The Intelligence Layer: Einstein Discovery, Stories & Models

We have moved data in, shaped it into datasets, and designed lenses and dashboards to display it. Now comes the part that makes CRM Analytics genuinely intelligent: the intelligence layer, powered by Einstein Discovery. This is where the platform stops merely describing the past and starts predicting the future — and, importantly, explaining itself along the way.

Stories: insights on autopilot

Point Einstein Discovery at a dataset and it produces a story — an automatically generated analysis of what drives an outcome you care about (won deals, churned customers, resolved cases). You do not write formulas; Einstein does the statistical heavy lifting and hands you plain-language findings.

A story typically surfaces:

• Top predictors — the factors most strongly associated with the outcome.
• Top influencers — how specific field values push the result up or down.
• Explanations — human-readable reasons you can trust and share.

The model behind the story

Every story is backed by a model — the machine-learning engine that learned the patterns in your data. The story is the friendly, explorable surface; the model is the predictive core underneath. When you are happy with a story's quality, you turn its model into something operational.

Deploying predictions to score records

A model is only valuable when it touches real work. Deploying a model pushes its predictions onto live records, so that when a user opens an opportunity, case, or account, they see:

• A score — the prediction (a probability, a likely value).
• An explanation — why the score is what it is.
• An action — a recommended next step to improve the outcome.

This is the whole philosophy from Lesson 2 made real: intelligence delivered to the CRM end user, understandable and actionable, not locked in a lab.

Monitoring models

A deployed model is a living thing, and data drifts over time. Customer behavior changes, markets shift, and a model that was accurate last year can quietly degrade. That is why CRM Analytics lets you monitor your models:

• Volume — how many predictions are being made.
• Alerts — flags when something looks off.
• Accuracy over time — is the model still performing, or does it need retraining?

Good practice is to keep an eye on these metrics and refresh a model when accuracy slips, so the guidance your users rely on stays trustworthy.

Einstein Discovery for Reports (EDI insights)

Not every insight needs a full CRM Analytics dashboard. Einstein Discovery for Reports — sometimes referred to as EDI insights — brings Einstein's analysis directly onto standard, operational Salesforce reports. You get predictive and descriptive insight layered right onto the reports your team already uses.

The best part: it is included free with the CRM Analytics license. That makes it a low-friction way to introduce augmented analytics to users who live in reports and are not yet ready to jump into a dedicated analytics app.

Bringing it together

The intelligence layer completes the platform picture:

1. Story — Einstein Discovery auto-generates insights (top predictors, influencers, explanations) on a dataset.
2. Model — the predictive engine behind each story.
3. Deployment — pushes scores, explanations, and actions onto records, masking the complexity.
4. Monitoring — tracks volume, alerts, and accuracy so predictions stay reliable.
5. Einstein Discovery for Reports — the same intelligence, free, on operational reports.

With data, design, and intelligence all covered conceptually, it is time to see it in action. The next lesson is a hands-on tour of Analytics Studio, the Data Manager, and a real story.