Our Methodology

Last updated: November 17, 2025

PivotIntel tracks AI data center infrastructure investments, community impact, and labor market effects using transparent, verifiable research methods. This page documents how we collect data, verify sources, and maintain accuracy.

Data Collection Methods

1. Automated Monitoring System

We operate a custom-built PHP-based scraper that runs daily at 6:00 AM EST, monitoring:

RSS Feed Sources (20+ feeds):

  • Data Center Knowledge
  • Data Center Frontier
  • CRN Tech News
  • TechCrunch AI coverage
  • State/regional business journals
  • Environmental news outlets

Company Press Rooms:

  • Amazon Web Services newsroom
  • Microsoft news center
  • Google Cloud blog
  • Meta newsroom
  • Oracle announcements

Google News Alerts:

  • “AI data center” + investment terms
  • “data center” + location + dollar amounts
  • State-specific infrastructure announcements
  • Permit filing notifications

2. Regional Intelligence Network

Michigan-Specific Sources:

  • Bridge Michigan
  • Detroit Free Press
  • MLive (statewide)
  • Michigan Advance
  • County planning commission websites
  • Michigan EGLE (environmental)
  • DTE Energy and Consumers Energy announcements

Great Lakes Regional:

  • Wisconsin State Journal
  • Milwaukee Journal Sentinel
  • Cleveland Plain Dealer
  • Indianapolis Star
  • Chicago Tribune

National Coverage:

  • State business journals
  • Local investigative outlets
  • Community opposition groups
  • Public hearing records

3. Manual Verification

Every project in our database undergoes human review:

  • Source verification (minimum 2 independent sources)
  • Cross-referencing with public records
  • Fact-checking dollar amounts and job numbers
  • Timeline validation
  • Status confirmation

Source Archiving

Every source URL is archived using the Internet Archive’s Wayback Machine within 24 hours of discovery.

Why we do this:

  • Press releases get deleted
  • Company pages get updated/removed
  • Public records websites change
  • Permits expire from online systems

What this means for you:

  • Every claim we make has a permanent archived source
  • You can verify our data even if original sources disappear
  • We maintain source credibility over time
  • Legal/research use cases protected

Access: All archived sources linked in our project database.

Data Fields We Track

Project Information:

  • Project name (if disclosed)
  • Company/operator
  • Location (city, county, state)
  • Investment amount (announced vs. actual)
  • Status (announced → permit → approved → construction → operational)
  • Timeline (announcement date, permit date, construction start, operational date)

Employment Data:

  • Construction jobs (temporary)
  • Permanent operational jobs
  • Local residents hired (vs. out-of-state workers)
  • Average wages
  • Hiring timeline
  • Skills required

Financial Impact:

  • Tax abatements (amount and duration)
  • Infrastructure subsidies
  • Utility rate discounts
  • Grants and incentives
  • Annual tax revenue (projected vs. actual)
  • Cost-per-job analysis

Environmental Impact:

  • Energy demand (megawatts)
  • Water usage (gallons per day)
  • Cooling system type
  • Renewable energy percentage
  • Land use (acres)
  • Previous land use

Community Input:

  • Public hearing attendance
  • Comments submitted (opposed vs. support)
  • Petition signatures
  • Vote counts (if applicable)
  • Opposition groups
  • Community benefit agreements

Verification Standards

Three-Tier System:

Tier 1 – Confirmed:

  • 2+ independent credible sources
  • Official press release + third-party news coverage
  • OR public permit filing + news coverage
  • Archived sources available
  • ✓ Green badge in database

Tier 2 – Credible:

  • Single credible source (major news outlet or company announcement)
  • Unable to independently verify all details
  • Awaiting additional confirmation
  • ⚠️ Yellow badge in database

Tier 3 – Unverified:

  • Rumor or unconfirmed reports
  • Social media claims without official source
  • Anonymous tips requiring investigation
  • ⚠️ Red badge in database (not published until verified)

We only publish Tier 1 (Confirmed) and Tier 2 (Credible) projects. Tier 3 remains in research until verified.

Calculation Methodologies

Cost Per Job:

Total Public Cost / Permanent Jobs Created

Where Total Public Cost =
Tax Abatement +
Infrastructure Subsidy +
(Utility Rate Discount × Abatement Years) +
Grants

Payback Timeline:

Total Public Cost / Annual Tax Revenue = Years to Break Even

Local Hire Percentage:

(Local Residents Hired / Total Permanent Jobs) × 100

Temp-to-Perm Ratio:

Construction Jobs / Permanent Jobs = Ratio

Example: 4,000 construction jobs / 50 permanent jobs = 80:1 ratio

What We Don’t Track

We explicitly do NOT track:

  • Proprietary business information not publicly disclosed
  • Personal information about workers
  • Security-sensitive facility details
  • Non-AI-related data centers (unless specifically relevant)
  • Speculative projects without credible sourcing

Update Frequency

Daily: Automated scraper runs, new sources identified Weekly: Database verified and updated with new information As-needed: Status changes, breaking news, community actions

Newsletter publication: Weekly (Sundays) Major analysis pieces: As warranted by data

Corrections Policy

We make mistakes. When we do, we correct them transparently.

If you find an error:

  • Email: angela@pivotintel.org
  • Subject: “Data Correction Request”
  • Include: Project name, specific error, correct information, source

Our response:

  • Investigate within 48 hours
  • Update database if verified
  • Publish correction in next newsletter if material
  • Maintain correction log on this page

Correction Log: [None to date – launched November 2025]

Data Limitations

What we acknowledge:

  1. Not all states disclose tax incentive data – 12 of 32 states with data center subsidies don’t report publicly
  2. Companies use NDAs and shell corporations – Some project details remain hidden
  3. Actual vs. promised jobs – We track both, but “promised” numbers often change
  4. Real-time vs. delayed data – Some information only becomes public months later
  5. Local hire percentages – Rarely disclosed; we estimate when data unavailable

We clearly mark estimates, projections, and incomplete data in our reporting.

Funding & Independence

PivotIntel is funded by:

  • Premium Newsletter subscriptions (future – portions will always be free)
  • Premium community analysis services (B2B, future)
  • Consulting for workforce development agencies (future)

We have ZERO:

  • Corporate sponsors
  • Affiliate relationships with job platforms
  • Advertising from companies we cover
  • Paid placements or promotional content

Why this matters: Our analysis isn’t influenced by financial relationships with the companies or platforms we evaluate.

Research Standards

We follow investigative journalism standards:

  • Verify before publishing
  • Disclose sources when possible
  • Archive all evidence
  • Correct errors transparently
  • Avoid conflicts of interest
  • Protect source confidentiality when requested

Technical Infrastructure

Our systems:

  • PHP-based automated scraper
  • MySQL database (schema documented)
  • Cron-scheduled daily collection
  • Wayback Machine API integration
  • Hosted on A2 shared hosting (U.S.-based servers)

Data retention: Indefinite (historical tracking critical to analysis)

Security: Standard web hosting security, no PII collected from users

Contact & Feedback

Questions about our methodology: angela@pivotintel.org

Data requests: Researchers, journalists, and community advocates: Contact us for specific data exports or custom analysis.

Suggestions for sources to monitor: We’re always expanding our source network. Send recommendations to angela@pivotintel.org.

This methodology is a living document. Last updated: November 17, 2025

Scroll to Top