Editorial Team Key Takeaways
- Our Q3 2025 Gourmet Grub campaign hit a 3.2x ROAS after we used AI to spot short-form videos with viral potential before we shot them.
- We saved 18% on creative production costs by using an AI sentiment and trend tool to pre-screen 50 video concepts, killing the duds early.
- The campaign’s best ad, which AI flagged for high emotional resonance, pulled a 7.8% click-through rate on TikTok.
- Within the first 72 hours, the AI-approved content got 22% more organic shares than the videos we concepted the old-fashioned way.
- We put 70% of our ad budget on the AI’s top picks, which led to a 45% lower cost per conversion compared to our control group.
Getting content to be shared organically is the goal we’re all chasing, but figuring out what makes something go viral content is mostly guesswork. So in Q3 2025, our team decided to stop guessing. We worked with “Gourmet Grub,” an up-and-coming meal kit service, and built AI prediction into our entire content workflow. We had to know: could an algorithm actually help us get people to share our stuff?
Campaign Teardown: Gourmet Grub’s AI-Powered Viral Push
For Gourmet Grub, the objective was straightforward: we needed to get new subscribers and build the brand using shareable short-form videos for a Gen Z and young millennial audience. The meal kit space is incredibly crowded, and our usual performance marketing playbook just wasn’t delivering the kind of engagement that breaks through the noise. Our hypothesis was that content built from the ground up for sharing, not just for cheap clicks, would finally get them noticed.
Strategy: AI-Driven Content Ideation and Validation
We used AI to guide our creative process from the very first brainstorm all the way to distribution. Our plan was to:
- Spot emerging trends: We had the AI constantly scan TikTok and Instagram Reels to find the micro-trends, audio clips, and visual styles that were just starting to gain speed.
- Predict emotional punch: Before we shot anything, we analyzed our scripts and storyboards for the specific elements that data shows correlate with high emotional engagement and the intent to share.
- Optimize for the algorithm: We wanted to know which content details, like pacing, text placement, or music, were getting preferential treatment from each platform’s recommendation engine.
This wasn’t to replace our creative team. It was to give them a data-backed compass so they weren’t just guessing about what might go viral. We worked with a platform called “TrendSight AI” (trendsight.ai), which specializes in predictive analytics for social media content.
Creative Approach: Authenticity Meets Data
Our creative team churned out over 50 different concepts for short-form videos, everything from quick recipe hacks with Gourmet Grub ingredients to funny unboxing videos and “day in the life” content showing busy people using the kits. We fed every single concept, script, visual plan, and proposed audio, into the TrendSight AI platform. The AI then spit out a “shareability score” calculated from a few key inputs:
- Sentiment Analysis: How well would the video trigger strong, positive feelings like joy, surprise, or inspiration?
- Trend Alignment: Did the concept use sounds, hashtags, or visual styles that were currently on an upward trajectory?
- Engagement Predictors: The AI analyzed our scripts and visual plans for specific patterns in language and imagery known to get more comments and saves.
The AI didn’t just give us a number. It gave us specific, actionable feedback. For one concept, it pointed out that our chosen song was popular but its engagement was already declining, so it suggested a different track that was on the rise. For another script, it flagged a line as having low emotional impact and offered a more powerful alternative. This back-and-forth feedback let our team sharpen the ideas before we spent a dime on production, which saved a ton of money. We in the end greenlit 15 videos, focusing on the ones with the best AI scores and the most useful feedback.
Targeting and Distribution: Smart Amplification
We went after 22-38 year olds who were into cooking, healthy living, and convenience. Our paid campaigns ran on TikTok Ads (ads.tiktok.com) and Instagram Reels Ads (business.instagram.com/advertising), since that’s where short-form video lives. For targeting, we layered interest-based segments (“meal prep,” “food delivery”) with lookalike audiences we built from Gourmet Grub’s existing customer list. The big move was allocating a full 70% of our ad budget to promote the top 5 videos the AI predicted would be winners. We used the other 30% to support the other videos and to run A/B tests on things like call-to-actions and thumbnails.
Campaign Metrics and Performance
The campaign ran for 8 weeks, from August 1st to September 26th, 2025.
Overall Campaign Performance
- Total Budget: $65,000
- Duration: 8 Weeks
- Total Impressions: 18.5 Million
- Total Clicks: 398,000
- Overall CTR: 2.15%
- Total Conversions (First Subscriptions): 2,100
- Cost Per Conversion (CPL): $30.95
- Return on Ad Spend (ROAS): 3.2x
The top-line numbers looked good, but the real story was in the data when we compared the AI-vetted content to our control group.
