Smart Living 2026: AI Boosts ROAS 35%

Our client, a DTC electronics brand that’s been growing fast, brought a new smart home device to market in Q1 2026. The goal was simple on paper: get strong initial sales and make people aware of the product. We decided to use fractional attribution to get a true read on every touchpoint in what we knew would be a messy customer journey. We just knew that sticking with last-click data would make us blind to all the early engagement that’s so important for a product that needs a bit of explaining. So how do you properly credit all the little interactions that lead to a purchase when your AI marketing tools are busy creating thousands of different customer paths?

Campaign Teardown: “Smart Living” Device Launch

The “Smart Living” device was a premium, easy-to-use piece of home automation hardware. Our campaign ran from January 1 to March 31, 2026, and was built to educate people, get them interested, and in the end drive them to buy. The brand’s old reports were always based on last-click attribution, which meant their upper-funnel work always looked like it was failing. Our whole pitch was to use a data-driven fractional model to see how every channel and ad actually worked together.

Strategy and Objectives

We built a strategy on the fact that buying a smart home device isn’t an impulse decision. It’s a process of research, comparison, and learning to trust the brand. So we set three main objectives for the campaign:

1. Awareness: Get the “Smart Living” device in front of a wide but still relevant audience.
2. Consideration: Teach potential buyers about its specific features and why it was better than the other options out there.
3. Conversion: Push for direct sales on the brand’s own e-commerce site.

We set up KPIs for every stage, tracking impressions and CTRs, but also video views, time on page for content, email sign-ups, and of course, final sales. The whole thing was backed by a $750,000 budget for the quarter.

Creative Approach: AI-Powered Personalization

We leaned hard into AI marketing for our creative, using a platform that works a lot like Persado to spin up and test ad copy, plus an AI-driven tool for the visual ads. For the awareness phase, the creative was all about aspirational lifestyle shots and quick 15-30 second video ads that showed the convenience of smart living. We ran headlines like “effortless control” on Meta, TikTok, programmatic display, and YouTube. When we moved to the consideration phase, the creative got much more detailed. We had 2-3 minute videos that demonstrated specific uses for the device, interactive product tours on our landing pages, and even some comparison charts. The ad copy here was all about solving pain points and showing off the device’s technical specs and easy setup, which we pushed through retargeting, email, and native ads. For conversion, it was all business: direct calls to action, some limited-time launch offers, and a lot of social proof like early reviews. These ran in our lower-funnel retargeting, SEM, and direct email pushes.

Targeting and Channel Mix

Our targeting was layered. We started with broad demographic and interest targeting (think “smart home enthusiasts” or “tech early adopters”) for the awareness stage. As soon as someone engaged, we segmented them into a consideration or conversion audience for retargeting. The channel mix was pretty standard:

• Social Media Ads (Meta, TikTok): Mostly for awareness and consideration, with a heavy dose of short-form video.
• Programmatic Display: Used for broad reach at the top of the funnel and retargeting in the middle.
• Search Engine Marketing (Google Ads): Focused on high-intent keywords for the conversion stage, with a few broader terms for consideration.
• YouTube Video Ads: We used a mix of pre-roll for awareness and longer in-stream demos for consideration.
• Native Advertising: This was our vehicle for getting our consideration-phase articles and reviews out there.
• Email Marketing: The workhorse for nurturing leads from their first sign-up all the way to purchase.

Implementation of Fractional Attribution

This is what made the whole campaign different. We ditched last-click for a custom fractional model that assigned credit based on where an interaction happened in the journey and how much influence we thought it had. It was a data-driven model, similar to what you get with GA4’s data-driven attribution, but we customized it for this specific product’s longer sales cycle. This is how we split the credit:

• First Touch: Got 20% of the credit, because we had to acknowledge where the journey started.
• Mid-Journey Touches (e.g., video views, blog reads, interactive demos): Got a combined 50% of the credit, which we distributed based on how long someone engaged with the content.
• Last Touch: Got the remaining 30% of the credit, since it did the job of closing the deal.

