It’s no surprise that Salesforce’s Dreamforce event in San Francisco this year was all about artificial intelligence (AI), specifically generative AI.
Also not surprising, nearly every facet of the Salesforce portfolio received an injection of GenAI – with the Marketing Cloud platform perhaps receiving the most generous dose. What is now ‘generally available’ in the platform amounts mainly to productivity improvements for go‑to‑market professionals, with embedded AI prompts that can automate or remove common steps in common GTM workflows — a tactical improvement to the platform. Salesforce’s longer-term vision, however, is strategic — with the company eyeing a model where AI powers a more seamless, multi-channel (and multi-persona), full journey orchestration of customer engagement.
There are more moving parts than meets the eye, but Salesforce has certainly set the table for a compelling AI-powered GTM automation story. To move beyond core productivity gains (i.e., beyond tactical improvements and towards this strategic vision), the company will have to continue aligning and integrating key components and making them seamlessly integrated, without significantly increasing complexity or costs for GTM teams.
We also believe the company needs a dose of a Perpetual Growth Engine strategy — i.e., a true methodology — to help guide the connectivity between these elements. If GTM software is going to deliver truly predictable productivity gains within the enterprise similar to the impact of ERP adoption, it needs a combination of software and methodology — a gap for Salesforce and for many of the GTM technology vendors in today’s marketplace.
Innovation Overview
The newly-announced Einstein 1 platform and tools allow for a more seamless integration into the core Salesforce business Cloud. We previously looked at the components of the core Einstein 1 platform, but let’s now dig into how these new AI tools enhance the Marketing Cloud and related GTM applications.
Common AI Use Cases Made Safer in Marketing Cloud
Salesforce previously announced that its Marketing Cloud would support generative AI use cases. At Dreamforce, we saw the first iteration of a more seamlessly integrated, prompt-based UI inside the Marketing Cloud via Einstein 1 Copilot. Copilot is essentially an embedded generative AI prompt interface that is becoming available out of the box inside tools like Marketing Cloud, Sales Cloud, etc.
The addition of Copilot does two positive things for marketers. First, it allows marketers to access AI tools more seamlessly from within Marketing Cloud. For example, rather than cutting and pasting or manually typing into third-party prompts, users can simply pick from a prompt menu and get more immediate and — with grounded data from the CRM — potentially more accurate or effective results. Second, leveraging the Trust Layer means marketers are not risking sensitive customer data being added to public large language models like OpenAI’s ChatGPT (and therefore available to anyone using the open-source tools). This increases productivity as well as the compliance and security of those marketing actions augmented by generative AI. It is a nice pair of benefits to add for users, positioning Salesforce as a trusted advisor as more and more teams adopt generative AI into their everyday tasks.
One of the new key elements of Marketing Cloud is the ability for the AI Cloud integration to perform “intuitive segmenting,” essentially looking at target lists and performing hyper-segmentation to improve results, or simply to better align a particular message or content with the right audience in the right context. The AI Cloud analyzes behavior and activities and combines with meta-data and CRM data, to better identify target lists. This, of course, is predicated on the fact that users have these data sources flowing inside their Salesforce universe, are compliant in their outbound communications actions to restrict potential GDPR and other noncompliance complaints, and are keeping data sanitized for accuracy.
This is still a very new process today, and somewhat manual. The roadmap is to automate a lot of this process — allowing marketers to simply set the AI against a large group of customers and let the predictive models (via integrations with SageMaker and others) analyze against unified profile data in Data Cloud to create hyper targeted segments based on propensities, affinities, etc.
Clearly there is a great deal of foundational work that needs to be done to make this type of automated campaign creation happen. Decisions made by end-user organizations in their processes, configurations, and underlying data have tremendous potential to both increase and decrease actual performance. Regardless, Salesforce believes it is laying the foundation by integrating the AI Cloud, Data Cloud and Einstein 1 into the core business Cloud.
GTM Implications
Improved Full Journey Orchestration + Personalization
The AI Cloud already included some prompts and tools for email subject line and content generation. Now, this is not only embedded via Copilot, but some new capabilities have also been added. A new deep integration with Typeface allows Marketing Cloud users to take insights from the AI Cloud’s intuitive segmentation and quickly build personalized emails, either on a 1:1 or more scalable nature. The Copilot prompt can suggest content (e.g., a product image or other content), and users can input that into the Typeface engine to create AI-powered imagery and content within seconds. In theory, these new images and content are designed to power personalized outreach across the full customer journey.
The potential productivity gains here are significant, when you consider what it takes to create a Converged Growth engagement model that spans the full journey. Theoretically, these tools unlock capabilities and data access that is typically locked away in tools to which few marketers have access. The high usability here opens up what we think of as marketing-owned outreach to more departments focused on Growth — and this is a good thing. Enabling a more granular ownership of outreach — where outcomes-driven teams can focus on the journey stages and personas they know best and build up existing Conversation Track Architectures — is a compelling concept. Compliance and auditability are an early concern, but the pros may outweigh potential cons in this case.
