AI Partnerships: $150K Campaign Hits 3.2x ROAS in 2026

The way brands talk to customers is being completely upended by AI, especially through what we’re calling agentic commerce. This is a breakdown of a campaign we ran that was all about partnering with AI agent platforms to get right in front of consumers and drive sales. This kind of strategic change isn’t just about plugging in some new software. It’s about totally rethinking the customer journey, swapping out passive website clicks for active, AI-led conversations that are built to convert. So how do you actually make these new AI partnerships work without wasting a ton of money?

Our goal was pretty straightforward: get our product catalog plugged into these new AI agent platforms so people could find, ask about, and buy our products just by talking. We figured this method, “agentic commerce,” would cut out a lot of the usual friction in the buying process. The campaign itself, which we called “AI Direct Connect,” ran for six weeks between early March and mid-April 2026, and we put a $150,000 budget behind it. This was never meant to be a fluffy awareness campaign. It was a hardcore performance initiative with hard conversion goals.

Strategy: Embedding in Conversational Ecosystems

Our whole strategy was to find the AI agent platforms our target customers were already using and partner up with them. We weren’t about to build our own AI from scratch. We integrated into existing, popular systems. Our research showed a big slice of our audience, people aged 25-45 who are into sustainable home goods, were using AI assistants more and more for product research. We went after platforms that could handle direct transactions and had solid APIs for syncing our product catalog, which led us to integrations with Google Assistant’s shopping actions and Amazon Alexa’s purchasing features. We also tested the waters with some newer, specialized AI platforms for eco-friendly products. Their user numbers were smaller, but the communities were incredibly dedicated.

We made the call to stick with these bigger platforms because they already had the users and the tech was mature. We didn’t have the time or the money to mess around with platforms that were still in beta or would have required a massive custom development project, especially on a six-week timeline. The mission was to show up where the customer was already asking questions. In practice, this meant a lot of grunt work tailoring our product data feeds to match each platform’s unique schema so that every price, stock level, and description showed up correctly.

Creative Approach: Conversational Design and Dynamic Prompts

Our creative work here had nothing to do with banner ads or video. It was all about designing conversation flows that actually worked. We developed a whole library of “agent scripts” to walk users through discovering, comparing, and buying products. These scripts had to feel natural and helpful, sort of like a great retail associate. For instance, if a user asked, “Show me sustainable coffee makers,” the agent wouldn’t just dump a list of links. It would come back with a curated selection and ask, “Are you looking for a single-serve option or a larger capacity?” to narrow things down.

We set up dynamic prompt engineering so the agent’s responses would actually change based on what the user said and what they’d asked about before. This let us create a much more personal experience than a static product page ever could. Our creative team, which now included conversational designers, spent a ton of time obsessing over the tone to make sure the AI came across as trustworthy, not some creepy robot. We also had to build clear calls to action like “Add to cart” or “Compare features” right into the dialogue. Honestly, it was a steep learning curve for our creative department, who had to learn to think in dialogue trees instead of visual layouts, and our first scripts were way too stiff.

Targeting and Placement: Intent-Driven Conversations

Targeting for this campaign was a completely different animal than what we’re used to in digital ads. We weren’t targeting by demographics or interests. We focused on what we called intent-driven conversational triggers. We worked directly with the AI platform companies to figure out the common questions people were asking about our product categories. On Google Assistant, for example, we’d bid on phrases like “best eco-friendly kitchen appliances” or “where to buy sustainable cookware,” letting us pop up right when a user was actively trying to solve a problem.

There are no “ad slots” in this world. Placement is all about being the best, most relevant answer the AI can find for a user’s query. Our partnerships gave us preferred listings on some platforms, which made sure our products got top billing for relevant questions. We also built out “agent skills” or “actions” that a user could call up by name, like saying, “Hey Alexa, ask [Our Brand Name] about their latest sustainable products.” This required some existing brand recognition, of course, but it was a powerful way to engage with loyal customers. You just had to be there at that exact moment of consideration.

