Key Takeaways
- Einstein Lead Scoring is a machine learning model trained on your historical Salesforce data — not a generic industry model; it needs 1,000+ examples in each category to produce accurate predictions.
- Einstein Opportunity Insights surfaces at-risk deals based on engagement patterns — most useful for managers reviewing pipeline, not reps managing daily activity.
- Einstein Activity Capture reduces manual logging burden but stores data separately from standard Salesforce activities — understand the reporting limitations before enabling at scale.
- Most Einstein sales features are included in Enterprise and Unlimited editions — check existing licenses before assuming additional cost.
- Einstein features require good underlying data to produce good predictions — an org with poor adoption and incomplete data will not benefit significantly from Einstein features.
Salesforce Einstein AI was introduced with considerable fanfare in 2016 and has expanded significantly since. By 2020, it's a set of specific machine learning features embedded in various Salesforce products — not a general artificial intelligence, but purpose-built models for specific sales process tasks. The marketing framing — 'AI for your CRM' — is both accurate and somewhat overstated. The features are real, the predictions are useful (when the data is there), and the automation genuinely reduces manual work. But Einstein's value is conditional on the same requirement that all CRM features share: high-quality data and consistent adoption. This guide gives an honest account of what Einstein's core sales features do, what they require to work well, and which are worth enabling for different types of sales organisations.
Definition: Machine Learning Model Training in Salesforce Einstein
Machine learning models in Salesforce Einstein are trained on your organisation's historical Salesforce data — not on generic datasets or industry benchmarks. Einstein Lead Scoring trains by analysing the features (field values) of leads that converted to opportunities and leads that were discarded or decayed, identifying patterns that predict conversion in your specific business context. The model updates automatically as new conversion data accumulates in your org. This org-specific training means Einstein predictions are more relevant than generic industry predictions — a technology company's converted lead profile differs significantly from a financial services company's — but it also means the model requires sufficient historical data to train on. Orgs with insufficient historical data (fewer than 1,000 examples in each category) should not expect useful predictions from Einstein features that require training.
Einstein Lead Scoring and Opportunity Insights
Lead Scoring prioritises which leads to work first; Opportunity Insights flags which deals are at risk — both add value when lead and deal volumes exceed what managers can monitor manually.
Einstein Lead Scoring displays a score (1–99) on each lead record, visible in lead list views and on the lead detail page. Reps see which leads have high predicted conversion probability and can prioritise accordingly. The practical value: in a queue of 50 leads, the score tells a rep which 10 to call first — not because the rep doesn't know their leads, but because a machine learning model considering 20+ factors simultaneously identifies patterns human review of individual records often misses. The ROI is clearest when lead volume is high enough that not all leads can be worked equally — Lead Scoring enables triage.
Einstein Opportunity Insights appears on opportunity records in Lightning Experience — displaying three signal categories: Deal Risk (signals that the deal may be stalling — no response to last email, meetings declining, conversation sentiment declining), Key Moments (the opportunity passed a significant milestone — proposal sent, executive engaged, competitor mentioned), and Follow-up Reminders (based on last activity and expected deal timeline). For sales managers reviewing pipeline in weekly one-on-ones, Opportunity Insights provides a quick visual indicator of which deals to discuss most urgently, without requiring managers to read through every deal's activity timeline.
Both features require Lightning Experience — they don't appear in Salesforce Classic. If your organisation hasn't migrated to Lightning, enabling Einstein Lead Scoring and Opportunity Insights is a motivation to complete that migration. Both features also require Salesforce data quality: Lead Scoring needs complete lead records with consistent field values; Opportunity Insights needs activity data logged consistently. Orgs with poor data completeness should improve data quality before expecting Einstein to produce reliable predictions.
- Einstein Lead Scoring: requires 1,000+ converted and 1,000+ non-converted leads in your org for training
- Score displayed on Lead record and in Lead list views — reps prioritise by score
- Opportunity Insights: visible on Opportunity records in Lightning — Deal Risk, Key Moments, Follow-up Reminders
- Both require Lightning Experience — not available in Salesforce Classic
- Best for: high lead volume orgs (100+/month) and managers reviewing large pipeline ($2M+)
Einstein Activity Capture and Einstein Forecasting
Activity Capture reduces manual logging burden; Forecasting uses AI to improve sales forecast accuracy — both valuable in specific conditions.
Einstein Activity Capture connects Salesforce to Microsoft Exchange/Office 365 or Google Gmail/Calendar and automatically logs emails and calendar events to the related Salesforce records based on the email addresses of Contacts and Leads. When a rep emails a contact, the email is automatically logged to the Contact record and associated Opportunity without any manual action. The benefit: activity data completeness improves without requiring rep behaviour change. The limitation: EAC stores data in a separate data store ('Einstein Data Storage') rather than standard Salesforce Activities — this data doesn't appear in standard Salesforce reports, doesn't count toward validation rules based on activity completion, and isn't queryable with standard SOQL. Understand these limitations before making adoption decisions based on EAC data.
