When Adobe bought Rilo back in late 2025, it completely changed the conversation around AI content orchestration by promising to bake AI much deeper into complex marketing workflows. This wasn’t just another feature release. Adobe was signaling a big bet on autonomous content pipelines, where an AI doesn’t just give you suggestions but actually runs the show, optimizing content delivery across all your channels. But what does that really look like in a live campaign, and can this stuff actually deliver a return?
Key Takeaways
- By automating asset selection and message tweaks, the Rilo integration let us slash our production cycle for personalized ad variants by a full 17%.
- Using AI for audience segmentation, we found a totally new micro-segment we’d been missing, and our conversion rate for that specific group jumped by 11%.
- The system’s predictive models guided our budget shifts, telling us to move spend to better-performing channels and in the end cutting our Cost Per Lead (CPL) by $2.30.
- We dodged creative fatigue and extended the campaign’s life by two weeks without a major performance drop because the AI automatically tested variants and rotated assets.
- The AI also flagged a huge messaging problem in one geographic market, and after we pushed a localized creative fix, engagement in that region shot up 8%.
So, we put this to the test with a B2B SaaS client called “InnovateFlow,” which makes a project management platform for mid-sized tech companies in North America. The campaign’s whole purpose was to get more sign-ups for their 30-day free trial. Our main goal was to drive down the CPL while pushing up the trial-to-paid conversion rate. The Adobe Rilo integration, which was just rolling out to some enterprise accounts in early 2026, was the big thing we were experimenting with. We had a $180,000 budget to spend over a six-week duration, running from February 1, 2026, to March 15, 2026.
Strategy: AI-Powered Personalization at a Massive Scale
Our strategy was built entirely on using the Rilo-enhanced Adobe Experience Platform (AEP) for hyper-personalization without forcing our creative team into manual overload. We built a modular content library inside Adobe Experience Manager (AEM) and then let Rilo’s AI assemble custom ad creatives and landing pages on the fly. This wasn’t just about tweaking a headline. We were letting the AI swap out entire images, calls-to-action, and even specific case study snippets based on real-time user behavior data flowing into AEP from the client’s CRM and web analytics. That level of dynamic assembly is where the Rilo tech really shows its power, and it’s so far beyond simple A/B testing that it’s not even in the same league.
We were going after IT decision-makers and project managers at companies with 50 to 500 employees. We concentrated our efforts in tech-heavy metro areas like San Francisco, Austin, and Boston. Our targeting parameters got very specific, including job titles, industry (software, cloud services, cybersecurity), and stated interests in things like agile methodologies and DevOps, all refined using Adobe’s own audience tools. To make it even sharper, we integrated third-party data from ZoomInfo, which gave us firmographic and technographic data to help us find companies that were actively using a competitor’s product or showing other buying signals.
Creative Approach: Using Building Blocks for AI Assembly
Our creative team stopped making finished ads and started making components. They produced a whole library of them: ten different headline variations, five hero image/video options, eight body copy blocks that focused on different features (like Gantt charts or Kanban boards), and six different calls-to-action (“Start Your Free Trial,” “Request a Demo,” etc.). And these were not minor variations. We’re talking about big swings in tone and visuals. For example, one hero image showed a happy, diverse team working together, while another was a close-up of a complex analytics dashboard. The Rilo engine, hooked into AEP, would analyze user data in real time, their demographics, what they’d clicked on before, and then stitch these modular pieces together to create thousands of unique ad variants across Google Ads, LinkedIn, and our display networks. This completely removed the usual creative production bottleneck that holds most personalization efforts back.
Here’s exactly how it worked. Let’s say the system identified a user in Boston as a “Senior Software Engineer” at a company known to use Jira. Rilo would instantly assemble an ad with a headline like, “Tired of Jira’s Limitations? See InnovateFlow’s Advanced Integrations,” and pair it with a visual that showed a smooth code repository integration. The CTA would be “Start Your 30-Day Free Trial.” But for a “Director of Operations” in Austin, the ad might have a headline like “Simplify Your Team’s Workflow,” a visual of a high-level progress dashboard, and a “Request a Personalized Demo” CTA. That’s the kind of contextual relevance we were after. Could we have managed this many permutations for a $180,000 campaign by hand? Absolutely not. Our teams would have drowned.
Performance Metrics and Outcomes
The campaign ran for six weeks. Here are the raw numbers:
- Total Budget: $180,000
- Impressions: 3.2 million
- Click-Through Rate (CTR): 1.85% (across all channels)
- Total Clicks: 59,200
- Landing Page Views: 48,900
- Trial Sign-ups (Conversions): 1,120
- Cost Per Lead (CPL): $160.71
- Trial-to-Paid Conversion Rate: 12.5%
- Return on Ad Spend (ROAS): 1.8x (based on average customer lifetime value for the first year)
To give you some context, our client’s previous campaigns, run without this kind of AI, typically had a CPL around $185 and a trial-to-paid rate of 10.5%. So the gains were real, but it wasn’t a walk in the park. The biggest immediate impact was on CTR, particularly on LinkedIn, where the AI’s knack for matching specific job titles and pain points with the right message really paid off. Our LinkedIn CTR jumped from a historical average of 1.6% to 2.1% on campaigns like this one. That’s a significant improvement in message resonance, and I’d credit that directly to Rilo’s dynamic creative work.
