How the preliminary estimate works

Volume alone does not establish value. The model uses your answers to estimate potential usefulness to an AI developer and readiness for licensing. Its heaviest value factor is the reported chain from input to human decision to real-world outcome. No records or agreements have been inspected.

Factors are listed in order of influence within each group. Higher scores reflect more favorable model assumptions, not verified rights, quality, or demand.

Indicative aggregate licensing value across one or more licenses over roughly the first two to three years, not a sale price. The range combines reported scale, assumed usefulness for AI work, assumed buyer demand, and reported licensing readiness. It is not a price, offer, bid, or appraisal. Rights, uniqueness, quality, scale, freshness, provenance, permitted uses, and buyer demand need separate review.

1. 01

   Volume sets a starting point

   The model assigns each record band a dollar starting point that rises more slowly than record count. Confirmed volumes may receive higher starting points for terabyte-scale media. Unknown record counts use no more than the 10,000 to 100,000 band, with no media uplift. These are model assumptions, not observed license prices.

2. 02

   Quality multiplies it

   The value factors combine into an index that multiplies the starting point. The model gives more weight to paired decisions and outcomes in specialized domains than to generic record volume. This does not establish what a buyer would pay.

3. 03

   Demand assumptions adjust it

   Fixed assumptions about each industry and the reported signals adjust the estimate. They are not live buyer requests or evidence of demand for your dataset. Buyer fit must be tested separately.

4. 04

   Rights and readiness discount it

   The readiness factors apply 35% to 100% of the modeled amount. Unclear rights, personal information, and third-party material can reduce it. This adjustment is a scoring rule, not a determination that licensing is permitted.

5. 05

   Freshness adjusts it

   The model increases the amount by about 15% when updates are reported as daily or continuous. This assumes possible ongoing deliveries; it does not establish a subscription premium or licensing rights.

6. 06

   High rights risk reduces it further

   High rights risk applies 50% of the amount, multiplied by the capped score divided by the uncapped score. The range is closed and limited to 25% of the record-band ceiling, rounded down. The result lists the applied caps and policies.

7. 07

   Uncertainty sets the width

   The high end starts at about 3.5 times the low end. Each "Not sure" answer lowers the low end, up to a ratio of 7, without raising the high end. Bounds are rounded to familiar figures within the applicable ceiling. Unknown volume also keeps the range closed.

Each record band has a display ceiling set by the model. For known volumes without high rights risk, when the calculated range exceeds that limit, we show the ceiling with a plus sign. The + is a limit of this calculator, not evidence that a buyer would pay that amount or more. These ceilings are assumptions, not market benchmarks.

We show the score but no figure when

• You indicated the business likely does not hold the rights to license the data.
• The data is mostly simulated or test activity, outside this calculator's focus on operating records.
• The overall score is below 35.
• The modeled high end would fall below $25K, this calculator's minimum display threshold, not a measured cost of licensing.

Withholding a range reflects this calculator's scope and input rules, not a judgment on your business or on all synthetic data. The model also caps the score for some answers: likely lacking rights caps it at 45, mostly simulated data at 50, and extensive or sensitive personal information without confident rights at 65.

• Whether your contracts, privacy notices, and customer terms permit the proposed uses. Your counsel must review those rights; ownership alone does not establish them.
• How accurate, complete, and consistently recorded the data is, and whether its collection history is documented.
• Whether outcomes really join to records at the volume described, and how much work the join takes.
• Which buyers need this exact structure right now, and what they have already licensed elsewhere.
• What de-identification and redaction will cost, and how much useful signal they remove.
• Market conditions at the time a deal is negotiated.

Diligence under NDA may refine or rule out an opportunity. Any license fee is negotiated separately after rights, privacy, quality, and buyer fit have been reviewed. Every assessment records the model version that produced it, so estimates can be re-run when the model is recalibrated.