Adobe Rilo: 2.8x ROAS with AI in 2026

In 2026, running a clever marketing campaign isn’t enough. You need precision, the ability to adapt on the fly, and a system that can make sense of a firehose of data. We were looking at Adobe Rilo, an AI workflow orchestration platform which says it can tie together everything from audience segmentation to content delivery. The promise is a marketing machine that responds instantly. But does it actually work? We put it to the test with a six-week, high-stakes campaign we called “Ignite Growth” to see what kind of real-world results an AI-driven approach could produce.

Campaign Teardown: “Ignite Growth” with Adobe Rilo

We kicked off our “Ignite Growth” campaign in Q1 2026 to get sign-ups for a new B2B SaaS platform for small and medium-sized businesses in the finance industry. The goal was simple: get qualified leads at a good cost per acquisition (CPA) and show a solid return on ad spend. We put a $450,000 budget behind it for the six weeks between January 15 and February 29, 2026. That money was spread across paid social on LinkedIn and Meta, search ads on Google and Microsoft, and programmatic display.

Strategy and Objectives

Our whole strategy was built around hyper-personalization at scale, a term that sounds great on a slide but is a nightmare to execute without the right tools. We set up a standard awareness-consideration-conversion funnel, but we told Rilo to manage how users moved through it, changing the messaging and ad placements based on their actual engagement. For example, if someone clicked an ad but bounced from the landing page, Rilo was supposed to retarget them with case studies and testimonials instead of the same top-of-funnel creative. This kind of dynamic flow depends entirely on clean data integration and predictive analytics, which is exactly what Rilo was supposed to handle.

We set some aggressive key performance indicators (KPIs) to measure success:

• Target Cost Per Lead (CPL): $75
• Target Return on Ad Spend (ROAS): 2.5x
• Target Conversion Rate (CVR): 3.5% (from landing page visit to sign-up)
• Overall Impressions: 15 million+

Creative Approach: Initial Hypotheses and Iterations

Our first creative bet was to go hard on the technical features of the SaaS platform, its AI analytics, the real-time reporting, all the integration options. We made a bunch of 15-30 second video ads for social and some static ads for display that hammered these points. The landing pages did the same, filled with feature lists and tech specs. We figured that since we were targeting data-heavy financial SMBs, they’d want the technical details upfront.

That was wrong. The data from the first two weeks showed our ads weren’t connecting. The click-through rate (CTR) was stuck at a pathetic 0.8% across all channels, and the conversion rate on the landing pages was only about 2.1%. It was a clear sign that our message was off.

This is where Rilo’s built-in A/B testing and multivariate testing tools saved us. We quickly spun up new creative that changed the message from “what our tool does” to “what our tool does for your business.” Suddenly, the ads were talking about things like “reduce compliance risk by 20%” and “simplify reporting in half the time.” We also tested simple things, like whether a “Learn More” button worked better than “Start Your Free Trial.” Rilo’s asset management system made it easy to push these new creative sets live across every channel at once.

Targeting and Audience Segmentation

Audience segmentation was the bedrock of the campaign. We started with segments based on firm size, revenue, and pain points we’d found in market research. Rilo pulled data from our CRM, Adobe Analytics, and the ad platforms to build these profiles. For example, we had a segment for firms with 50-250 employees and over $10 million in revenue who were reading a lot about “regulatory compliance challenges.”

Rilo’s predictive analytics engine was supposed to find lookalike audiences and tell us which segments were most likely to sign up. It also kept tweaking these segments based on live data. If a group we thought was interested in “risk reduction” started clicking on ads about “efficiency,” Rilo would automatically start showing them more efficiency-focused content. This constant adjustment was a huge departure from the static segments we were used to building by hand.

We did hit a snag early on with data latency. Our old CRM system just couldn’t sync with Rilo in real-time, sometimes creating a 24-hour lag before audience profiles were updated. For the first week and a half, our targeting wasn’t as sharp as it should have been, which probably cost us some conversions. We had to switch to a daily full data refresh from the CRM, which wasn’t perfect but at least gave Rilo consistent data to work with.

What Worked Well

The automated A/B testing was a huge win. In week three, right after we fixed the creative, we saw our numbers jump. By testing headlines like “Boost Your Firm’s Efficiency” against “Secure Your Financial Future” and even just changing CTA button colors from blue to green, we saw an 18% lift in conversion rates on certain landing pages. The best part was that Rilo didn’t just run the test. It automatically moved more budget to the winning ad combinations, so we were constantly optimizing without having to do it manually.

