The Marketing Keynote at Dreamforce ‘23, titled “Marketing in Generation AI”, was full of new innovations that the Marketing Cloud engineering teams have been working hard on shipping, wrapped together by awesome demos. “Inside Marketing Labs” highlighted the ‘sneak peek’ nature of what was showcased, with expected general availability (GA) dates shared.
Steve Hammond, GM of Marketing Cloud, kicked off the keynote, emphasizing ‘efficiency’. What was once time-consuming and expensive, personalization at scale is set to become far more efficient. Plus, marketing teams are accountable to many other functions across the organization, with hand-offs of information to service, merchandising, sales, and the list goes on.
What we were treated to in this session was generative AI offerings that are set to take on these two challenges, head-on.
1. Prompt Engineering and Grounding for Marketers
The session started with context on prompt engineering – the art of writing prompts to get the most optimal answer. It’s a skill that barely existed 6-12 months ago, but is one that we need to all acquire in order to use AI effectively. The takeaway is that “AI is only as good as your data”, in this context, only as good as what you disclose to Einstein when writing prompts.
Here is the example that illustrates this point, the output with grounding – what CRM data should be used to produce outputs – and the comparison to without grounding. You have data about your customer, and so tap into for better outputs.
Of course, you don’t want to share the customer’s personal information with the large language model (LLM) for compliance reasons, but you can safely share segment data. This is where the Einstein Trust Layer comes in, taking care of the background processing such as data masking, zero-retention, and more.
2. Campaign Recommendations
Campaign Recommendations are quite self-explanatory, with Einstein able to drum up a suggested campaign which you can opt to start executing in Marketing Cloud. The image below shows this in the Marketing for Slack app which, in the spirit of Einstein Copilot, has become more proactive.
3. Campaign Briefs and Preview
Campaign Briefs are data-driven – in other words, they are created using AI and the wealth of historical campaign insight you have. These are interactive, meaning you can work with Einstein to reiterate the plan, for example, to include a new/extended segment (as shown below).
In this section, find a campaign overview ‘preview’ sub-tab. From what you and Einstein have just created, visually see the segment overview, affinities, channel preferences, and more. Also, if my eyes don’t deceive me, do we see a new Marketing Cloud user interface?
4. Content Collections
Content Collections within Content Builder group assets for a specific campaign, bringing together variants of an asset that you’ve created – and of course, the ability to create further AI-generated images and copy variations for different personas. This paves the way for more personalization and experimentation.
Assets within Content Collections can be rendered in different dimensions and formats for use across different channels.
5. Real-Time Campaign Updates, Experiments, Goals (in Slack)
Salesforce have been promoting Slack heavily as the key place for collaboration – the control panel to surface insights and kick-off workflows. As mentioned, Slack has become more proactive, making smart suggestions to you.
In the Marketing for Slack app, we will be alerted when a campaign experiment has completed. Experiments include those crafted with the support of generative AI functionality, such as subject line experiments and banner experiments.
The dashboard will also feature a goal summary with metrics such as the impact on CLV (customer lifetime value) and other custom metrics important to your business.
Demo #1: Slack for Marketing + Briefs + Content Builder
Here are the products featured in the first demo, stitched together in a cohesive user story:
- Slack for Marketing App
- Campaign Recommendations
- Campaign Briefs
- Campaign Collections (Content Builder)
- Content Builder approval via Slack
- Campaign Updates, Experiments, Goals (in Slack)
6. Segment Intelligence for Data Cloud
Generally available in October 2023, this tool enables rapid segment building without having to jump between different interfaces. With generative AI, Segment Intelligence serves up recommendations on which attributes to target in your next campaign.
For example, a marketer may see that one segment is performing poorly, and drill down into further attributes that paint a picture of customers buying a product infrequently because of where they are based.
- Segment performance charts in a pre-built dashboard that connects segment, engagement, paid media, and revenue data.
- Attribute breakdowns (e.g. how long ago the purchase was made) and double attribute breakdowns which pair two pieces of data together (e.g. how long ago the purchase was made, by location).
- “Einstein Recommends” is a tooltip that suggests changes to improve the engagement for that segment, without the marketer having to trawl through masses of data.
7. Email Content Creation and Typeface Partnership
Now we come to the Typeface partnership that Salesforce announced earlier this year. Typeface delivers generative content (images and layouts) that are specific to your brand guidelines.
In Content Builder, you can see custom image content being generated for an email with the user inputting conversational prompts. Like/dislike the outputs until you find one that is suitable.
8. Einstein for Account Engagement
On a side note, Segment Creation, Email Content Creation, and Typeface partnership will be available for Account Engagement (formerly Pardot) in February 2024.
9. Journey Re-Mapping
Marketing Cloud Personalization (formerly Interaction Studio), is Salesforce’s real-time interaction management offering.
A decision node can be added to Journey Builder to listen to data from Marketing Cloud Personalization, and react rapidly to input the contact on the journey path that’s most relevant.
10. Real-Time Customer Event Stream
This will be a two-way connection between Marketing Cloud journeys/Marketing Cloud Personalization and Salesforce core. Insights are served up within page layout components. For example, if a prospective customer raises a case while purchasing a product, the components with a case record page can guide agents to support them based on their affinities and interaction behavior.
- Affinities: Filter by the type of interaction event to paint a picture of their product preferences within a certain timeframe.
- Event Stream: Filter by interaction event types, within a certain timeframe.
That’s a wrap on the marketing updates and features you may have missed at Dreamforce ’23! The generative AI offerings are very exciting and show us a ‘sneak peek’ into the future.
Which was your favorite? Were there any announcements you were hoping for that didn’t happen? Let us know in the comments…