Editorial Team Getting in front of customers in 2026 isn’t the hard part. Making them care is. To build a real emotional connection, you need what we’re calling AI brand affinity. Here’s a breakdown of how an up-and-coming sustainable home goods brand, “Bloom & Branch,” pulled off a 22% jump in customer loyalty with an AI-powered campaign. They figured out how to stop being just another transaction and started building a real community.
Key Takeaways
- They spent $75,000 over 10 weeks and got a 3.5x ROAS, mostly from personalized video.
- The whole strategy was built on using AI to analyze customer service chats to figure out what to talk about in our ads.
- We A/B tested AI voiceovers on user-generated content and they beat professional voice actors, lifting CTR by 1.8%.
- Broad targeting was a money pit, hitting a $12.50 CPL. Lookalike audiences did way better at $4.10 CPL.
- We quickly moved 30% of the budget from weak static ads into interactive quizzes, which gave us a 15% bump in conversion rates.
The Challenge: Cultivating Connection in a Crowded Market
Bloom & Branch popped up in late 2024 selling ethically sourced home decor. Their early marketing was all about product features and sustainability, which got them some attention but didn’t make anyone fall in love. We saw it in the data: high bounce rates on product pages and almost no repeat buyers. The brand’s leadership knew that in a market drowning in “eco-friendly” claims, the only way to stand out was to build an actual bond with people. For the Q2 2026 campaign, the objective was clear: raise customer lifetime value (CLTV) by creating a genuine emotional connection, tracked through repeat purchase rates and social media chatter.
Campaign Strategy: AI-Driven Empathy at Scale
The entire strategy for Bloom & Branch was about using artificial intelligence for both efficiency and empathy. We wanted to understand our customers’ unspoken needs and values by analyzing their behavior and then reflecting those values back at them with personalized content. This went far beyond just recommending products. We were aligning the brand’s story with a customer’s personal goal of having a sustainable and comfortable home. We built the campaign on three main ideas:
- Finding Emotional Hot Buttons: We used AI to analyze customer service transcripts and social media comments to find out what people were actually feeling.
- Creating Personalized Micro-Experiences: We served up custom content sequences to users based on what they looked at on the site and how they’d interacted with us before.
- Building a Community with AI: We used AI to spot potential brand fans and give them a little nudge to get more involved.
The campaign itself ran for 10 weeks straight, from April 1 to June 9, 2026. We had a total budget of $75,000 to work with, which we split across creative, media spend on Meta (Meta Business Suite) and Google Ads (Google Ads), and a smaller experimental budget for TikTok.
Creative Approach: Storytelling Through Data
The insights from the AI completely changed how the creative team approached the content. Instead of just product shots, they built lifestyle stories around the emotional themes the AI uncovered. For example, the sentiment analysis showed that the audience had a strong craving for “sanctuary” and “mindfulness” in their living spaces, a desire that went much deeper than just “sustainability.”
- Video Series: “The Intentional Home” (60% of creative budget): We created a series of short videos (15-45 seconds) featuring real customers (who gave us permission, of course) showing how Bloom & Branch products made them feel a sense of personal calm. We then A/B tested AI-generated voiceovers against professional voice actors.
- Interactive Quizzes: “Find Your Home Sanctuary Style” (25% of creative budget): We put these quizzes on the website, where a conversational AI asked users about their lifestyle and taste, then gave them personalized product suggestions and styling tips.
- Personalized Email Journeys: After someone took a quiz, they were dropped into an email sequence that sent them content based on their “sanctuary style.” This included things like DIY projects, designer interviews, and stories from other customers who liked the same things.
One of the biggest surprises came from our early A/B tests. We found that AI-generated voiceovers, specifically the ones we picked for having a natural and slightly imperfect sound, actually beat the expensive studio voice actors, boosting CTR on our video ads by 1.8%. It seems that little bit of imperfection felt more authentic and connected better with the audience. This lines up with what the IAB has been saying in their reports (like the 2023 Podcast Advertising Revenue Study) about people being more open to AI voices as long as they don’t sound completely robotic.
Targeting and Audience Segmentation
We used a layered targeting strategy that we constantly changed based on what the performance data was telling us week to week.
- Initial Broad Targeting (Weeks 1-2): Like most campaigns, we started out broad on Meta and Google, targeting interests like “sustainable living,” “home decor,” and “mindfulness.” This was really just to get baseline data and see who would bite.
- Lookalike Audiences (Weeks 3-10): We took our initial website visitors and buyers and built 1% and 2% lookalike audiences from them. This turned out to be our goldmine.
- Retargeting (Weeks 1-10): Anyone who started a quiz or looked at a product but didn’t buy got hit with retargeting ads showing specific product features and customer testimonials.
- Custom Audiences (Weeks 4-10): We also uploaded our existing customer and email lists to create custom audiences, which we used for both exclusion and specific re-engagement campaigns.
The big lesson here is pretty obvious in hindsight: broad targeting is a fast way to burn cash. Our initial broad segment had a cost per lead (CPL) of $12.50, which was just painful compared to the consistent $4.10 CPL from our lookalike audiences. Seeing that massive difference made reallocating the budget a very easy decision.
