Analytics

Five endpoints return the numbers behind the Fuse insights tabs. They are read-only, cost no credits, and each covers one area. They need a key with the analytics scope.

Endpoint	Covers
`GET /business/analytics/campaigns`	Email and LinkedIn volume, replies and engagement
`GET /business/analytics/dialer`	Call volume, connects and outcomes
`GET /business/analytics/enrichment`	Emails and phones enriched over time
`GET /business/analytics/website-intent`	Visits, companies and people seen on your site
`GET /business/analytics/agents`	What your agents have delivered

Pick a window

Every endpoint takes one optional range:

`range`	Window
`7d`	Last 7 days
`30d`	Last 30 days (default)
`90d`	Last 90 days

curl -G https://api.tryfuse.ai/api/v1/business/analytics/website-intent \
  --data-urlencode "range=7d" \
  -u "$FUSE_API_KEY:"

{
  "range": "7d",
  "metrics": {
    "totalVisits": 1518,
    "uniqueCompanies": 182,
    "uniquePeople": 47,
    "highConfidenceVisits": 206
  },
  "timeSeries": [{ "date": "08/18", "companies": 76, "people": 13 }]
}

The range you asked for is echoed back, so a stored response is self-describing.

The shape of a response

Each endpoint returns one or more panels. A panel is either a set of summary cards or a time series, and the response always names the range it covers.

{
  "range": "30d",
  "agents": {
    "totals": {
      "totalAgents": 6,
      "activeAgents": 4,
      "totalProspects": 480,
      "peopleProspects": 310,
      "companyProspects": 170
    },
    "byType": [{ "monitorType": "job_postings", "prospects": 210 }]
  }
}

A panel can come back null. Each one is computed independently, so a slow or unavailable source degrades that panel rather than failing the whole call. Treat null as “not available for this window”, not as zero — and read a missing panel differently from an empty one.

Buckets in a time series

Series are bucketed by the window, and the bucket is named in the payload rather than implied:

{
  "emailsEnriched": {
    "totalSince": 644,
    "mode": "week",
    "series": [{ "date": "W30", "total": 5 }, { "date": "W31", "total": 141 }]
  }
}

mode tells you how to read date — W31 is an ISO week, a 08/18 is a day. Read mode rather than pattern-matching the label, since the same endpoint switches bucket as the range widens.

Deltas

Card values often carry a comparison against the previous window of the same length:

{ "totalCalls": { "value": 412, "deltaPercentage": null } }

deltaPercentage is null when there is no comparable previous window — a new workspace, or a range that reaches past the available history. Render that as “no comparison” rather than 0%.

Putting it in a dashboard

Fetch the areas you show, in parallel

The five endpoints are independent; nothing needs to be called in order.

Handle null panels per panel

One unavailable source should blank one card, not the page.

Read `mode` before formatting dates

It changes with the range.

Cache for the window you display

These are roll-ups over days; polling them frequently spends rate limit for numbers that barely move.

Get started

Concepts

Guides

CRM sync

Power dialer

Deep Research

Signals

MCP server

Pick a window

The shape of a response

Buckets in a time series

Deltas

Putting it in a dashboard

​Pick a window

​The shape of a response

​Buckets in a time series

​Deltas

​Putting it in a dashboard

Pick a window

The shape of a response

Buckets in a time series

Deltas

Putting it in a dashboard