Attribute definitions · 3 active · demo data
| Attribute | Type | Asset scope | Coverage | Status |
|---|---|---|---|---|
| hook_type | Enum | Video | 98% | Complete |
| text_density | Number | Image · Video | 94% | Processing |
| voiceover_present | Boolean | Video | 100% | Complete |
AI creative element intelligence
Attribute Analysis
Watch Lemon read the creative itself: who appears, what happens, how the product is framed, and how the message lands. Then see which choices are tied to better ROAS, CPI, CTR, or conversion.
No card · Up to 2 apps · Includes 3 generation credits + 1.2 attribute credits.
SOURCE CREATIVE1080 × 1920 · VIDEO
MECHANIC · CASCADECHARACTER · MASCOTFRAMING · BOARD DOMINANT12 attributes assigned986 creatives compared
METRICROAS · D7
CREATIVE COHORTGame mechanic
OUTCOME · ROAS D7How it works
Teach Lemon what to look for. It handles the rest.
Choose what you want to learn
Ask about the hook, character, setting, offer, visual style, or any creative choice your team cares about.
Let Lemon scan every creative
Lemon reads matching images, videos, and ad copy across your library. You do not have to label them by hand.
See what moves performance
Compare each creative choice against ROAS, conversion value, CPI, CTR, or another metric you use to make decisions.
Your creative language
Analyze the ideas that matter to your brand
Track the hooks, settings, characters, offers, and visual styles your team already talks about. Add a new idea once and Lemon finds it across the library.
From pattern to creative
Click any result to see the ads behind it
Go straight from a winning pattern to the creatives that use it. See the examples, the metric, and the campaign context without rebuilding your filters. Every result carries the number of creatives behind it and the range the sample supports, because the sample size in an attribute comparison is the number of creatives, not the number of impressions. Before a tag enters that comparison, the creative tagging integrity protocol checks population, failed fields, coverage, and label version.
| Selected value | ROAS · D7 | Assets | Compared with |
|---|---|---|---|
| UGC hook, first 3s | 148% | 214 | All other values |
| Creative · demo data | Network | ROAS D7 |
|---|---|---|
| ugc_hook_reveal_v3.mp4 | AppLovin | 152% |
| gameplay_boss_30s.mp4 | AppLovin | 95% |
| static_carousel_04.png | AppLovin | 143% · estimated |
| playable_tutorial_v2 | Unity | 114% |
Turn every winning pattern into your next brief.
Send the best-performing hook, format, or visual style straight into Creative Lab. Make new variants from the ads that already work, then choose what goes live.
From association to a test
Check the finding before you write the brief
An attribute can be associated with higher ROAS because it appeared in different campaigns, countries, or delivery conditions. The next creative test should challenge that explanation.
Check the labels
Open examples with and without the selected attribute. Confirm that the definition captures the creative choice you meant, and inspect missing or failed labels with the tagging QA protocol.
Match the comparison
Keep app, network, country, dates, and metric window consistent. Count distinct creatives and inspect the examples behind the difference. More impressions on one creative do not create more independent creative examples.
Change one creative choice
Use the finding as a hypothesis. Write down what changes, what stays fixed, and which result would contradict it. The creative test brief gives the team a reusable method.
Competitor Insights
Compare your creative choices with the market
Use competitor creative research to see which patterns recur in competing apps’ ads and how they compare with your analyzed library. Bring a new hypothesis back to your own account, where Attribute Analysis can relate your creative choices to measured performance.
FAQ
Questions before you get started
What is creative attribute analysis?
Creative attribute analysis measures how recurring elements inside ads relate to performance. Instead of comparing only whole creatives, Lemon lets teams examine attributes such as hook type, format, subject, pacing, voiceover, or text density against the metric and attribution window they select.
Do I have to label every creative manually?
No. You define the attribute and Lemon applies it across matching assets with AI. Manual attributes are also supported when a team needs a controlled taxonomy or wants to correct a value.
Which asset and attribute types are supported?
Attributes can be applied to image, video, and text assets. A definition can produce an enum, number, or boolean value, which supports both categorical questions and measurable properties.
Can I inspect the creatives behind an attribute result?
Yes. Every attribute value can be opened to inspect example assets under the same app, network, campaign, country, platform, date, and performance filters used by the analysis.
How does Attribute Analysis connect to Creative Generation?
The analysis identifies the elements associated with stronger results. Those findings can then become a generation brief or reference set in Creative Lab, keeping the creative iteration grounded in observed performance.
Bring your hardest growth question
See Attribute Analysis work on your data
Bring one app and the decision in front of you. We’ll show you what Lemon finds.