Pardot Einstein, the set of Einstein features in Salesforce Account Engagement (formerly Pardot), includes several capabilities designed to help marketers make better decisions about their prospects and campaigns. These can make parts of marketing automation more data-driven, but getting value from them depends on more than activating them. Each feature works with different Salesforce records, draws on different types of data, and has specific requirements that determine how useful its predictions and insights will be.

That makes it worth understanding what each Einstein feature does before building it into your marketing operations or paying for an edition that includes it.

We take a look at the practical applications, setup requirements, limitations, and costs of these capabilities. We’ll also assess where you can start using them, where a pilot makes more sense, and where your data needs attention first.

How does Salesforce Einstein enhance marketing automation?

Einstein replaces some of the manual guesswork in marketing automation with predictions learned from your own CRM and engagement data.

For instance, instead of a marketer deciding what a form fill is worth, a model studies which past prospects became opportunities and weights activity to match.

From fixed rules to prediction

The standard Prospect Score in Account Engagement runs on static factors and weights that are set by default or edited by an admin. A rule might add points for every pricing-page visit, whether or not those visits ever led to deals.

Einstein Behavior Scoring instead uses a model that changes as it learns, and it weighs frequency, recency and the lag between activities. For example, the model could rate an email open from last week above a form submission from 90 days ago.

Where Einstein sits in Account Engagement

Einstein in Account Engagement covers three areas:

• Lead scoring: Einstein Lead Scoring, which scores Salesforce lead records on how well they fit past conversions.
• Behavior scoring: Einstein Behavior Scoring, which scores prospects from 0 to 100 on engagement.
• Campaign analytics: Einstein Campaign Insights and Einstein Attribution.

Other Einstein features, such as Send Time Optimization and Key Accounts Identification, are outside this guide. For a wider view, see our overview of Account Engagement AI features. All of these are separate from the Einstein tools in Marketing Cloud Engagement, which is a different product, and from the credit-based Agentforce features that Account Engagement customers can reach through Marketing Cloud Next. Our guide to Account Engagement and Agentforce explains how the two fit together.

What you need in place first

Before you activate Einstein Behavior Scoring, Campaign Insights or Attribution in Marketing Setup, confirm these five prerequisites:

• Account Engagement Advanced Edition, with Salesforce Enterprise, Performance or Unlimited Edition; Premium also qualifies
• A verified Salesforce connector in Account Engagement
• Prospects syncing to leads, contacts or both, for Behavior Scoring
• Connected Campaigns, for Campaign Insights and Attribution
• Customizable Campaign Influence, for Attribution

Pardot Einstein lead scoring: When it beats manual scoring

Einstein Lead Scoring scores Salesforce lead records, not Account Engagement prospects, by learning which past leads converted. A custom model needs at least 1,000 leads created in the last 200 days, with 120 or more converted (Salesforce Help, 2026). Below that, if you score all leads together, Einstein uses a global model built from anonymized data across many Salesforce customers.

It is worth testing against a manual model that has not changed in a long time, as long as you have the lead volume it needs. The prospect-side predictive score is Behavior Scoring, covered in the next section.

How it learns from CRM data

Einstein Lead Scoring studies historical field data on the Lead object, including custom fields, to find patterns in leads that converted. It then scores each open lead on how closely it matches those patterns and shows which fields moved its score most. For example, a manual grade says a VP in fintech deserves an A because someone decided so; Einstein says it only if your converted leads show it. Our guide to Pardot scoring and grading explains how manual grades work.

In other words, behavior scoring shows how interested a prospect is in you, and lead scoring shows how interested you should be in the lead.

Einstein Lead Scoring is available through Account Engagement, which lists it among its AI features, and through Sales Cloud Einstein, which is included in Performance and Unlimited Editions and an extra cost in Enterprise. Confirm which contract covers you.

Where it wins

It suits teams with a steady inflow of leads and a wish to spend less time maintaining grading rules. Each lead gets a score and the fields that affect it most, so reps know where to focus first.

Suppose a team’s grading profile gives an A to any director-level title. After a quarter, Einstein’s top-scored leads cluster around company size and lead source instead, and the director-title A-grades convert no better than average. The team rewrites its grading criteria around what Einstein surfaced.

Where it falls short

Low lead volume and long sales cycles make the custom-model threshold harder to reach, which leaves you on the global model. It scores only leads, so contacts and prospects that have not synced as leads get no Einstein Lead Score.

Verdict: pilot first, beside your manual grade for one quarter, before you change any routing.

Einstein behavior scoring in Account Engagement: what prospects do, not just who they are

Einstein Behavior Scoring rates eligible Account Engagement prospects from 0 to 100 based on their engagement history. It weighs each activity by frequency and recency and learns from your data. A tailored model needs six months of engagement data and at least 20 prospects linked to opportunities.

Lead scoring says who fits. Einstein Behavior Scoring says which actions mattered in your own data.

What Einstein behavior scoring surfaces

The model reviews all prospect engagement activity, marks each interaction as positive or negative, and weights it by frequency and recency for each activity type and asset. Each score comes with rationales, such as email opens, file, form and video views, event registrations and check-ins, unsubscribes and spam complaints.

