week.
All behavioral signals decay. The pricing page visit score drops by 10 percent every seven days. The demo request stays at full score for thirty days then starts decaying. The decay curve is per-signal and tunable so the model reflects the real conversion half-life for each behavior type.
Account-level cap
One account cannot flood the queue.
If an account has five leads in the system, the model caps the account's contribution to the queue so one hot account does not displace five other accounts' leads. The rep gets the primary champion plus the two strongest secondary contacts, not every form fill the account produced.
Role priority
Decision maker beats individual contributor.
When two leads from the same account score similarly, the tie-break prefers the higher-seniority title. VP of Sales beats SDR when both scored 70. The model reads the enriched title field and applies the priority at the ranking stage.
Freshness tie-break
Newer behavior wins when scores match.
Two leads both at 85 points: the one with the most recent high-value action goes first. A pricing page visit three hours ago beats a demo request from two weeks ago at the same score. The queue respects recency as the secondary sort.
Explicit DNC
Do not contact flag: minus 500 points.
A hard do-not-contact flag drops the lead to the bottom of the queue permanently until a human removes the flag. The system never surfaces a DNC lead for outbound regardless of other scoring signal. Compliance and brand both demand the hard floor.
Model drift check
Monthly conversion review on each criterion.
Strkr AI runs a monthly review of each scoring criterion against the actual closed-won rate for leads that passed the threshold. Criteria whose predictive power dropped surface in the admin dashboard with a suggested re-weight. The model stays accurate over time instead of ossifying.
A working lead score, built from signals that actually convert.
Strkr ships the scoring engine, the enrichment layer, the behavior tracker, the email engagement signal, and the third-party intent integrations on every paid tier. Strkr AI reviews the model monthly against real conversion rates and suggests re-weights grounded in the book. Open the model builder and ship a working first version in an afternoon.
What is the right starting point for picking lead scoring criteria?
Start with the shape of your closed-won customers over the last twelve months. Pull the top ten firmographic attributes that correlate with won deals: industry, headcount band, revenue band, geography, technology stack, growth signal, funding stage, title of the primary buyer, product line purchased, and deal cycle length. Those become the first wave of firmographic criteria. Then add the three or four behaviors that preceded a demo request in the last ninety days of pipeline: pricing page visits, comparison page visits, demo form submissions, specific content downloads. Those become the behavioral criteria. The initial model does not need to be perfect. It needs to be in production so the thirty-day review has real data to tune against.
How many lead scoring criteria is too many?
The practical ceiling is around twenty criteria for a mid-market B2B motion. Beyond that, individual criterion weights become too small to meaningfully move a lead in the ranking, and the model becomes harder to explain to the SDR layer that has to trust the output. Twelve to fifteen criteria is a healthy range: four to six firmographic, four to six behavioral, two or three intent, and two or three negative. The constraint keeps the model readable and makes the monthly tune cycle manageable. Strkr's model builder supports more criteria than that, but the UI surfaces a warning when the model exceeds twenty active signals because the marginal signal from each additional criterion tends to shrink below the noise floor.
Should intent data be weighted higher than behavioral data?
Not usually. First-party behavioral data (what the lead did on your site, your product, your emails) typically converts at a higher rate than third-party intent data (what the lead did on the open web, aggregated through a vendor). Intent data is strongest when it stacks: three intent signals from different providers in the same week carries composite weight that outperforms any single signal. The honest configuration weights behavioral data heavier at the individual-signal level and relies on intent data for coverage of accounts that are researching quietly without a form fill. Strkr's model builder supports both and exposes the per-criterion contribution over time so admins can see which signals are earning their weight.
How do we handle lead scoring criteria that work for one segment but not another?
Build segmented models. Strkr supports multiple scoring models that run in parallel, each scoped to a segment. An enterprise model weights revenue band and executive-title engagement heavier. An SMB model weights individual-contributor behavior and credit-card-eligible firmographic signals heavier. A product-led model weights in-product usage signal heavier than inbound-marketing engagement. The lead carries both scores on the record. The router reads the model that matches the segment and surfaces the right queue. Segmented models prevent one global score from averaging away the signal that matters inside each motion.
What is a decay rule and why does every criterion need one?
A decay rule reduces the weight of a scoring signal over time. A pricing page visit from this morning is a strong buy signal. The same visit from forty-five days ago is probably not. Without decay, old high-value behaviors stay at full weight and inflate the queue with stale leads. Each criterion in Strkr's model builder carries a decay curve: linear, exponential, or step. Short-half-life signals like pricing page visits decay faster. Long-half-life signals like demo requests decay slower. The decay rules are the single most important element of keeping a scoring model fresh over quarters without a quarterly rebuild cycle.
How does Strkr AI help tune the criteria over time?
Strkr AI runs a monthly review of each scoring criterion against the actual closed-won conversion rate for leads that passed the model threshold. The review surfaces three outputs: criteria whose predictive power dropped (suggest re-weight or retire), criteria that are over-weighted relative to their contribution (suggest a lower weight), and criteria the data suggests are missing from the model (suggest add). The admin reviews the suggestions, accepts or rejects each one, and the model updates on the active production flow. The learning loop keeps the model honest as the market, the product, and the ideal customer shape change over time, without requiring a quarterly model rebuild.
Related features
Pair this with...
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.
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.