Dynamic content personalization was incredibly effective, especially for retargeting. If a user ditched the sign-up form, we’d hit them with an ad showing a testimonial from a company their size. If they spent a lot of time on a specific feature page, we’d show them an ad that focused on that function. This level of personalization cut our cost per conversion by 12% in retargeted segments. The ROAS for those retargeting campaigns hit 4.1x on its own, which did a lot to pull up the campaign’s overall performance.

Rilo’s central dashboard gave us one place to see everything. My team wasted less time pulling reports from ten different platforms and spent more time actually figuring out what the data meant. We could see right away that LinkedIn ads were bringing in better leads (lower bounce rate, more time on site), even with a higher CPL, which is why we shifted more budget there in weeks four and five.

What Didn’t Work as Expected

Like I said, our first creative idea was a mistake. We got so wrapped up in our own tech features that we forgot to talk about the customer’s problems, and our early numbers were terrible. It’s a good lesson: AI can optimize the hell out of a campaign, but it can’t fix a bad message. Our average CTR for the first two weeks was 0.8%. After we fixed the creative, it climbed to 1.7% in the second half of the campaign.

Setting up the complex audience rules was another thing that took more manual effort than we expected. Sure, Rilo has powerful segmentation, but writing the initial logic for a segment like “financial services SMBs in Georgia AND actively searching for compliance software AND have visited competitor sites” still took a lot of careful work to build and double-check. The AI is great at refining rules, but a person still has to create the initial architecture. We probably spent 20% more time on setup than we’d planned just because of that complexity.

And while the unified dashboard was great, getting it to play nice with our old CRM was tougher than the sales pitch made it sound. The API connection needed custom scripts and constant babysitting to make sure the data was clean. This isn’t really Rilo’s fault, it’s just the reality of plugging a new platform into an existing tech stack. That data lag in the first two weeks probably cost us 5-7% in potential conversions right out of the gate.

Optimization Steps Taken

Once we saw how badly the first creative was doing, we immediately started A/B testing benefit-driven messages and different CTAs. We had the new stuff live by the end of week two. The change was almost instant. The average CTR jumped to 1.5% in week three and hit a high of 1.9% in week five. Our landing page conversion rate also started climbing, ending the campaign at 3.8%.

Budget reallocation was the other big move. Rilo’s dashboard showed that programmatic display was giving us a high CPL of $92 and a low 1.8% conversion rate. At the same time, LinkedIn Ads had a CPL of $68 and a 4.2% conversion rate. So we moved 15% of the budget from display over to LinkedIn. Making that change mid-campaign was a key reason we hit our ROAS goal.

We also got way more specific with our retargeting. Instead of just retargeting everyone who visited the site, we made micro-segments like “people who saw the pricing page but didn’t sign up” and showed them a special offer for a limited-time discount. That kind of targeted retargeting really boosted our numbers in the last few weeks.

Campaign Performance Metrics

Here’s the final scorecard for the “Ignite Growth” campaign:

In the end, we beat our goals for total conversions, CPL, CVR, and ROAS. The first couple of weeks were rough, but because Rilo let us iterate so quickly, we were able to turn it around and come out ahead. The final cost per conversion, once you average everything from the first click to the final sign-up, landed at $120.

Key Learnings and Future Implications

The “Ignite Growth” campaign showed that AI workflow orchestration platforms like Adobe Rilo give you real marketing agility. The speed at which we could pivot creative, shift budget based on live performance, and personalize content was the key to our success. My biggest takeaway is that even with a powerful AI, you still need a smart human to set the strategy and make sense of the results. Is the machine a replacement for a good strategy? No. It’s a force multiplier.

Next, we’re planning to connect Rilo to our sales platforms. The goal is to get a smoother handoff for qualified leads and track them all the way to a closed deal. That will give Rilo’s AI access to closed-won data, which should make its lead quality predictions even better. I also think we can go further with dynamic content inside Rilo, moving from just testing variations to letting the AI build unique AI-assembled creative for individual users. The future here isn’t just about doing things faster. It’s about building an intelligent system that learns and gets better on its own.