INTENTKEY PLATFORM
Inside an Audience Model
The IntentKey Platform delivers AI-powered audience models that go far beyond surface-level demographics or outdated data sources. Each model provides a real-time, concept-based view into what drives consumer intent, without relying on cookies or personal identifiers.
Building an Audience Model
The IntentKey Platform is designed to make it easy to build instant AI-powered audience models that are more precise than stagnant third-party data, and that don’t rely on cookies or any personal identifiers. Let’s dive in and break down an audience model for car shoppers looking for an SUV for their family.
Describe your intended audience. Clear, focused inputs lead to better model results.
Add a relevant URL to help the model better understand your audience’s interests and context.
This was our prompt:
Shoppers in the automotive space, searching for an SUV that fits their needs for family and travel. They are focused on reliability and safety.
This was our example URL: https://www.caranddriver.com/rankings/best-suvs/family
Seed Concept Summary
Review the key concepts the model identified from your input. These guide audience targeting.
Manually add concepts to further refine and customize your audience model.
Remove concepts that aren’t relevant to your intended audience.
After editing concepts, regenerate model based on your current input.
The Seed Concept section represents the foundation of your audience model. These concepts are pulled directly from your initial input, whether that’s a written prompt or a relevant URL. They reflect the themes most aligned with the audience you’re trying to reach.
For example, a model built for “family SUV shoppers” might surface seed concepts like SUV, safety, fuel efficiency, stroller, and reliability.
You can remove irrelevant concepts by clicking the minus (-) icon next to any item. This moves the concept to a section labeled Concepts Selected for Removal. You can also add custom concepts using the Add Concepts button. After any edits, click Refresh to update the model with your changes.
Key takeaway: Seed concepts give IntentKey AI the context it needs. Ensure they’re relevant and focused to produce the most accurate audience model.
Focus Model Training
Select a category to focus model training and narrow your audience to a more specific category.
This section allows you to focus audience predictions within your model. By selecting one area of focus, you help the model fine-tune its targeting to the ideal audience you want to reach.
One focus category can be applied at a time. Once selected, click Refresh to see updated insights throughout the model.
Key takeaway: Use this feature to gain insights into a specific demographic within your target audience. These insights can help inform your targeting and messaging strategy across multiple platforms.
Concepts
Core concepts are highly relevant to your input. Trending concepts reflect current interest. Core + Trending are both highly relevant and currently gaining traction.
Score indicates the strength of each concept’s relevance to your input. Higher scores suggest closer alignment with your intended audience.
Click a concept to visualize how concepts are related to each other. Connections indicate higher contextual overlap and audience alignment.
This section shows the building blocks of your audience model. These are AI-identified concepts that represent the motivations, interests, and intent behind the content your audience engages with.
Concepts are categorized into three types:
- Core Concepts are stable and central to the audience’s behavior.
- Trending Concepts are rising in popularity but may fluctuate.
- Trending Core Concepts are both influential and currently gaining momentum.
Each concept includes a score on a 100-point scale. This score combines its importance and popularity within the audience model.
Key takeaway: Concepts provide transparency into what your audience cares about. Don’t dismiss seemingly unrelated concepts. They may point to secondary interests that expand your targeting opportunities.
Groups
Represents the share of the audience associated with each thematic concept group. Higher percentages indicate stronger influence of that group in the model.
View the main themes identified in your audience. Expand each group to see the individual concepts that define it.
Groups are collections of related concepts clustered under shared themes. For instance, this model built for SUV shoppers includes groups like Road Safety Essentials and Child Restrain Systems.
These groupings offer a high-level view of the types of content and topics that consistently appear in your audience’s journey.
Key takeaway: Groups help marketers understand how different ideas are connected within the audience’s mindset. Use them to guide creative messaging or identify additional targeting angles.
Audience & Sentiment
Shows the percentage of your audience actively engaging with the core concepts of the model.
Estimated daily online audience interacting with the model’s concepts.
Represents the overall emotional tone of content associated with your audience’s concepts. Ranges from -100 to +100. Positive values reflect predicted positive sentiment from your audience; negative values reflect negative sentiment.
These real-time metrics show your model’s potential size and sentiment. The data updates automatically as the model is created or refreshed. It reflects daily average size from the last 30 days.
- Max Audience Size estimates the number of people actively engaging with the concepts in your model.
- Core Audience shows the percentage of users engaging with the core concepts.
- Sentiment Score ranges from -100 to +100. Positive values indicate overall favorable engagement with the model’s content themes.
- Audience and Sentiment Trends visualize how interest levels change over time, especially in response to new content, trends, or events.
Key takeaway: This section gives you a dynamic snapshot of your model’s reach and relevancy. Use it to determine timing, sentiment shifts, and campaign potential.
Demographic Data
Demographics are inferred from audience’s concepts based on content consumption. These estimates reflect traits—like age and gender—of audiences engaging with similar topics.
This section shows how different demographic groups align with your audience model, using an index where 100 represents the national average.
- Values above 100 indicate higher-than-average relevance.
- Values below 100 suggest underrepresentation.
For example, a model focused on family SUVs may show elevated engagement from users aged 35–44 or households with children, but it also reveals high engagement in ages 65+, revealing that the demographic is highly engaged and serves as a potential audience to focus on.
Key takeaway: Demographics don’t define your model, but they offer valuable context. They can guide creative and messaging strategies while reinforcing that intent signals transcend traditional audience segmentation.
States & Zip Codes
Shows where in the U.S. your audience is most engaged with the model’s concepts. Higher indexing (dark green) indicates stronger relative interest in those regions.
Lower indexing (dark red) indicates less interest in those regions.
Zip code insights are derived from analyzing content engagement and concept signals across the open web, enabling geographic-level audience understanding without using personal data.
This geographic breakdown reveals where your model is likely to perform best. Regions are ranked based on the density of online content consumption that matches your model’s concepts.
A family SUV model may show stronger alignment in suburban and Southern regions, while urban areas with heavy public transit usage may under-index.
Key takeaway: Geo data helps with regional targeting, budget allocation, and understanding where your audience lives and browses.
Bringing It All Together
The IntentKey Platform gives advertisers a clear, privacy-safe view into real audience intent. Every section of the model is built to help you uncover what motivates consumers, where they’re most engaged, and enable you to reach them in the moments that matter.
Powered by our proprietary AI, IntentKey is trained on over 110 billion pages of content, with 1 million new URLs added daily. Our models refresh every five minutes to stay on top of trending topics and delivers the most precise intent signals, mapping more than 25 million concepts across the open web — all without relying on cookies or personal identifiers.
Ready to see it in action?
Start building your own custom model at platform.inuvo.com, contact us to ask questions or book a personalized demo.
