The Problem
Every day, we experience around 20,000 moments — and we're bombarded with up to 10,000 ads fighting for a piece of that attention.
The result? 41% of people remember less than 10% of the ads they see.
Your ad gets one chance to break through.
The Project
So I gave myself a project.
Research Foundation
Using the same LLMs we all use today, I identified and analyzed over 1,000 marketing effectiveness studies from the past decade across 14 industries — pulling insights from Kantar, IPA DataBank, Nielsen, System1, and many more.
Benchmark Extraction
The LLMs extracted hundreds of thousands of quantified benchmarks: what makes an ad memorable, what builds trust and persuasion, what actually drives action, sales growth, and loyalty.
Award-Winning Campaigns
But even with all that research, there were gaps — certain industries and metrics where the academic data ran thin. So I tapped the LLMs again, this time to gather and analyze thousands of award-winning campaigns from the Effies, Cannes Lions, and IPA Effectiveness Awards. I had them break out the work by industry and metric, then identify patterns that appeared consistently in campaigns that actually moved the needle.
Signal Distillation
From all of this, I worked with an LLM to distill the findings into 80 discrete creative signals — the specific elements that predict ad performance. Then we calculated industry-specific weights for each one, because what works in fast food doesn't work in luxury, and what drives awareness doesn't always drive conversion.
Neural Network Training
Finally, I fine-tuned pre-trained neural networks — NIMA for aesthetic quality, ResMem for memorability, EmoSet for emotional tone — on a labeled dataset of thousands of ad images. Through multiple training iterations, the models learned to detect the 80 signals, with weights adjusting at each pass. I validated the system on held-out data to ensure it generalizes across industries and ad formats.
The Scoring Engine
The detection layer feeds into a scoring engine built on the research-backed weights — applying industry-specific logic to predict how each ad will perform on awareness, consideration, and conversion.
The result is
How It Works
A tissue session for your creative — without the conference room.
Upload any static creative and TissueSesh gets to work. It automatically identifies your brand and the industry you operate in, then evaluates your creative against benchmarks trained on what actually drives results in your category — because what works in fast food doesn't work in luxury.
You get scores across three outcomes:
Awareness
Will people remember this?
Consideration
Will people trust and connect with this?
Conversion
Will people take action?
How scoring works:
Each outcome is scored 0–100.
Your creative is set up to deliver on this outcome.
There's a creative or strategic element that could strengthen this outcome.
This asset isn't currently optimized to deliver here.
If something's missing, you'll know exactly what — and why it matters for your industry.
Recommendations aren't generic. They're grounded in what's been proven to work in your category, backed by the research that built the model.
Industries
14 Industries.
Different Rules.
The same signal can help one industry and hurt another.
TissueSesh knows the difference.
Industry-specific weights ensure recommendations match what actually works in your category.
See example: What drives results by industry
| Awareness | Consideration | Conversion | |
|---|---|---|---|
| Luxury | Heritage imagery, aspirational lifestyle, minimal text | Craftsmanship cues, celebrity endorsement, premium visuals | No price visibility, subtle CTAs, exclusivity signals |
| QSR | Bold colors, food photography, humor | Behind-the-scenes authenticity, diverse representation | Price prominence, limited-time offers, clear value |
| Finance | Distinctive visuals (breaks category norms), emotional storytelling | Trust badges, real testimonials, security signals | Single clear CTA, benefit-focused copy |
| Tech B2B | Minimalist design, brand consistency | Emotional messaging (yes, even B2B), customer logos | Demo/trial CTAs, specific metrics, single CTA |