Introducing new generative AI capabilities in Adobe Experience Manager Sites

Adobe is excited to introduce new capabilities that allow users to generate variations of content using generative AI within Adobe Experience Manager Sites. The new functionality will help brands create new, personalized pages and content faster and:

• Instantly create engaging content variations with built-in generative AI prompts
• Deliver personalized content within brand and compliance guidelines
• Quickly test content variations to reach the winning content faster and drive continuous optimization

As brands aspire to tailor more of their digital touchpoints to customers’ needs, the burden to deliver effective content for each audience increases. Brands struggle to achieve this level of content creation given the lack of resources on their teams and the inefficiencies in their current tools and processes. Furthermore, understanding the impact of content requires continuous testing and experimentation, which becomes unwieldy given the massive volume of content variations needed to support all segments. Generative AI offers a path to accelerate content creation and empower teams to accomplish more.

https://video.tv.adobe.com/v/3427751?quality=12&learn=on

Create high-performing content variations without generative AI prompt expertise

As powerful as generative AI models are, your output is only as good as your input — or prompt. One of the big obstacles to generative AI adoption is the need for prompt engineering expertise to craft detailed inputs that will result in content that meets your objectives. Without a dedicated prompt engineer, brands often go through several prompting sessions and iterations to get the right output.

For generative AI to truly gain traction, it’s important to reduce the barriers to adoption for practitioners and assist them with out-of-the-box tools for prompt engineering. Prompt inputs need to capture a variety of factors — including user intent, brand guidelines, and context on page — without using proprietary information to train the model.

In Experience Manager Sites, content authors can access prompt templates optimized for enterprise use cases. These templates provide content authors with a set of pre-defined fields designed to minimize the information required to generate a quality output. The prompts are also customizable — meaning users can choose to add additional input fields if teams want to generate more contextual content. Any content that is entered into the prompts is stored separately and not used to train the model, ensuring data privacy and security for customers.

For example, let’s say a marketer is launching a new campaign, but is struggling to create content that resonates and drives results. That marketer needs to potentially test hundreds of content variations to optimize the campaign for effectiveness and maximize ROI. Manually writing the copy and designing the ad creative for that campaign would typically take hours of hands-on work for creative teams.

However, by using the variant generation capabilities in Experience Manager Sites, marketers can bypass that manual work by entering a few pieces of information into the AI-powered tool. The tool would recommend multiple variants of engaging banner experiences, where minor revisions can be made on the optimal copy selected by the marketer.

Content authors can also use Adobe Firefly to generate visual assets — giving them a full experience that includes both images and text.

Deliver personalized content at scale without compromising on brand and compliance guidelines

Generative AI may change the content management workflows enterprises are used to, but it still needs to deliver value. Many brands are faced with complex questions when looking for a generative AI solution to support content management systems such as:

• How do we ensure the content we generate is brand-aware, legally compliant, and tailored for our different audiences?
• What does governance, including reviews and approvals, look like in a world where marketing teams now create exponentially more content variations with generative AI?

Ensuring content is brand-aware requires a content author to input their brand guidelines and requirements into the prompt. To make audience-specific content, brands can import segments directly from Adobe Target, or define the audiences in a spreadsheet. The tool analyzes the characteristics or preferences of these segments and produces audience-specific variations. This can be done at scale for thousands of segments, unlocking the ability for brands to truly personalize at scale.

Typically, creating even just a handful of promotional materials means several weeks of work for creative teams. In a traditional workflow, teams work closely with many different stakeholders, including designers, content creators, and developers to determine audience segments, write copy, design creative assets, and launch campaigns. The ability to automatically generate variations helps eliminate much of the back and forth between teams due to the high quality of the initial output cycles.

Before Adobe Creative Cloud’s own product growth team started using the variant generation capability, it could take up to 75 days to create five banners. Now, the team can launch more than 60 personalized banners in just five days.

Teams can even accelerate localized content production by using the feature to deliver contextual, relevant messages faster than manually localizing content. Unlike typical translations, generative AI doesn’t translate copy word for word. Instead, it takes cultural context into account to develop variations that best connect with the new audience.

Out-of-the-box testing and optimization capabilities allow quick iterations tailored to segmented audiences

One of the challenges brands face today is the lack of testing within the organization. While they may invest in a dedicated testing tool or optimization team, they can’t always move with the speed desired with the testing process sometimes taking weeks even for a simple test.

To get the most out of content created with generative AI, brands must continually test the generated variations to validate the lift they are providing over the baseline experience. In Experience Manager Sites, content authors have access to native A/B testing capabilities that they can use to quickly learn how the generated variations are performing.

Adobe’s product growth team saw big lifts in content performance. Banners created with the new capability performed better, with half of the content delivering up to 30% higher click-through rates compared to the baseline. Because the team could generate a greater number of variations, it could personalize messages for more audiences and find new options that delivered even better performance.

Teams are empowered to assess content performance and iterate quicker without any dependency on a separate team. The more tests a content author can run, the faster they can learn and scale the use of variations across other aspects of the experience.

Learn more about generative AI tools in Experience Manager Sites

Wherever you are in your generative AI adoption journey, Adobe can help you scale the creation of high-performing, personalized web experiences. With out-of-the-box tools for prompting, tools to create personalized content that’s on brand, and native testing and experimentation capabilities, you can maximize the impact your content has on your customers and scale at the pace your business demands.

To learn more about the variant generation capability in Adobe Experience Manager Sites, check out our technical documentation.