We put this model together using our internal data warehouse connected to a marketing attribution platform. It was a ton of work to get user IDs tracking correctly across every touchpoint and required integrating data from our CRM, all the ad platforms, and our website analytics.

Performance Metrics and Analysis

Here’s a look at the numbers from the three-month campaign:

The results speak for themselves. The budget didn’t change, but the campaign blew away the old benchmarks from last-click models. That 40% jump in ROAS, from 2.0x to 2.8x, was the big one. It wasn’t just a number. It was a direct result of us being able to move money around more intelligently.

What Worked Well

1. Fractional Attribution’s Clarity: The biggest win was finally seeing the whole customer journey clearly. Social media awareness campaigns, which last-click always labeled as “top-of-funnel fluff,” were actually contributing almost 20% of the total conversion credit. Our detailed YouTube explainer videos, which didn’t drive many immediate clicks, were earning 15% of the credit in the consideration phase, which gave us the ammo to defend (and even increase) the budget for these early-stage ads.
2. AI-Driven Creative Personalization: Being able to quickly test different ad copy and visuals for each stage was a huge advantage. For example, the AI figured out that headlines about “energy savings” worked great for people who’d already read sustainability content, while “smooth integration” worked better on users who had been looking at competitor sites. That kind of granular personalization is what got us the 28% CTR increase.
3. Integrated Data Approach: Pulling data from our ad platforms, web analytics, and the CRM into one place gave us a complete picture. We could actually follow a user from a TikTok ad to a blog post, then to an email, and finally to a purchase, with every step getting its proper weight. This made our retargeting segments incredibly precise.

What Didn’t Work as Expected

1. Initial Over-reliance on Broad Targeting for Awareness: In month one, our programmatic display campaigns were way too broad. We got a ton of impressions, but the CTR was low and our CPL for early leads was higher than we wanted. We had to get in there and refine the audience segments pretty quickly.
2. Underestimation of Content Production Needs: The AI was great for ad copy, but it still needed a steady diet of high-quality, long-form video and interactive content for the consideration phase. We didn’t have enough stuff ready at the start to feed the personalization engine properly, which caused some delays in the first few weeks.
3. Data Integration Challenges: Even with a plan, getting data to talk to each other across Meta Ads Manager, Google Analytics, Salesforce, and our own attribution platform was a headache. We had some early data discrepancies that took a lot of manual work and API tweaking to fix. It happens on every project like this, but it definitely ate up more engineering time than we budgeted for.

Optimization Steps Taken

We used the data from our fractional model to make a few key changes on the fly:

1. Budget Reallocation: The model showed that broad programmatic display wasn’t pulling its weight, so we moved 15% of that budget into more targeted social media video ads and native content for the consideration phase, which we knew was having a strong influence.
2. Refined Audience Segmentation: We stopped using pure interest-based targeting for our awareness campaigns and tightened them up, focusing on lookalike audiences built from existing customers and engaged site visitors. The quality of our initial traffic improved immediately.
3. A/B Testing for AI-Generated Content: We started running more A/B tests on the AI’s headlines and visuals, especially for the top and middle of the funnel. For instance, we tested emotional appeals against functional benefits in our short video ads and found that emotion got a 15% higher initial engagement rate on TikTok, a finding that helped us lower our CPL for those stages by 12% over the campaign.
4. Enhanced Content Production Pipeline: We had to get more content into the system, so we brought on some freelance video editors and content strategists to crank out more demo videos and interactive guides for the AI to work with.
5. Attribution Model Fine-Tuning: We were constantly tweaking the weights in our fractional model. We saw that visits from review sites were having a big impact, so we bumped up the credit for those interactions by 5% in the consideration phase, just one small adjustment to make the model better reflect what was actually happening.

The real power here was combining fractional attribution with dynamic AI marketing which changed our whole understanding of what was working. It got us past just counting clicks and into valuing the entire journey. Marketers are often blind without this kind of attribution. This full-picture approach to AEO marketing is what let us get ahead of market shifts and hit a CPL under $15.

FAQ Section