The optional Einstein Behavior Scoring dashboard in B2B Marketing Analytics shows which engagement activities affect opportunities most and which assets perform best. It displays each prospect’s top four factors, and the model is retrained every 10 days.

The results can contradict what your team assumes. For example, a webinar your team views as high-intent may carry less weight than a click on a pricing email.

Before you act on those insights, plan around these limits:

• Until you have six months of engagement data and 20 linked prospects, Einstein uses a baseline model.
• A prospect needs engagement activity from the past year to be scored.
• Scores usually refresh every four hours.
• Prospects linked to an opportunity, or moved to the Sales Qualified Lead lifecycle stage, count as converted and stop being scored.
• Duplicate prospects’ scores are not combined.

The score reports patterns in historical data. It shows correlation with past conversions, not a reading of what a buyer intends.

From insight to rules and nurture triggers

Behavior Score can be used as criteria in Account Engagement automation rules and Engagement Studio, and it appears on lead and contact list views and record pages in Salesforce. The score card sits on Lightning record pages, so if your team still works mainly in Account Engagement Classic, our Classic vs. Lightning comparison is worth a read first.

Use what it reveals in three places:

• Re-weight your Prospect Score points toward the high-signal actions.
• Rebuild Engagement Studio branches around those actions.
• Tune the threshold at which sales gets an alert.

Behavior Score does not replace the Prospect Score, so keep both.

Review the dashboard each quarter, and log every scoring or nurture change with its date and the metric you expect it to move.

After two quarters, the log shows which changes helped.

Einstein campaign insights and attribution in Account Engagement

Einstein Campaign Insights flags audiences and assets that over- or under-perform inside your campaigns, based on engagement and audience data. It needs Connected Campaigns, and Salesforce recommends at least 50 connected campaigns. Einstein Attribution is separate and shows which campaigns influenced opportunity value.

Attribution only works when Salesforce campaign and opportunity data are tied together properly.

What it shows

Campaign Insights reads engagement with list emails, forms and landing pages, plus job title, industry, country and state. It flags activity that is unusually high or low against a baseline, as global insights (one form outperformed your other forms) or segment insights (one audience in one region submitted a form less than others).

Insights appear on Lightning home and campaign record pages, refresh up to every four hours, and can feed a custom report. They answer questions like which asset is underperforming, and which persona or region should be the next campaign target.

Dashboards and attribution

Campaign Insights shows which assets and audiences stand out, on Lightning home and campaign pages. Einstein Attribution shows how much opportunity value each campaign influenced, in Campaign Influence related lists and reports. B2B Marketing Analytics adds dashboards on top: the Multi-Touch Attribution dashboard compares attribution models, including Einstein’s, and the Behavior Scoring dashboard shows how the scoring model works.

Einstein Attribution’s Data-Driven Model borrows the Shapley value from cooperative game theory to split an opportunity’s amount across the campaigns that touched it. By default it looks at campaign activity up to six months before opportunity creation, and you set the opportunity stage that counts as success. B2B Marketing Analytics Plus and the standard B2B Marketing Analytics app are bundled differently by edition, so check your contract before you plan dashboards around them.

Limits to plan around

Attribution also needs the standard Opportunity object with Amount and Close Date filled in, and at least 100 opportunity contact roles. Einstein fills some gaps by creating virtual contact roles from tasks, emails, events and email domains, but that does not replace campaign membership discipline. Deactivating the stage you chose as the success milestone stops the model.

Insight quality equals CRM data quality. Before paying for anything extra, pull the last two quarters’ closed opportunities and check how many have a contact role and a campaign influence record. If most do not, fix that first.

Pardot Einstein cost: which features are worth paying for?

Pay only where you already have the data the model needs. Behavior Scoring, Campaign Insights and Attribution come with Advanced and Premium Editions. At US list prices billed annually as of November 2026, Plus is $2,750, Advanced $4,400 and Premium $15,000 per org per month. Plus is also marketed with AI-powered scoring, so ask your account executive exactly which Einstein features your contract includes.

Feature	Best for	Needs	Verdict
Einstein Behavior Scoring	Teams tuning scoring rules and nurture	6 months of engagement data, 20+ prospects linked to opportunities	Activate now
Einstein Campaign Insights	Teams improving assets and targeting	Connected Campaigns, about 50 connected campaigns	Activate now, if those are in place
Einstein Lead Scoring	Steady lead flow, stale manual grades	1,000 leads in 200 days, 120 converted; Sales Cloud Einstein or an Account Engagement edition that includes it (check your contract)	Pilot first
Einstein Attribution	Leaders who must show pipeline influence	100+ contact roles, Customizable Campaign Influence, Amount and Close Date filled	Skip for now, until the data audit passes

Conclusion

Once your edition is covered, most of these verdicts turn on one question: is your data ready for the model?

To find out, start with three steps:

1. Audit CRM and campaign data: contact roles, campaign membership and Connected Campaigns.
2. Activate Behavior Scoring and start a quarterly review log.
3. Run Einstein Lead Scoring beside your manual grade for one quarter, then decide.

If you want a second pair of eyes, Mavlers can review your Account Engagement setup and tell you which of these steps applies to your org first.