> ## Documentation Index
> Fetch the complete documentation index at: https://docs.leadinfo.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Set up scoring profiles

> Draft an ICP Builder and an Intent Builder profile from your own data, save them, and watch the unified score explain itself.

Out of the box the score uses engagement and people. Two profiles make it yours: the ICP Builder describes who you sell to, the Intent Builder which pages mean buying intent. Both can be drafted from your data through the API and saved in the Portal or through the [Leadinfo MCP](https://docs.leadinfo.com/product/mcp/overview.md).

**1. Check what is in force**

```bash cURL theme={null}
curl --request GET \
  --url "https://data.leadinfo.com/v1/scoring-model" \
  --header "Authorization: Bearer <token>"
```

`icp_profiles` and `intent_profiles` are empty when a pillar has no active profile yet; `pillars` shows the effective weights and `reading_guide` explains the method.

**2. Draft the ICP profile**

```bash cURL theme={null}
curl --request GET \
  --url "https://data.leadinfo.com/v1/profiles/icp/suggestions?tracker_code=LI-5A14070B93509" \
  --header "Authorization: Bearer <token>"
```

Review `rules` against `evidence`: each value shows its share of your customers and its lift against all visitors. Drop values that describe your past rather than your target. Every rule must match, so keep a profile to the rules that define the segment.

**3. Save it as an active profile**

Save the draft in the Portal's ICP Builder, or ask the Leadinfo MCP to save it; it shows you the rules and saves after you confirm. The draft names sector groups (SIC major groups); the ICP Builder holds full four-digit sector codes, so pick the codes within those groups, and leave the draft's revenue floor out.

A new profile is active. Up to five ICP profiles count, the five oldest active ones; each company keeps its best match, so add one profile per segment instead of one profile that covers everything.

**4. Draft and save the Intent profile**

`GET /v1/profiles/intent/suggestions?tracker_code=...` groups the pages your visitors actually viewed in the last 180 days into families such as pricing, demo or trial, contact, comparison, integrations and product pages. Rules use the page path, so `url contains ["/pricing", "/tarieven"]` matches either language. The draft ends with an exclusion rule (`url not_contains [...]`) for login, blog, careers and legal pages; extend it with pages of your own that never signal interest. Save it in the Intent Builder or through the Leadinfo MCP.

**5. Read the result**

Within a minute (stored scores: at the next scoring run), company rows carry `icp_score`, `intent_score` and a recomputed `score`, `meta.scoring` lists the active profiles, and `GET .../companies/{company_id}/score` lists which rules each company matched and, with several active profiles, which profile it matched best.

> [!NOTE]
>
> Personas are drafted the same way (`kind=persona`) and saved in the Portal's Persona Builder or through the Leadinfo MCP. The API returns the drafts in Persona Builder format.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.