Heading 1
Heading 2
Heading 3
Heading 4
Heading 5
Heading 6
Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur.
Block quote
Ordered list
- Item 1
- Item 2
- Item 3
Unordered list
- Item A
- Item B
- Item C
Bold text
Emphasis
Superscript
Subscript
The decision on the table is which AI features to switch on, in which order, and how to prove they pay off. Switch on everything at once and you cannot tell which feature moved your results. Start with the predictive layer and you get measurable gains on journeys you already run, plus cleaner engagement data for the generative features that follow.
Predictive features decide who, when and how often
Marketing Cloud Engagement, the Salesforce email and mobile messaging platform, includes several predictive Einstein features that personalise without writing a word. Einstein Engagement Scoring predicts how likely each contact is to open, click or unsubscribe, and groups contacts into engagement personas. Journeys can then branch on those scores, sending loyal openers one path and lapsing contacts another.
Einstein Engagement Frequency uses the same signals to show which contacts are under-saturated, on target or saturated, so you can adjust how often each group hears from you. Einstein Send Time Optimization scores every hour of the week for each contact and sends when that person is most likely to open. None of these needs new content. They change the timing and volume of messages you already send.
Volume decides how much the predictive layer returns
These models learn from engagement history, so their value grows with the number of sends and contacts. Engagement Frequency can fall back on the behaviour of your whole contact base when a new subscriber has no history yet, which makes it usable early in the customer lifecycle. For a small list with a monthly newsletter, the gains are modest. For a retailer sending triggered journeys every day, small improvements in timing and frequency repeat across millions of messages.
That is the scale our marketing automation work runs at, with campaigns that reach two million recipients and journeys built on behavioural data. At that volume, frequency control protects the list as much as it lifts engagement, because every avoidable unsubscribe is a contact you cannot reach again. How predictive and generative AI divide the work in Salesforce more broadly is set out in predictive and generative AI compared.
Recommendations personalise the content block
Einstein Recommendations fill an email or web block with the products or articles each contact is most likely to want. You define the rules, such as which catalogue items qualify, which ones to exclude and which to boost, and the model weights the rest by each contact's behaviour and affinities. In Marketing Cloud Personalization, the same logic runs through recipes that combine those rules.
Recommendations are only as good as the catalogue and behaviour data behind them. Missing categories, stale stock or untracked browsing all show up as irrelevant suggestions. Fix the feed before you judge the model. Where behaviour sits in several systems, it needs unifying first, which is the case made in Data Cloud as a customer foundation.