Einstein Forecasting provides an AI-generated sales forecast alongside the traditional rep-submitted and manager-adjusted forecasts. It analyses historical forecasting accuracy (how often rep and manager forecasts were correct at various points in the quarter), opportunity pipeline trends, and win rate patterns to generate a predicted close amount for the quarter. The AI forecast is displayed as a range with confidence intervals, complementing rather than replacing the human judgment forecast. It's most valuable in organisations where forecasting accuracy is strategically important (public company guidance, investor reporting, significant resource allocation decisions) and where historical Salesforce data quality is high.
Einstein Conversation Insights (a premium Einstein add-on) transcribes and analyses sales call recordings — identifying competitor mentions, objections raised, and coaching opportunities across all calls without requiring managers to listen to recordings manually. It surfaces patterns like 'calls where competitor X was mentioned had a 40% lower close rate' or 'calls where rep followed the discovery script had a 25% higher close rate'. This feature requires Sales Cloud Einstein license (additional cost beyond standard Enterprise/Unlimited plans) and integration with your call recording tool.
Experience Signal
The Einstein features that deliver clear value in most organisations we've worked with are Lead Scoring (when lead volume is high) and Activity Capture (when manual logging compliance is consistently below 70%). The features that sound compelling in demos but deliver less in practice for most mid-market orgs are Opportunity Insights (reps often ignore the signals because the deal context isn't in the CRM) and Einstein Forecasting (requires high historical data quality that many orgs don't have). The universal requirement remains: Einstein makes good data better; it doesn't fix bad data.
Frequently Asked Questions
Einstein Lead Scoring is a machine learning model trained on your organisation's historical Salesforce data — specifically, the characteristics of leads that converted to opportunities versus those that didn't. It analyses factors like company size, industry, job title, lead source, and engagement behaviour, and produces a score (1–99) for each new lead predicting conversion likelihood. The model is built from your data, not generic industry data — it reflects your actual conversion patterns. Minimum requirement: Salesforce needs at least 1,000 converted leads and 1,000 non-converted leads in your org to build an effective model. Orgs with less historical data receive less accurate predictions.
Einstein Activity Capture (EAC) automatically syncs emails and calendar events between Outlook/Google and Salesforce — logging activities to the related Salesforce records without manual input. The benefit: reps don't need to manually log emails and meetings, improving activity data completeness. The limitations: EAC stores synced data separately from standard Salesforce activity data — it's not queryable through standard reports, it doesn't appear in the standard Activity Timeline in the same way as manually logged activities, and the sync is one-directional (email to Salesforce, not Salesforce to email). For organisations with low manual activity logging compliance, EAC is a practical fix. For orgs where activity data quality for reporting matters, the data storage limitations are important to understand before enabling.
Einstein Lead Scoring and Einstein Opportunity Insights are included in Salesforce Sales Cloud Enterprise and Unlimited editions — no additional license required. Einstein Activity Capture is also included in Enterprise and Unlimited. Einstein Forecasting (AI-augmented sales forecasting) is available in Enterprise and above. The premium Einstein add-ons — Einstein Conversation Insights (analyses sales call recordings), Einstein Relationship Insights (surfaces relationship context from external data), and Einstein Prediction Builder (custom AI predictions) — require additional licensing beyond standard Sales Cloud plans. Check your current Salesforce contract for which Einstein features are included before purchasing add-ons.
Sources
Want to Get More from Your Salesforce Investment?
Webnixon configures Salesforce Einstein features, improves data quality, and builds the adoption processes that make AI predictions useful — not theoretical.
Book a Salesforce ConsultationAbout the author
Jai Paek
Creative Director
Jai leads brand identity and UX design at Webnixon, bringing 20+ years of experience building digital design systems for agencies and enterprise teams. He has shipped design systems and visual identities for over 200 brands across Canada and the US, with deep expertise in conversion-focused UI, WCAG 2.1 accessibility compliance, and responsive web design for service businesses and ecommerce brands.
Related Articles
Business Technology
Salesforce Lightning vs Classic: Why You Need to Make the Switch
Salesforce has made its direction clear: Salesforce Classic is in maintenance mode and all new features, AI capabilities, and innovations are being built exclusively for Lightning Experience. If your organisation is still on Classic, you're running on a platform that will only fall further behind. Here's what Lightning offers and how to switch without losing user adoption.
Business Technology
Measuring Salesforce ROI: How to Prove the Value of Your CRM Investment
Salesforce is a significant investment. Leadership wants to know it's paying off. The challenge is that CRM value is diffuse — it improves processes, visibility, and consistency across the sales organisation, but connecting that to a revenue number requires specific measurement. Here's how to do it.
Business Technology
Salesforce CRM: Is It Right for Your Business?
Salesforce is the most capable and most expensive CRM platform on the market. For businesses with complex sales processes, large sales teams, and integration requirements, it's frequently the best tool available. For businesses without those characteristics, it's frequently over-engineered and over-budget. Here's how to make the call honestly.