Segments
A contextual segment is a curated list of URLs grouped around a topic, interest, or intent signal. Classify builds segments by analyzing seed inputs — a text prompt, example URLs, or both — and finding the broader set of web pages that match the same context at scale.
How segments work
You describe what you're looking for. Classify finds it.
- You provide a seed — a natural language prompt, a set of example URLs, or a brand story describing your target audience
- Classify builds the list — our models analyze your seed and identify URLs across the web that match the same contextual signal
- You receive a URL list — pushed into our data supply chain and exposed as a Deal ID you can target directly in your DSP
Segment creation is asynchronous and typically takes 1–2 hours.
What makes a good segment
The more specific your seed, the more precise the segment. You can combine a prompt with example URLs for best results:
- Prompt only — good for broad conceptual targeting ("people researching electric vehicles")
- Example URLs only — good when you have known high-performing pages to expand from
- Prompt + example URLs — best results; the model uses both signals together
You can also supply a block_tlds list to exclude specific domains from the segment output.
Segment lifecycle
Create → Processing (1–2 hrs) → Complete → Activate → Live in DSP
| Stage | What's happening |
|---|---|
| Create | You submit a segment request; receive a segment ID immediately |
| Processing | Classify builds the URL list in the background |
| Complete | Segment is ready; you can now activate it |
| Activate | Classify pushes the segment and returns a Deal ID |
| Live | Target the Deal ID in your DSP |
Anti-MFA filtering
Made-for-advertising (MFA) sites are low-quality publishers that generate content purely to capture ad spend. Classify automatically filters MFA domains from all segments, so you never need to manage exclusion lists manually.
See the Segment Creation API to get started.