Introduction to Creating Hyper-Targeted Custom Models – Audiencelabhelp

IMPORTANT: If you want to just get started with AI Model builder, then you can skip head here

💰 The Power (and Value) of Custom Models

Custom models are one of the most powerful and monetizable tools inside AudienceLab. They’re not just about better targeting — they’re about building high-value data assets that can be sold, reused, and activated across any channel.

Unlike static segments or pre-built lists, custom models let you define exactly who you want to reach based on behavior — not demographics, not job titles, and not guesswork.

When used correctly, they become incredibly valuable:

🔍 Target real-time intent — people actively researching, comparing, or ready to buy

⚡ Eliminate noise — outperform broad audiences in outbound, email, and paid media

💼 Sell them as premium data — agencies and clients will pay for niche, high-performing audiences

🔁 Activate on any platform — Facebook, LinkedIn, Google, CRMs, or direct export

AudienceLab scans over 200 billion URLs per week. That’s a lot of behavioral signals — but only if you feed the system with a clean, specific description. This is where most models fall short, and where you gain the advantage.

🔑 The 3 Core Elements of a Great Custom Model

1. 🎯 Choose One Specific Topic

Start by defining one clear behavior someone is doing — not who they are.

✅ Good Examples:

“Zero-down HVAC financing”

“Cold email inbox rotation”

“In-person divorce consultations”

“VA refinance loan options”

🚫 Avoid:

“Doctors looking for CRM tools” ← ICP included (filter later)

“Marketing tools like HubSpot and ActiveCampaign” ← brand stacking

“SaaS founders who need lead generation” ← too broad and persona-based

✅ Tip: AudienceLab is behavior-first. If you describe what someone is doing, the system will find them. You can always filter who they are later.

2. 🔍 Apply a Single Lens

A lens frames how the system interprets the topic. Use only one per model.

Lens	Use When...
Brand	You want to target behavior around one brand
Product	You’re focused on a specific feature or offer
Function	You want to model what the solution does
Service	You're modeling a consultation or delivery format
Solution	You’re modeling the problem being solved
Event	The behavior is tied to a trigger or trend

3. 🎚️ Set the Right Granularity

Granularity = how deep or narrow the behavior should be.

Level	Example Behavior
1	“CRM software” (broad category)
3	“Cold email inbox warm-up tools”
5	“Automated sender rotation for domain health”

💡 Think about what someone is doing right before they make a buying decision. That’s your Level 5.

❌ Don’t Include Your ICP

Avoid writing descriptions like:

“SaaS companies looking for lead tools”

“Medical practices interested in hydrogen therapy”

Why? Because intent and ICP are separate.

Custom models are built on what people are doing, not who they are. After your model is live, you can apply filters by:

Industry

Company size

Domain list (e.g., Dream 100)

Location

Job title

Let the model do its job. Keep the description clean.

✅ Ready to Build?

Now that you understand the structure, you're ready to use the AI bot — trained on our system — to generate your paragraph for you.