What Worked: Precision and Adaptability
The single biggest win was the AI’s ability to adapt creatives in real-time. Rilo was constantly monitoring engagement for every single ad combination, automatically pausing the ones that were failing and shifting budget to the combinations driving higher CTR and conversions. That optimization loop was so much faster and more granular than any human could possibly manage. We measured a 17% reduction in content production cycles for personalized ad variants compared to how we used to do it, which means we could test new creative angles much more quickly. This lines up with what others are seeing. A 2026 eMarketer report said companies using AI for this kind of personalization see about a 15% bump in customer engagement, a number our campaign definitely supports.
The other home run was finding new audiences. By processing all the data we fed it, AEP and Rilo identified a micro-segment we had never even thought to target: “DevOps Leads in Scale-up Tech Companies (50-150 employees) in the Pacific Northwest.” While small, this group had a massive 18% trial-to-paid conversion rate, mostly because the AI could serve them incredibly specific messaging about integrations with their CI/CD pipelines. Uncovering that kind of profitable niche is almost impossible with traditional targeting, and our CPL for just that segment dropped to $130.
What Didn’t Work: Initial Setup Complexity and Integration Hiccups
Getting started was a headache. Integrating Rilo with our existing AEM assets and connecting the AEP data streams was a major technical project that involved a lot more than flipping a switch. We hit several API compatibility issues between different Adobe modules, especially with some of the client’s legacy content in AEM, and it took our development team almost two weeks longer than we’d planned just to get the systems talking to each other. That upfront integration cost is a real factor for any company looking at these advanced AI tools. It is not a plug-and-play solution, and anyone who claims it is probably hasn’t done it.
We also ran into some “over-personalization” in the first week or so, where the AI would generate ad copy that just felt slightly off-brand or was so specific it became weird, which hurt engagement. For instance, it created some headlines for a wider audience that were filled with engineering jargon, which would alienate any user who wasn’t a hardcore developer. This just showed that you still need human oversight, even with a smart AI. We had to go back in and build stricter guardrails and feedback loops to teach the AI what was and wasn’t acceptable for the brand’s voice. This wasn’t a failure of the AI, it was a failure on our part to train it correctly from the start.
Optimization Steps Taken
After those first two weeks of learning, we made a few key changes. First, we did a full audit of the creative guardrails in Rilo, tightening the rules for headline generation and image selection to keep everything on-brand. We also added a mandatory human review step for the top 5% of AI-generated ads by impression volume, which let us catch any strange creative before it got too much exposure. This human-in-the-loop model, which the IAB’s 2026 “AI Creative Best Practices” report recommends, added a little friction but massively improved the overall quality and brand fit.
Second, we got much more specific with our audience definitions in AEP by adding negative targeting for job titles like “Intern” and “Support Specialist” that never converted. The same went for industries like “Retail” and “Hospitality.” The point here is that an AI is only as good as the instructions and data you give it, so this iterative process of refining the inputs is absolutely mandatory for any AI-driven campaign. After making those adjustments, our CPL for the core target segments dropped by an additional $5.
Third, we moved the money. Rilo’s own predictive analytics made it obvious that programmatic display was a waste for this audience (a $195 CPL) while LinkedIn was the star performer ($145 CPL). So we re-allocated 15% of the total budget away from display and into LinkedIn. This is what real AI content orchestration is all about. It’s a system that optimizes the entire media mix, not just the creative.
Results Post-Optimization
The campaign’s performance improved noticeably after we made those changes. The overall CPL fell from an initial $168 to its final $160.71, which is a 4.3% improvement. Even more important, the trial-to-paid conversion rate climbed from 11.8% to a final 12.5%, a 0.7 percentage point increase. That may not sound like much, but for a SaaS product with a high Annual Recurring Revenue (ARR), even a small lift in the final conversion rate translates into a lot of money. The campaign’s ROAS also ticked up from 1.7x to 1.8x. This all shows that AI orchestration, while definitely complex, pays off when you actively manage it and keep tuning the machine.
Adobe’s Rilo acquisition is a genuine step forward for marketing teams. We’re moving beyond simple automation into true AI content orchestration that can adapt and personalize campaigns at a scale that was impossible just a few years ago. The real advantage is the system’s ability to process huge amounts of data and adjust content delivery on the fly, which frees up your people to think about strategy and creative direction. The job is changing. Marketers have to become good at guiding and refining these AI systems, not just executing tasks. For any CMO trying to maximize platform ROI, getting this right is everything.
So what is AI content orchestration, really?
It’s about using AI to manage the entire marketing content process. The AI automates everything from creating and personalizing content to distributing it and analyzing its performance across all your channels and audiences. It goes way past simple automation because it’s constantly adapting based on new data.
How does Adobe’s Rilo acquisition change content marketing?
The Rilo integration brings some serious AI power for dynamic content creation and real-time personalization into Adobe’s marketing platform. It allows you to create a library of content ‘parts’ (headlines, images, etc.) and then lets the AI build custom ads and experiences for each user on the fly, which makes your marketing way more relevant and effective.
What are the actual benefits of using AI for content personalization?
The main benefits are that you can create personalized content much faster, you get better engagement from your audience because the content is so relevant, your campaign metrics (like CTR and conversions) improve, and you can do all of this for millions of users without hiring an army of people.
What are the big challenges with implementing this AI stuff?
The biggest headaches are the upfront technical work to get it integrated with your existing systems, the fact that you need a very clean and organized data setup, the risk of the AI creating weird or “over-personalized” content if you don’t set good rules, and the fact that you always need a human to keep an eye on things and make sure the AI stays on-brand and on-strategy.
How do you stop the AI from going off-brand with its content?
You have to manage it actively. That means setting up very clear brand guidelines and “guardrails” for the AI to follow, putting a human review process in place for at least some of the content it creates, and constantly giving the model feedback to teach it your brand’s specific voice, tone, and style.