The updates to Marketing Cloud also include some new low/no code journey-building tools. As the Data Cloud and Marketing Cloud become more combined, new activity and demographic data can be utilized to create more dynamic experiences at more personalized levels. In addition, the new Flow capabilities included in Einstein 1 present some interesting no code journey and workflow-building capabilities that can allow for solid personalization as well. While still early going, Salesforce is presenting very intelligent journey-management tools that could and should become more and more automated: responding intuitively to buyer behaviors and engaging with the right content, at the right time, and via the right channel for every member of the buying committee and beyond.
While some of these features are becoming ‘GA’ in the near term, we would be remiss not to call out a few concerns:
- ANNUITAS clients have voiced concerns over the legacy capabilities of Marketing Cloud components when it comes to personalization, integration with Salesforce or other CDPs (Data Cloud or more legacy offerings). This is due to multiple factors, including but not limited to the varying synchronization times between legacy Pardot and CDP components, etc., that restrict the ability to truly generate real-time offers based on prospect behaviors across channels. Said differently, real-time personalization ‘fails’ due to lags in data. Performing more traditional batch nurture activities is not an issue, however.
- ANNUITAS clients have voiced issues with the need for more IT expertise than expected in configuration and management of an integrated GTM workflow that incorporates multiple Salesforce entities. Their issue is that for many marketing teams, the coding or even deep configuration skills are above their aptitude as business users — for example, when GTM workflows rely on multiple components, there might arise differing database administration skills and requirements that are only found in IT, thus increasing the complexity in making even small tweaks to workflows and campaigns in general. The whole intention of new tools like Flow and Prompt Studio are designed to alleviate some of this, but we still foresee a bit of a learning curve with these new concepts as they get rolled out to the general user base.
In the short term, Salesforce Marketing Cloud users will easily be able to take advantage of some of the GenAI capabilities for tactical use cases. For example, leveraging Copilot to create content and then manually inserting into campaigns — despite not being fully automated, this still offers significant productivity gains.
Landscape Implications
Salesforce continues to outpace its peers when it comes to both generative AI and a completeness of GTM automation vision. Other enterprise players, like HubSpot and Adobe/Marketo have been talking GenAI for some time, but Salesforce has both the strongest feature set we’ve seen, and the most “compliance-focused” tool set on the market — at least on paper. However, no matter how strong the feature set, end users still need to develop cogent, compliant strategies for injecting GenAI into their GTM processes or risk internal disruption, out-of-compliance fines, or worse.
The smaller players like SugarCRM and Zoho are adding some features, but not to the platform-level extent of Salesforce, nor with their ear to the compliance grindstone. For midsize deployments, this may be OK — users simply wish to access the tools to more quickly create or optimize existing GTM content and workflows. But at scale, we do not see any provider even talking with the level of maturity of vision as Salesforce. However, the pace of change when AI, especially GenAI, is involved is incredibly rapid — so we will continue to monitor the progress in the field of players.
The Takeaway
Salesforce points to statistics that claim generative AI can make Salesforce users up to 30% more productive. Though these exact claims are hard to substantiate, there is no doubt that some of the generative AI injected into the Marketing Cloud drives significantly more streamlined go‑to‑market actions. However, these remain ‘tactical’ improvements, reducing the time and energy required to execute tactical GTM programs. There is not similar data behind the vision of full-journey orchestration and optimization — largely because it represents a move to a Perpetual Growth Engine mindset — which is not a productivity gain but is more about driving Lift through go-to-market programs.
The introduction of the Einstein Trust Layer is an important move for Salesforce. While many competitors in the CRM and adjacent sectors are scrambling just to offer some sort of generative AI functionality, Salesforce is playing the “We’re the adults in the room” card, offering what it claims as a more secure and compliant way to quickly inject AI into your existing go‑to‑market workflows. It has even adopted the mantra “Data + AI + CRM + Trust” to support the position.
In theory, the infrastructure and core technology concepts around the Trust Layer make sense, but the proof will be in the execution, especially with the kind of scale AI usage brings to the table. Salesforce is already touting “trillions” of AI actions supported and terabytes of data processed. It will be interesting to see if the system can hold up to serious usage by its 185,000+ customers as generative AI becomes even more commonplace.
Data is the lynchpin, but it’s always been an Achilles heel when it comes to truly breaking down silos and creating a Customer Data Value Chain, regardless of the GTM tech stack involved. Salesforce is saying all the right things and is building out a vision of free-flowing, but also more easily managed, unified data profiles around each customer. It is not an easy undertaking, and as noted earlier, requires real methodology to improve the approach taken by Salesforce user organizations in their go‑to‑market. The proof will be in the pudding, when we see some major production accounts start leveraging these tools on a regular basis to engender consistent success.
