Compliance first: We never take HIPAA/PHI data or personal customer data. Everything is de-identified before it leaves your control.
Home / Insights / Analysis
Analysis

From Open Data Policies to Private Data Markets: What Cities Taught Us About Data Value

From Open Data Policies to Private Data Markets: What Cities Taught Us About Data Value

For more than a decade, open data policies pushed governments to publish machine-readable data for free. The same principles of inventory, standard formats and privacy protection now shape a new private market where businesses license de-identified operational data to AI builders.

OpenDataPolicies.org began as an index of city, county and state open data policies, the ordinances, executive orders and directives that committed governments to publishing their data. That archive still lives here in our guidelines and browse page. This article connects that history to what the site covers today: a private market where businesses can be paid for de-identified operational data. The two are more closely related than they look.

By OpenDataPolicies Editorial · November 19, 2026 · 6 min read

What the open data movement got right

The open data movement argued that data held by government is a public asset and should be released proactively, in formats machines can read, with privacy protected. Policy groups codified those ideas into model guidelines. The Sunlight Foundation's Open Data Policy Guidelines became a widely used reference for cities drafting policies; the foundation stopped operating in September 2020, and its guidelines are maintained as a static archive (Sunlight Foundation).

At the federal level, Data.gov launched in 2009 and is managed by the U.S. General Services Administration (Data.gov). The OPEN Government Data Act, Title II of the Foundations for Evidence-Based Policymaking Act of 2018, requires government data to be made available in open, machine-readable formats while continuing to ensure privacy (Data.gov; GovInfo, Public Law 115-435).

Three lessons from that era carry over directly:

  • Inventory first. Good policies required agencies to list what data they held before deciding what to release.
  • Machine-readable formats matter. Data locked in PDFs or proprietary formats is far less useful than structured, documented data.
  • Privacy is a precondition, not an afterthought. Every serious open data policy carved out personal and sensitive information.

You can explore how individual governments put those principles into practice in our browse page and the revived guidelines.

Why open data alone could not feed the AI boom

Open government data is valuable, but it describes government: budgets, permits, inspections, transit, crime reports. It says little about how a private dental office runs its front desk, how a contractor manages change orders, or how a support team resolves a billing dispute.

Meanwhile, the broader open web that trained the first generation of large language models is being used up and fenced off. Epoch AI estimated the stock of quality-adjusted, human-generated public text at around 300 trillion tokens and projected it would be fully used sometime between 2026 and 2032 if trends continue (Epoch AI). The Consent in Crisis study found that, within a single year, about 5% or more of all tokens in the C4 dataset, and 28% or more of its most critical sources, became fully restricted by robots.txt (arXiv).

So AI builders face two gaps at once: the open web is running dry, and open government data was never designed to describe private operations. We explore the first in why the AI data window is closing.

The private data market that emerged

Where free data runs short, licensed data fills in. Public reporting since 2023 shows the shape of that market: Google's reported $60 million-a-year deal for Reddit data (The Verge), and the Associated Press licensing part of its text archive to OpenAI (AP). These are well-known public deals between large organizations; they are not ODP transactions. We cover them in more detail in what AI companies have paid for data.

The next phase, in our view, is about operational data: the email, chat, tickets, CRM records, projects, accounting and industry software that show how businesses actually work. That data is private by default, which is exactly why it is valuable. As founder Jeff Cline puts it, "Every industry is a geek away from being uberized," and the builders doing that work need to see real workflows. See operational data vs web data.

Same principles, different deal

The private market is not open data. It is licensed, paid for and controlled by the data owner. But the disciplines that made open data work translate almost one-for-one:

  • Data inventory → data readiness. Cities listed their datasets; businesses list their systems and record counts. See the data readiness checklist.
  • Metadata and documentation → header sample. Open data portals publish field descriptions; data sellers share a header sample with field names, counts and date ranges.
  • Machine-readable formats → clean exports. Structured exports from systems like Salesforce or Zendesk are worth more than unstructured dumps.
  • Privacy exemptions → de-identification. Public bodies withheld personal data; private sellers remove it before anything leaves. We never take HIPAA/PHI data or personal customer data. Everything is de-identified before it leaves your control.
  • Open licenses → commercial licenses. Instead of public-domain or open licenses, private data is shared under data licensing agreements that set term, exclusivity and permitted uses.
  • Proactive release → proactive valuation. Open data advocates urged governments not to wait for records requests; we urge businesses not to wait until the window closes.

