Delivery

The Signal Database is delivered as structured flat files in your choice of format. This page covers file formats, delivery mechanisms, and refresh schedules.

File Formats

JSON (NDJSON)

Newline-delimited JSON (NDJSON) format—one signal record per line. Ideal for streaming ingestion and simple parsing.

File extension: .ndjson or .json

Example:

[
  {
    "signal_id": "a1b2c3d4-5678-90ab-cdef-1234567890ab",
    "insight_subtype": "workExperienceJobChange",
    "entity_type": "contact",
    "entity_identifiers": {
      "contact_email": "[email protected]",
      "contact_linkedin_url": "https://linkedin.com/in/sarahchen",
      "company_domain": "newco.io"
    },
    "detected_at": "2024-12-10T14:22:00Z",
    "source": "Work Experience",
    "variables": {
      "new_job_title": "VP of Sales",
      "new_job_company_name": "NewCo",
      "previous_company_name": "OldCorp"
    }
  },
  {
    "signal_id": "b2c3d4e5-6789-01bc-def2-3456789012cd",
    "insight_subtype": "newsFunding",
    "entity_type": "company",
    "entity_identifiers": {
      "company_domain": "techstartup.com",
      "company_name": "TechStartup Inc"
    },
    "detected_at": "2024-12-12T09:15:00Z",
    "source": "News",
    "variables": {
      "insightTitle": "TechStartup Raises $50M Series B"
    }
  }
]

Pros:

• Human-readable
• Easy to parse in any language
• Streaming-friendly (process line by line)
• Simple to debug and inspect

Recommended for: Real-time ingestion pipelines, smaller datasets, debugging

Parquet

Apache Parquet columnar format. Optimized for analytical queries and data warehouse ingestion.

File extension: .parquet

Schema:

signal_id: STRING
insight_subtype: STRING
entity_type: STRING
entity_identifiers: STRUCT<
  contact_email: STRING,
  contact_linkedin_url: STRING,
  company_domain: STRING,
  company_linkedin_url: STRING,
  company_name: STRING
>
detected_at: TIMESTAMP
source: STRING
variables: STRING (JSON-encoded)

Note: The variables field is stored as a JSON string within Parquet. This allows flexibility across 350+ signal subtypes while maintaining a consistent columnar schema.

Pros:

• Highly compressed (typically 5-10x smaller than JSON)
• Columnar format enables fast analytical queries
• Native support in Spark, Snowflake, BigQuery, Databricks, etc.
• Efficient for large-scale batch processing

Recommended for: Data warehouses, analytical workloads, large datasets

Delivery Mechanisms

Push to Your S3 Bucket

We deliver files directly to your AWS S3 bucket on a scheduled basis.

Setup:

1. Create a dedicated S3 bucket (or prefix within an existing bucket)
2. Provide us with a cross-account IAM role ARN with s3:PutObject permissions
3. We deliver files on your chosen schedule

File structure:

Files are organized by insight type, with separate folders for each signal category you're subscribed to. This structure ensures consistent schemas within each folder and simplifies ingestion pipelines.

s3://your-bucket/autobound-signals/
├── news/
│   ├── 2024-12-15/
│   │   └── news_2024-12-15.parquet
│   └── 2024-12-16/
│       └── news_2024-12-16.parquet
├── job_changes/
│   ├── 2024-12-15/
│   │   └── job_changes_2024-12-15.parquet
│   └── ...
├── hiring_trends/
│   └── ...
├── linkedin_activity/
│   └── ...
├── technographics/
│   └── ...
├── financials/
│   └── ...
├── website_intelligence/
│   └── ...
├── sec_filings/
│   └── ...
└── full/
    ├── news_full_2024-12-01.parquet
    ├── job_changes_full_2024-12-01.parquet
    └── ...

Why organize by insight type? Each signal category has a distinct schema with different variables fields. Separating by type allows you to define strongly-typed tables for each category and avoid parsing heterogeneous data within a single file.

Pull from Autobound S3

Access files from our S3 bucket using temporary credentials.

Setup:

1. We provision read-only credentials for your account
2. You pull files on your own schedule
3. Files are retained for 30 days

Access pattern:

# Example: List available insight type folders
aws s3 ls s3://autobound-signals-{your-org-id}/

# Example: List available dates for job changes
aws s3 ls s3://autobound-signals-{your-org-id}/job_changes/

# Example: Download today's news signals
aws s3 cp s3://autobound-signals-{your-org-id}/news/2024-12-15/ ./signals/news/ --recursive

# Example: Download all signal types for a specific date
for type in news job_changes hiring_trends technographics; do
  aws s3 cp s3://autobound-signals-{your-org-id}/$type/2024-12-15/ ./signals/$type/ --recursive
done

Refresh Cadence

Signal refresh frequency varies by category. Higher-velocity signals (news, job changes) refresh more frequently than stable signals (tech stack, business model).

Incremental vs. Full Loads

Incremental (Default):

• Daily/weekly files contain only new or updated signals since last delivery
• Use detected_at timestamp to identify new records
• Smaller file sizes, faster processing

Full Snapshot (On Request):

• Complete database dump
• Useful for initial load or periodic reconciliation
• Delivered monthly or on-demand

File Naming Convention

Files follow a consistent naming pattern based on insight type:

{insight_type}_{date}.{format}

Examples:

• news_2024-12-15.ndjson — News signals (funding, M&A, launches) from Dec 15
• job_changes_2024-12-15.parquet — Job change signals from Dec 15
• technographics_2024-12-15.parquet — Tech stack signals from Dec 15
• news_full_2024-12-01.parquet — Full news snapshot from Dec 1

Available insight types:

Data Volume Estimates

Typical daily delivery sizes per insight type (varies by coverage):

Delivery SLA

• Daily signals: Delivered by 6:00 AM UTC
• Weekly signals: Delivered Sunday by 6:00 AM UTC
• Availability: 99.9% uptime SLA
• Support: Issues resolved within 4 business hours

Getting Started

1. Select insight types: Choose which signal categories you want to subscribe to (news, job changes, technographics, etc.)
2. Choose your format: JSON for simplicity, Parquet for analytics
3. Set up delivery: Provide S3 bucket details or request pull access
4. Configure filtering: Optionally filter by geography or company size
5. Test with sample: We provide sample datasets for each insight type for integration testing

Contact [email protected] to configure your delivery.