Performance Metrics: What Worked and What Didn’t

The campaign pulled in 3,529 conversions over six weeks. On a $150,000 budget, that put our cost per conversion (CPC) at $42.50. Yes, that’s higher than our standard search campaigns, but the much higher average order value (AOV) from these agent sales made it worthwhile. All told, we generated $480,000 in revenue directly from the AI agents, giving us a return on ad spend (ROAS) of 3.2x. Our initial goal was 2.5x, so we were pretty happy with that result.

Initial Phase (Weeks 1-2):

• Impressions: 1.2 million (conversational prompts delivered)
• Click-Through Rate (CTR): 4.5% (users engaging with the agent’s initial response)
• Conversion Rate: 1.8%
• Cost Per Lead (CPL): $8.50 (for users who progressed beyond initial interaction)
• Average Order Value (AOV): $120

The direct engagement was great right out of the gate. Anyone who started a conversation about a specific product was obviously a super qualified lead. The novelty factor of shopping with an AI also seemed to help. But that initial 1.8% conversion rate was a problem. Looking at the chat logs, we could see people bailing when the conversation got too complicated or when they couldn’t get a straight answer fast enough.

Optimization Phase (Weeks 3-4):

We immediately started A/B testing our conversational scripts. We made the product descriptions shorter, put the “buy now” options earlier in the flow, and added a “speak to a human” escape hatch for anyone getting stuck. We also worked on the prompt engineering to make the AI sound more empathetic and less like a cash register. For instance, we changed “Do you want to buy this?” to the much softer “Would you like to add this to your cart, or explore other options?”

• Impressions: 1.5 million
• CTR: 5.1%
• Conversion Rate: 2.7% (a significant jump)
• CPL: $7.20
• AOV: $135

The jump in conversion rate and AOV in this phase came directly from those tweaks. The big lesson was that users in these AI chats want efficiency and clarity more than anything else. And that “speak to a human” button? Almost no one used it, but having it there was a huge trust signal that seemed to reduce the perceived risk of the interaction.

Refinement Phase (Weeks 5-6):

In the final two weeks, we got a bit more sophisticated and started personalizing recommendations inside the conversation itself. If someone was asking about coffee makers, the agent would follow up by suggesting our reusable coffee filters or organic coffee beans. We also did some backend work to optimize our product data feeds for speed, cutting down the load times within the AI interface to make the whole experience feel snappier.

• Impressions: 1.8 million
• CTR: 5.8%
• Conversion Rate: 3.5%
• CPL: $6.50
• AOV: $145

What didn’t work at first was our assumption that a purely transactional script would be enough. People expected some nuance, almost like they were talking to a human expert. We also completely fumbled out-of-stock items at first, which just annoyed people by telling them something was “unavailable.” We quickly patched that by having the AI proactively suggest good alternatives or offer to notify the user when the item came back in stock.

Optimization Steps Taken

Our optimization was a constant, data-obsessed process. We were in the conversational logs every single day, looking for the exact points where users were getting frustrated and abandoning the chat. This qualitative data was gold. We ran weekly A/B tests on everything from single phrases to the structure of the AI’s responses. We found tiny changes could make a big difference. For example, switching the opening line from “What are you looking for?” to “How can I help you find the perfect item today?” gave us a 10% bump in initial engagement.

We also spent a lot of time and effort fixing our product data infrastructure. An AI agent is only as smart as the data it’s fed, so making sure our product catalog was clean, properly categorized, and easily searchable was absolutely critical. That meant standardizing attributes and adding more descriptive tags. A solid data foundation is completely non-negotiable for agentic commerce. I see so many brands get excited and jump right to the AI part without cleaning up their data first, and it’s a recipe for failure every single time.

Lessons Learned and Future Outlook

So, what did we learn from the “AI Direct Connect” campaign? It proved that partnering with AI agent platforms can be a legitimate, high-performance sales channel. The trick is you have to respect the unique rules of conversational commerce. It demands a totally different creative skillset, a focus on user intent, and a relentless commitment to optimizing the chat flows. Commerce is becoming more and more conversational, and the companies who figure this out now are going to have a massive head start. This isn’t just another marketing channel to test. It’s a real change in how people find and buy things, and the brands investing in conversational design and AI integration today are the ones who will define the next wave of customer experience.