> For the complete documentation index, see [llms.txt](https://docs.thousandeyes.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.thousandeyes.com/product-documentation/internet-insights/provider-intelligence/data-sources-and-coverage.md).

# Data Sources and Coverage

## Overview

Provider Intelligence's recommendations are only as good as the data underpinning them. This article explains:

* **Where the data comes from**: ThousandEyes Cloud and Enterprise Agents and third-party provider databases.
* **Coverage gaps**: Why some providers appear with "no coverage" and how to interpret this.
* **Agent proxies**: How Cloud and Enterprise Agent data serves as a proxy for end-user experience.

Understanding data sources is critical for evaluating the reliability and objectivity of Provider Intelligence results.

## Primary Data Source

ThousandEyes operates a global network of Cloud and Enterprise Agents, which are virtual machines deployed in data centers worldwide that continuously generate synthetic tests. These tests measure:

* **Network-layer metrics**: Latency, packet loss, jitter.
* **Application-layer metrics**: Time to First Byte (TTFB), HTTP response times, page load times.

### Coverage

Cloud and Enterprise Agents are deployed in hundreds of cities across six continents, providing broad geographic coverage. However, coverage is not universal:

* **High Coverage**: Major cities in North America, Europe, and parts of Asia-Pacific (for example, New York, London, Tokyo).
* **Moderate Coverage**: Secondary cities and emerging markets (for example, Bangalore, São Paulo, Johannesburg).
* **Low Coverage**: Remote or smaller cities might have limited or no agent presence.

To check current coverage:

1. Navigate to **Internet Insights > Catalog Settings > Providers** or **Internet Insights > Catalog Settings > Packages.**
2. Click any package or provider to view its **Coverage Map** (if available).

### Test Frequency and Volume

* **Granularity**: Tests run continuously, with data aggregated at 1-hour intervals.
* **Volume**: Provider Intelligence processes over 8 billion data points daily from Cloud and Enterprise Agent tests.
* **Duration**: Historical data is retained for up to 6 months.

## Third-Party Data Sources

Not all Autonomous Systems (ASNs) represent providers that you can purchase internet services from. Some are:

* **Private ASNs**: Used by individual organizations for internal routing (for example, a bank's internal network).
* **Content Delivery Networks (CDNs)**: Not traditional networks (for example, Cloudflare, Akamai).
* **Hosting Providers**: Might not offer commercial connectivity contracts (for example, AWS, Google Cloud).

To ensure Provider Intelligence only compares **commercially available** **providers**, ThousandEyes uses two authoritative databases:

* APNIC (Asia-Pacific Network Information Centre)
* PeeringDB

### APNIC

Website: <https://stats.labs.apnic.net/aspop>

#### What It Provides

APNIC maintains a database of ASNs with metadata including:

* Geographic presence (countries, cities).
* Provider type (ISP, CDN, hosting provider).
* Commercial availability status.

#### How Provider Intelligence Uses It

APNIC data is used to:

1. Identify which ASNs are ISPs vs. other network types.
2. Determine geographic footprint (for example, "Verizon operates in 50 US cities").
3. Filter out private or non-commercial ASNs from user-facing results.

### PeeringDB

Website: [https://www.peeringdb.com](https://www.peeringdb.com/)

#### What It Provides

PeeringDB is a community-maintained database of:

* Internet exchange points (IXPs).
* Peering relationships between ASNs.
* Provider facilities and points of presence (PoPs).

#### How Provider Intelligence Uses It

PeeringDB data is used to:

1. Validate which providers have peering relationships with major cloud providers (for example, AWS, Azure).
2. Identify providers with extensive peering, which often correlates with better performance.
3. Cross-reference APNIC data for accuracy.

### Legal and Attribution

ThousandEyes complies with the licensing terms of both APNIC and PeeringDB:

* **APNIC**: Re-use is permitted with attribution.
* **PeeringDB**: Data is used in accordance with PeeringDB's community guidelines.

## Providers with No Coverage

In the **Summary View**, when at least one application is selected as a destination, Provider Intelligence can show providers that are present in the selected location but have no test coverage for the current query. These no-coverage providers are identified from broader provider-availability data gathered for the universal application summary, which scores providers with test coverage to *any* application from a given location.

These additional providers are therefore known to exist in a location but do not have test data for the applications you specified in your query. Adding providers without coverage to your results enables you to view a complete set of providers at a location even when your specific query cannot score them for lack of test data, helping to prevent coverage bias.

### Why Providers Have No Coverage

A provider might have no test coverage in a query when:

* No ThousandEyes test data traverses that provider’s network to the selected destinations.
* The provider is present in the location, but the current destination set has no matching data.
* Routing or peering paths to the selected destinations use other provider networks.

### How to Interpret No Coverage

No coverage does not mean a provider is unavailable, unsuitable, or experiencing an outage. It means Provider Intelligence cannot make an objective performance comparison for that provider within the current query parameters.

### How to Access Scores for No-Coverage Providers

In the **Summary View**, no-coverage providers do not receive scores or metric values for the query, and you cannot select them for reports, monitoring, or time series.

While the **Summary View** cannot score no-coverage providers due to the restrictions of your query parameters, you can switch to the **Universal Application Summary** view to find scores for these providers, plus all other providers in a location that have tests running to at least one application. This is because all applications with test coverage are taken into account in the **Universal Application Summary** view, not just the applications selected for your specific query.

For more information about the different views, see [Destination Views](https://docs.thousandeyes.com/product-documentation/internet-insights/provider-intelligence/view-results/destination-views).

## Agent Proxies

### The Proxy Model

Cloud and Enterprise Agents are not end-user devices. They are centralized test infrastructure deployed in data centers. However, they serve as **proxies** for end-user experience because:

* **Location**: Agents are deployed in cities where your end-users are located (for example, agents in Austin represent end-users in Austin).
* **Provider network path**: Agents send tests through the same provider networks that your end-users would use.
* **Synthetic tests**: Tests mimic real user actions (for example, loading a web page, connecting to a cloud service).

### Limitations

* **Last-mile differences**: Cloud Agents are in data centers with high-quality connectivity. Your end-users on residential or cellular networks might experience different performance.
* **Device variability**: Agents use standardized configurations. Real users have diverse devices, browsers, and provider network conditions.
* **Geographic granularity**: An agent in "Austin, TX" might not represent performance in specific neighborhoods or suburbs.

## Data Freshness and Updates

### Aggregation Schedule

Provider Intelligence data is updated on a **daily basis**:

1. **Hourly tests**: Cloud Agents generate test data every hour.
2. **Daily aggregation**: At midnight UTC, the platform processes the previous 24 hours of data, calculating scores and trends.
3. **Historical queries**: When you run a query, pre-aggregated scores for your selected time frame (1, 3, or 6 months) are retrieved from the database.

**Implication**: Data reflects performance up to 24 hours prior. Real-time data (for example, last hour) is not available in Provider Intelligence; use **Network & App Synthetics > Views** for real-time monitoring.

### Data Retention

* **6 months**: Standard retention for all Internet Insights customers.
* **Longer retention**: Contact ThousandEyes Sales to discuss data retention.