Campaign Performance: Metrics and Insights
The campaign worked, and it really showed how AI can help build actual brand affinity. Here’s a quick look at the key numbers:
Overall Campaign Performance (10 Weeks)
- Total Budget: $75,000
- Total Impressions: 8,200,000
- Overall CTR: 1.5%
- Total Conversions (Purchases): 2,850
- Cost Per Conversion: $26.32
- Return on Ad Spend (ROAS): 3.5x
Segment Performance Comparison
| Segment | Impressions | CTR | CPL | Conversions | ROAS |
|---|---|---|---|---|---|
| Broad Interest | 2,500,000 | 0.9% | $12.50 | 400 | 1.8x |
| Lookalike Audiences | 4,000,000 | 2.1% | $4.10 | 1,800 | 4.8x |
| Retargeting | 1,200,000 | 3.5% | N/A (engaged users) | 500 | 6.2x |
| Custom Audiences | 500,000 | 1.8% | $6.70 | 150 | 3.1x |
The numbers speak for themselves: lookalike and retargeting audiences were our big winners. The incredibly low CPL and high ROAS from the lookalike groups proved how good the platform’s AI was at finding new people who acted just like our best existing customers.
What Worked and What Didn’t
What Worked Well:
- AI Sentiment Analysis: This was everything. The AI’s discovery that customers wanted a “sanctuary”, and not just more “eco-friendly” talk, changed our entire messaging approach. That single insight led directly to the “Intentional Home” video series, our best-performing creative by a long shot.
- Personalized Quiz Funnels: The interactive quizzes were a huge hit, with a 45% completion rate and a solid 15% conversion rate from quiz completion to a first purchase. This approach segmented users perfectly and let us follow up with super-relevant emails.
- User-Generated Content (UGC) with AI Voiceovers: Combining real customers’ photos and videos with a polished AI voiceover was a killer combination of authenticity and quality. This ad type had a 30% lower cost-per-view than our professional studio content.
- Iterative Optimization: We were in the accounts daily, making tweaks weekly. After week 3, for instance, we moved 30% of the budget out of static image ads and pushed it into the quiz funnels, which instantly boosted our conversion rates.
What Didn’t Work as Expected:
- Broad Demographic Targeting: As I mentioned, the initial broad targeting was inefficient. We needed it for a week or two to gather data, but the CPL was far too high to justify keeping it running any longer than that.
- Slick Studio Content: Our first batch of professionally shot product videos looked great but felt totally sterile to the audience next to the real-life UGC. It’s a classic mistake. Sometimes, trying too hard to be perfect backfires because it lacks authenticity.
- Generic Email Nurturing: The early email flows that weren’t tied to quiz results or user behavior had terrible open rates (18% vs. 45% for the personalized ones). It was a stark reminder that if it’s not personal, it’s ignored.
Optimization Steps Taken
We made several big changes on the fly based on what we were seeing in the data:
- Budget Reallocation: We pulled about 40% of our total budget from the weak broad targeting and static ads and funneled it directly into our top-performing lookalike and retargeting campaigns.
- Creative Refresh: We went all-in on the UGC-style videos and interactive content, and pretty much stopped making traditional product ads. The new videos we made leaned even more into customer testimonials and emotional benefits.
- AI Model Refinement: We kept feeding the sentiment analysis AI new data from customer service chats and social comments. This helped it get even better at spotting subtle emotional cues and language, making our content ideas more precise over time.
- Website Personalization: We used a third-party AI tool (Optimizely) to set up a dynamic content system on the homepage. It would show different banners and products to users based on their quiz answers or what they’d been looking at, making the whole site feel more personal.
These changes made a real difference. During the second half of the campaign (weeks 6-10), our CPL dropped another 18%, and the overall ROAS climbed from an early 3.2x to a final 3.8x. It just goes to show that you can’t set and forget a campaign like this. Constant, data-backed tweaking is just as important as having a good plan to start with.
Conclusion
The Bloom & Branch campaign shows that AI is about more than just automation. They used it as a tool for building real customer loyalty by digging into emotional triggers. The AI helped them listen to what people actually wanted, and in doing so, they turned one-off buyers into a connected community. For any brand that wants to compete in 2026, this is the playbook: use AI to get past the surface-level data and actually connect with your customers.
So how does an AI actually figure out customer emotions?
It uses something called natural language processing (NLP) to read through tons of text, think customer reviews, social media comments, and support ticket transcripts. The AI is trained to spot patterns, recurring words, sentiment (is this person happy, angry, or just neutral?), and specific emotional phrases. This gives you a data-backed look at what people are actually feeling about your brand.
What specific AI tools were used in this campaign?
We used a few different AI-powered tools. There was a custom sentiment analysis engine that plugged into their customer service software, a conversational AI platform for the quizzes, AI voice synthesis software for the videos, and of course the machine learning built into the Meta and Google ad platforms for building lookalike audiences.
Can a small brand do this without a huge budget?
Absolutely. You don’t need a $75,000 budget to get started. Small brands can begin by using the AI features that are already baked into platforms like Google Ads or the Meta Business Suite for audience targeting. There are also affordable AI tools for basic sentiment analysis or for getting starter ideas for ad creative. The key is to focus on one or two things at first instead of trying to do everything at once.
What’s the point of user-generated content (UGC) if you’re just using AI anyway?
UGC provides the authentic, human stories that AI can then scale and personalize. The AI’s job is to find the best pieces of UGC, figure out what emotion it conveys, and then get it in front of the right person at the right time. For example, the AI can match a specific customer testimonial to a user segment that’s struggling with the same problem, making the connection much stronger through that peer validation.
How do you measure brand affinity besides just ROAS?
ROAS is important, but it doesn’t tell the whole story. We also tracked metrics like customer lifetime value (CLTV), repeat purchase rate, and Net Promoter Score (NPS). On top of that, we watched for an increase in positive brand mentions on social media, a longer average session duration on the website, and higher engagement rates on our personalized emails. All those other metrics are what really tell you if you’re building a deeper connection and loyalty.