For a longer treatment of this bridge, see our guide from open data policy to data value.

What changes when data has a price

Open data asked, "Should this be public?" The private market asks, "What is this worth, and to whom?" That shift brings new questions:

  • Rights. Do you own or control the data, and do contracts or vendor terms limit its use? See the data rights checklist. This is informational, not legal advice.
  • Pricing. Buyers weigh volume, uniqueness, recency, labeling and exclusivity. See how buyers price data.
  • Timing. Open data keeps its civic value indefinitely; training data loses value as models learn what it teaches.
  • Competition. According to founder Jeff Cline, ODP seeks offers from multiple buyers so a seller can take the highest bid, bids typically arrive within 72 hours if the data is correct and a header sample is provided, and deals typically close in about 90 days. Results vary; not a guarantee.

Lessons from where open data struggled

Open data programs did not always deliver on their promise, and their shortfalls are instructive for anyone entering the private market.

  • Publishing is not the same as being useful. Portals filled with datasets that were stale, poorly documented or inconsistently formatted saw little reuse. In the private market, buyers discount data for the same reasons, so documentation and consistency translate directly into price.
  • Quantity is not quality. Counting datasets released became a vanity metric for some programs. Buyers of training data care far more about uniqueness and completeness than about how many files you have.
  • Privacy mistakes are expensive. Releases that were later found to expose personal details damaged trust in open data. That is why de-identification must happen before anything leaves the data owner's control, and why some categories, such as protected health information, are excluded entirely.
  • Maintenance matters. Datasets that were published once and never updated lost value quickly. In the private market, recent data and the option of ongoing delivery tend to command better terms.

Each of these lessons points to the same conclusion: preparation pays. A business that inventories its systems, documents its fields and confirms its rights before it goes to market tends to get faster, stronger bids.

For the legacy open data audience

If you came to OpenDataPolicies.org for the policy archive, it is still here, and we have kept those pages as a historical record. Researchers, civic technologists and public officials can still use the guidelines and browse page.

If you also run, advise or work for a business, the skills you built around open data (inventories, metadata, formats, privacy review) are exactly what the private data market rewards. The difference is that this time, the organization that holds the data can be paid for it.

Next step

Curious what your organization's operational data could be worth? Requesting a bid is free. Try the valuation calculator or go straight to a free bid request.

Our compliance promise: We never take HIPAA/PHI data or personal customer data. Everything is de-identified before it leaves your control.

Frequently asked questions

What is an open data policy?

A law, executive order or administrative directive that commits a government to publishing its data proactively, in machine-readable formats, while protecting privacy and security.

How is a private data market different from open data?

Open data is published free for anyone. In a private data market, a business licenses a de-identified copy of its data to specific buyers under contract, for payment, while keeping control of the original.

Does OpenDataPolicies.org still cover government open data policies?

Yes. The legacy policy archive remains available through the guidelines and policy browser pages.

Is personal data ever part of a private data deal?

Not with ODP. We never take HIPAA/PHI data or personal customer data. Everything is de-identified before it leaves your control.

Sources

  1. Sunlight Foundation — Open Data Policy Guidelines (archive)
  2. Data.gov — About
  3. GovInfo — Public Law 115-435, Foundations for Evidence-Based Policymaking Act of 2018
  4. Epoch AI — Will we run out of data?
  5. Longpre et al. — Consent in Crisis (arXiv 2407.14933)
  6. The Verge — Google cut a deal with Reddit for AI training data
  7. AP — AP, OpenAI agree to share select news content and technology

Updated November 19, 2026.

OpenDataPolicies.org connects businesses with buyers of de-identified operational data. We are not a law firm, broker-dealer or investment adviser, and nothing on this site is legal, tax or financial advice. Statements about amounts, timing and outcomes are Jeff Cline's statements based on past transactions. Results vary; nothing here is a guarantee of any bid, price, buyer or closing. Every deal depends on the data, its quality, its rights and buyer demand at the time.