Einstein Lead Scoring
Einstein Lead Scoring is the Salesforce machine learning feature that ranks Leads on a 1-99 score predicting which ones are most likely to convert into Opportunities.
Definition
Einstein Lead Scoring is the Salesforce machine learning feature that ranks Leads on a 1-99 score predicting which ones are most likely to convert into Opportunities. The model trains on your org's own historical Lead data: which Leads converted, which fields they had populated, how long they took, what Lead Source brought them in. The output is a score field plus an explanation panel showing the top factors driving each Lead's score, so reps can see why one Lead is hotter than another.
The feature is part of Sales Cloud Einstein, an add-on license tier above standard Sales Cloud. Einstein Lead Scoring is one of several scoring products (Opportunity Scoring, Account Recommendations, Activity Capture). It is the most mature and the easiest to deploy, because it requires no configuration beyond enabling it; the model auto-trains on the org's existing data and starts producing scores within 24-48 hours. Orgs without enough historical conversion data (under a few hundred converted Leads in the last six months) cannot use it; the model needs training examples.
How Einstein Lead Scoring builds a per-org conversion model from historical Lead data
The score, the explanation, and what reps see
Each Lead carries an Einstein Score field (0-99) and an Einstein Score Tier (Hot, Warm, Cold). The Lead detail page shows the score, the tier, and a Top Positive Factors and Top Negative Factors panel that lists the Lead fields driving the score. Reps see "This Lead scored 87 because Industry is Software, Number of Employees is 1000+, and Lead Source is Inbound." The explanation is what turns a numeric score into actionable triage; without it, scoring is a black box.
The training data requirement
Einstein trains on the org's last 12 months of Lead conversion data. The minimum requirement is 1000 Leads created in the past 6 months with at least 120 converted to Opportunity. Orgs below this threshold cannot enable the feature; the model has nothing to learn from. Most enterprise orgs have plenty of data; small-business orgs and new Sales Cloud deployments often have to wait six months to a year before Einstein Lead Scoring becomes available.
Auto-retraining and model drift
The model retrains automatically every 10 days using the most recent conversion data. This catches market shifts (new product launches, new buyer segments, new high-converting industries) without admin intervention. The retrain is invisible to users; the score field updates silently. For audit purposes, the explanation panel always reflects the current model; you cannot see "this Lead scored 87 last month based on the old model" historically.
Predictor exclusions and the field selector
Admins can exclude specific Lead fields from the model under Setup, Einstein Lead Scoring, Excluded Fields. Common exclusions: PII fields (email, phone) where presence is universal and contributes nothing to discrimination; legacy fields that no longer reflect business reality; fields with extreme cardinality (free-text comment fields) that confuse the model. Inclusion is implicit; everything not excluded is candidate predictor. Audit the field list before enabling; one bad field can skew the model.
Score tiers, customizable thresholds, and reporting
Hot, Warm, and Cold tiers use default thresholds (Hot 80+, Warm 30-79, Cold under 30) that admins can customize per org. The tier field drives list view filtering, dashboard segmentation, and assignment rule routing. Reports group Leads by tier to show "where are our hot leads coming from" or "Hot Leads not contacted in 7 days." The tier is what makes the score operational; raw numeric scores need bucketing to drive decisions.
Einstein Lead Scoring vs Pardot Lead Scoring
Two scoring systems coexist. Einstein Lead Scoring (Sales Cloud Einstein) uses ML on historical conversions. Pardot Lead Scoring (Marketing Cloud Account Engagement) uses rule-based points: open an email +5, visit pricing page +10, click an ad -2. Both produce a number on the Lead. Many orgs use both, with Pardot scoring activity engagement and Einstein scoring fit/conversion likelihood. Combining them via a custom formula field is the standard pattern for blended scoring.
Limits, caveats, and known weaknesses
Einstein Lead Scoring works well when the conversion criteria are field-based and the past predicts the future. It struggles with brand-new market segments (no training data), with rapidly shifting buyer behaviour (the 10-day retrain lags), and with orgs where most conversion happens via channels Einstein cannot see (offline meetings, partner referrals). Pair the score with rep judgment; do not auto-disposition Cold Leads without human review. Salesforce documents the feature as a recommendation engine, not an oracle.
Enabling and tuning Einstein Lead Scoring
Einstein Lead Scoring is one of the easiest Einstein features to enable. The setup is mostly verifying the org has enough training data and picking which fields to exclude. Initial scores appear within 24-48 hours of enable.
- Verify the data threshold
Setup, Einstein Lead Scoring shows a Data Readiness check. Confirm at least 1000 Leads in the past 6 months and 120 conversions. Below threshold, the feature stays grayed out; collect more data or check with Salesforce Support about edge cases.
- Enable the feature
Setup, Einstein Lead Scoring, Toggle Enable Einstein Lead Scoring on. The platform begins training on existing data. First-time training takes 24-48 hours; scores appear on Lead records after.
- Add the Score and Score Tier fields to layouts
Object Manager, Lead, Page Layouts. Add Einstein Score and Einstein Score Tier to the Lead Information section. Add the Einstein Score Insights component to the Lightning record page so reps see the Top Factors panel inline.
- Configure excluded fields
Setup, Einstein Lead Scoring, Excluded Fields. Exclude PII, legacy fields, and high-cardinality free-text. Review the auto-included predictor list and remove anything irrelevant.
- Update assignment rules and reports
Add Einstein Score Tier to the Lead Assignment Rule entries (route Hot Leads to a senior rep, Cold to a nurture queue). Build a Hot Leads dashboard component. Set up alerts on aging Hot Leads that have not been contacted.
Lead fields to leave out of the model. Always exclude PII (email, phone), legacy fields, and high-cardinality text. Inclusion is implicit.
Customize the Hot, Warm, Cold cutoffs from defaults (80/30). Match to your org''s rep capacity and definition of high-priority.
Automatic every 10 days. Not user-configurable. Re-enabling the feature triggers an immediate retrain.
Field-level security on Einstein Score controls which users see the score. Hide from junior reps if the org wants to filter by tier centrally.
- Minimum data threshold: 1000 Leads in 6 months, 120 conversions. Below this, the feature stays unavailable.
- The model retrains every 10 days. Rapid market shifts (new product launch, new buyer segment) lag the model by up to two weeks.
- Einstein cannot see conversion happening outside Salesforce. If most deals close offline or via partners without proper Salesforce tracking, scoring will misrank Leads.
- Free-text custom fields with high cardinality (long open-text comments) confuse the model. Exclude them.
- Pardot Lead Scoring and Einstein Lead Scoring are separate. Without a custom formula, reps see two different scores on the same Lead and have to reconcile manually.
Trust & references
Cross-checked against the following references.
- Einstein Lead Scoring OverviewSalesforce Help
- Set Up Einstein Lead ScoringSalesforce Help
Straight from the source - Salesforce's reference material on Einstein Lead Scoring.
- Data Requirements for Einstein Lead ScoringSalesforce Help
- View Einstein Lead Scoring InsightsSalesforce Help
About the Author
Dipojjal Chakrabarti is a B2C Solution Architect with 29 Salesforce certifications and over 13 years in the Salesforce ecosystem. He runs salesforcedictionary.com to help admins, developers, architects, and cert/interview candidates sharpen their fundamentals. More about Dipojjal.
Test your knowledge
Q1. What does Einstein Lead Scoring do?
Q2. Why is Einstein scoring better than manual A/B/C/D ratings?
Q3. What's a good way to validate the model?
Discussion
Loading discussion…