We are at Dreamforce 2026, Industry Pavilion at San Francisco
January 24, 2026
Einstein for Marketing Cloud is a set of AI features that predict who will engage, when to send, how often to send, and which content to show. The main ones are Einstein Engagement Scoring, Send Time Optimization, Einstein Engagement Frequency, Einstein Content Selection, Einstein Copy Insights, and Einstein Messaging Insights.
One thing to settle before you go looking for them in your org. Those predictive features belong to Marketing Cloud Engagement. Salesforce now also sells newer core-platform editions, and the AI you get there is not the same set. If you are on the newer editions and cannot find Engagement Scoring, you are not missing a setting.
The naming has changed more than once, and it matters here because it determines which Einstein features exist for you at all.
Here is how the names line up as of this writing. Treat it as a map rather than gospel, because Salesforce has relabeled this portfolio repeatedly.
If you are unsure which you are on, the giveaway is where journeys are built. Journey Builder with its own contact model points to Engagement. Journeys built in Flow on core platform objects, with Data Cloud underneath, point to the newer editions. Sorting this out is the first hour of any Marketing Cloud implementation conversation, and it saves weeks of chasing features that were never in your edition.
These are the predictive features that made Einstein worth talking about in email marketing. Each one answers a different question.
Scoring predicts each subscriber's likelihood of opening, clicking, and unsubscribing, then groups them into four personas: Loyalists who open and click, Window Shoppers who open but rarely click, Selective Subscribers who open rarely but engage hard when they do, and Winback or Dormant contacts who do not engage.
The personas are the useful part. Selective Subscribers get over-mailed by teams that only look at open rates, and Window Shoppers usually signal a content or offer problem rather than an audience problem.
Rather than sending everything at 9 a.m., this predicts each contact's best send window from their own history and staggers delivery accordingly.
It is the lowest-effort feature to adopt because you add it to a journey and change nothing else. It is also available on the newer editions, unlike the predictive scoring models.
Frequency tells you whether contacts are saturated or undersaturated, which is the question most teams answer by guessing.
Use it in both directions. Suppressing saturated contacts protects your deliverability; adding sends for undersaturated ones is where the missed revenue usually sits.
You upload a pool of assets, tag them, and Einstein picks the best one for each subscriber when the email is opened. You can cap how often someone sees the same asset.
One template then adapts across thousands of recipients. The catch is that it needs a genuine pool of distinct assets. Three near-identical banners give it nothing to optimize.
Copy Insights analyzes your subject line language and predicts which phrasing, tone, and punctuation drive opens. The generative side drafts subject lines and body copy from a message and a brand tone.
Treat generated copy as a first draft. It is fast at producing options and indifferent to whether your brand actually talks that way.
This watches performance and flags anomalies, up or down, so a broken link or a deliverability problem surfaces in hours rather than at the end-of-month review.
It is the feature most worth enabling early, because it costs nothing to leave running and catches the failures that cost real money. Community sessions on Apex Hours are useful for seeing how teams wire these into their operating rhythm.
This is the part missing from most feature lists, and it explains nearly every disappointed rollout. Predictive models need history to learn from, and yours may not have enough yet.
Check thresholds against current Salesforce documentation rather than assuming, since minimums differ by feature and change over time. The duplicate row is the one that quietly ruins results: identity resolution problems upstream become bad predictions downstream, which is why a properly modeled Data Cloud foundation does more for Einstein accuracy than any setting inside Marketing Cloud.
Enable the features, then wait. Judging predictions in week one is the most common self-inflicted failure here.
If you are moving to, or already on, the newer core-platform editions, the AI story is different rather than simply bigger. This is the distinction almost no feature roundup makes.
Read the first row carefully if you are planning a move. Teams that rely on Engagement Scoring personas for suppression and targeting need to know that the newer editions approach this differently, and to plan how they will replicate that logic before they migrate rather than after.
The trade is real in both directions. You give up some mature predictive models and gain a Data Cloud foundation, Flow-based journeys, and native Agentforce capability. Verify the current feature matrix with Salesforce for your own edition and contract, because this portfolio is still moving. Practitioner threads on Forcetalks are a reasonable place to see how other teams are handling the transition.
The genuine benefits are narrower than the marketing suggests, and still worth having.
Better timing with almost no effort. Send Time Optimization is the closest thing to free improvement in this list. You add it to a journey and stop arguing about send windows.
Fewer unsubscribes. Frequency and scoring together let you stop mailing people who are about to leave. Protecting list health compounds, because deliverability affects every future campaign.
Personalization without more templates. Content Selection means one email adapts per recipient rather than your team building six variants.
Faster first drafts. Generative copy shortens the blank-page problem for subject lines and body text.
Earlier warning. Messaging Insights turns a month-end surprise into a same-day fix.
What Einstein does not do is fix a weak offer, repair a bad list, or make an irrelevant campaign relevant. It optimizes distribution of what you already have. If the underlying offer is wrong, better timing delivers the wrong thing more punctually. The same logic applies on the sales side, which we covered in our guide to Einstein lead and opportunity scoring.
Five patterns account for most of the disappointment.
The third row is the most common by far. Scores get reviewed in a meeting and never reach a segmentation rule, so nothing about the actual sending changes. Certification and enablement paths on SaaSGuru help teams build the habit of acting on the output, and practical walkthroughs on Salesforce Geek cover the configuration side.
It is a set of AI features that predict engagement and personalize sending. The main ones are Engagement Scoring, Send Time Optimization, Engagement Frequency, Content Selection, Copy Insights, and Messaging Insights. Predictions are built from your own account data, not pooled across customers.
Availability depends on your edition and contract. Some Einstein capability is included at certain tiers while other features sit higher up the range, and the newer editions meter AI usage with credits. Check your own contract rather than a feature list, because this varies more than most teams expect.
Growth gives you generative AI for content and segment creation plus send time optimization, but not the predictive engagement models that Engagement provides. If persona-based scoring is central to how you segment, plan for that difference before migrating.
Engagement is the long-standing platform with Journey Builder and its own contact model. Growth and Advanced are newer editions built on the core platform with Data Cloud and Flow, sometimes referred to as Marketing Cloud Next or Agentforce Marketing. Advanced adds capacity and features such as path experiments.
Models need send history and a training period after you enable them, and the exact minimums differ by feature. The practical answer is to enable early, leave them alone through at least one full campaign cycle, and only then judge the output.
No. Einstein optimizes distribution of options you supply, while testing tells you whether the options themselves are any good. Copy Insights predicts which subject line language performs; it does not tell you whether your offer is compelling.
Einstein for Marketing Cloud earns its place by answering the four questions marketers otherwise guess at: who to send to, when, how often, and with what content. The features are mature and the effort to adopt them is low.
Two things decide whether you see the benefit. Knowing which edition you are on, so you look for features that exist rather than ones that do not. And having clean, deduplicated subscriber data with enough history for the models to learn from, because prediction quality is a data problem dressed as an AI problem.
At Minuscule Technologies we sort out that foundation first, then enable Einstein feature by feature so you can see what each one actually moved. Talk to our Marketing Cloud team about an Einstein readiness review for your edition.
You've seen what's possible. Now, let's make it happen for your business. Whether you need an end-to-end Salesforce solution, a complex integration, or ongoing managed services, our team is ready to deliver.
Schedule a Free Strategic Call