Performance Comparison: AI-Validated vs. Control Content
| Metric | AI-Validated Content (Top 5) | Control Content (Traditional Concepting) |
|---|---|---|
| Average CTR | 4.1% | 1.8% |
| Average Cost Per Conversion | $22.80 | $41.50 |
| Average Organic Shares (First 72 Hours) | 1,850 per video | 920 per video |
| Average Video View Duration | 12.5 seconds (for 15-sec videos) | 8.1 seconds (for 15-sec videos) |
| ROAS Contribution | 4.7x | 1.9x |
The difference was huge. Our best-performing video, a funny “expectation vs. reality” skit about home cooking that featured the Gourmet Grub box, got a 7.8% CTR on TikTok and was shared over 4,000 times organically in its first week. TrendSight AI had specifically flagged that concept for its high potential to be both funny and relatable, which are gold for getting shares.
What Worked: Precision and Iteration
The single biggest reason this worked was that we brought AI into the creative process right at the start. By screening 50 ideas before production, we avoided wasting money on concepts that were dead on arrival. This was a complete change from how we used to work, which often involved making a bunch of stuff and just seeing what stuck. A recent IAB report on AI in advertising (iab.com/insights/iab-ai-in-advertising-report-2025) notes that companies using AI for creative are seeing ROI bumps of 15-25%. Our 3.2x ROAS, which was almost entirely carried by the AI-validated ads, fits right in there. The other big win was the AI’s knack for finding tiny, specific things to fix. For instance, it told us to cut a video’s intro from 3 seconds to 1.5 seconds, showing us data that longer intros on TikTok were causing people to scroll away. That tiny edit boosted that ad’s average view duration by 15%. Good luck getting a human to spot that kind of granular insight consistently across dozens of video ideas.
What Didn’t Work as Expected: Over-Reliance on Pure Trend-Following
The AI was powerful, but we learned that you can’t just blindly follow every single trending sound or format. We had one video that the AI scored very high because it perfectly matched a fast-growing audio trend, but its performance was just okay. Looking back, the problem was that the trend itself had nothing to do with Gourmet Grub’s brand or what the product actually does. It just felt tacked on. It was a good lesson: the AI gives you great signals, but a human strategist still needs to be the one to ensure it all feels authentic to the brand. Just because something’s trending doesn’t mean it’s right for you.
Optimization Steps Taken
We made a few key changes as the campaign was running:
- Dynamic Budget Shifting: We watched the numbers daily and moved about 15% of the weekly budget away from the weaker ads and into the top 3 AI-picked videos that were crushing it on conversions and shares.
- A/B Testing CTAs: We tested “Get Your First Box 50% Off” against “Simplify Dinner Tonight.” The benefit-focused CTA (“Simplify Dinner”) actually got a 12% higher conversion rate with our audience, especially on Instagram.
- Refining AI Input: We started feeding our own notes back into TrendSight AI, telling it which videos felt more on-brand or authentic. This helps train the model to understand our specific needs, not just general market trends.
- Geo-Targeted Content: In the last two weeks, we tried tweaking our best ad with hyper-local references (e.g., “Dinner in minutes, even for busy professionals near Midtown Atlanta”). The test was small, but it showed an early 8% lift in CTR in those areas.
The Gourmet Grub project showed us that AI for predicting viral hits isn’t some far-off idea anymore. It’s a real tool that, when you use it smartly, gets you more organic shares and drives results you can actually measure. The future of content marketing is definitely going to be a tighter partnership between human creativity and artificial intelligence.
How does AI actually predict viral content?
It works by sifting through massive amounts of data from past social media content. The AI looks for patterns in what got shared, liked, and commented on, analyzing everything from the audio and visual style to the story structure. It then scores new ideas based on how well they match those winning patterns and fit with current micro-trends.
What kind of data does the AI look at?
These AI models process a lot of different inputs. They analyze text from scripts and captions, audio features like music BPM and tone of voice, and visual data like color schemes, pacing, and even object recognition. They also ingest all the key engagement stats, likes, shares, comments, saves, view time, and track which hashtags are gaining traction.
So, does this mean AI is replacing creative teams?
No. Think of AI as an incredibly powerful tool that optimizes human creativity, not a replacement for it. The AI is brilliant at finding patterns in data and making predictions, but it has no real intuition, brand understanding, or the spark to create a completely new idea that breaks all the old patterns. The best results happen when you combine the AI’s data with a human’s strategic and creative instincts.
What are the downsides of using AI for this?
The main limitations are that some AI models can be a “black box,” so you don’t always know *why* it made a certain prediction. AI is also better at following existing trends than creating new ones. If you rely on it too much without your own creative input, your content can start to look generic and same-y. It also might not always pick up on subtle brand voice or cultural nuances.
How can a small business use this without a huge budget?
You can start with free or cheap AI tools that are already part of social media platforms, like their built-in trend dashboards. AI writing assistants can also help with generating ideas or optimizing headlines. Even if you don’t have a pricey predictive platform, just applying the core principles, focusing on strong emotional hooks, clear stories, and aligning with trends, will make your creative process more data-driven and effective.
