OCR License Agreement Scans: Grants, Territory, Royalties

To OCR a license agreement scan, convert the agreement, its schedules and every amendment as separate files, then locate four things in the text: the grant clause, the definitions the grant depends on, the territory and the royalty terms. Read each of them against the page image, because a license turns on single words and single digits.

What makes a scanned license agreement hard to work from?

A scanned license agreement is hard to work from because its meaning is spread across three places: a short grant clause, the capitalized definitions that clause relies on, and schedules at the back that list the licensed property and the rates. Amendments signed later then change one of the three without restating the others.

Take a fictional trademark license between Corvane Brands Ltd. and Nettlefield Outdoor Co. The grant runs to four lines:

Licensor grants to Licensee an exclusive, non-transferable license to use the Licensed Marks on the Licensed Products in the Territory during the Term.

Almost every working word in that sentence is defined somewhere else. "Licensed Marks" points to Schedule 1, a table of registration numbers. "Licensed Products" points to Schedule 2. "Territory" is a definition on page 3. The sentence cannot be read alone, and a scan gives you no way to jump between the parts until it has text.

Three features cause the trouble during conversion:

• Capitals carry meaning. "Products" and "products" may be different things in the same agreement. Check that recognition kept the capital letter wherever a defined term appears.
• Schedules are tables and lists. Registration numbers, patent numbers and product codes are strings with no dictionary word to help recognition, so a wrong character looks as plausible as a right one.
• The file is a set. The agreement you were handed may be the 2019 original, with a 2022 amendment that redefines the territory sitting in another folder.

The general routine for a newly signed agreement of any type is in the post on contract OCR at signing. The points here are the ones peculiar to a license.

Follow each defined term in a trademark license back to its schedule, and read the registration numbers on the image. Do the same when the intellectual property is a copyright or a patent, and when a sentence says work made for hire. If the grant is a source code license, note the section and confirm the product names character by character.

Finding the grant, field of use and territory in a license agreement scan

Find the grant by searching the converted text for "grants", "hereby grants" and "license to", then follow each capitalized term in the clause back to its definition. Field of use and territory may sit inside the grant sentence or in separate definitions, so search for "Field", "Field of Use", "Territory" and "Licensed" as well.

A search list for the core terms:

Search for both "sublicense" and "sub-license". A hyphen at a line break can also split a word in the recognized text, which is one of the reasons a search misses; the post on how to search scanned contracts for a clause covers those misses and how to log what you searched.

Record what you find in a term sheet with the page of the scan beside each entry:

Quote the territory exactly. "Excluding" and "including" differ by two letters, and the exception at the end of a definition is the part a summary drops.

Royalty schedules and rate tables in a scanned license

Royalty terms in a scanned license are part clause and part table. The clause states what the rate applies to, such as a defined "Net Sales", and the table gives the rates, tiers and minimums. Export the table to a spreadsheet and keep the clause text with it, because a rate with no base is not a term.

A fictional royalty schedule from the Corvane license:

When this goes to a workbook, check four things on the page image:

1. The tier boundaries. "Up to" and "above" must meet at the same figure.
2. The percent signs and decimal points. 6.5% read as 65% or 6.596 is a plausible recognition error.
3. Merged cells. A minimum that applies to two rows may land in only one of them.
4. Footnotes. A note under the table may say the rate falls when a patent expires or rises on a renewal.

In Schema Law the Excel export puts each table on its own sheet and the rest of the text on another, so the definition of "Net Sales" will not be on the sheet with the rates. Copy the section number of that definition into a column of the rate sheet. The wider method for schedules, including row-by-row source references, is in the post on converting a scanned schedule to Excel.

Franchise agreements also charge royalties, but they arrive inside a much larger disclosure package with its own layout. The post on OCR for a franchise agreement deals with that package.

Which license agreement terms should be read against the page image?

Read against the image every term where one wrong character changes the deal: royalty rates, minimums, tier thresholds, dates, notice periods, registration and patent numbers, and the list of countries. Recognized text gives no warning when a character is doubtful, so these are checked by eye, one at a time.

A checklist for a license:

• Every percentage in the royalty clause and schedule.
• Every currency amount, including advances and minimum guarantees.
• The effective date, the term, each renewal term and the date the first royalty report falls due.
• The number of days in each notice, cure and audit period.
• The payment and reporting interval: "quarterly" against "monthly", "within 30 days" against "within 80 days".
• Each registration, application or patent number in the schedule of licensed property, against the image character by character.
• Each country or region in the territory, and each exclusion.
• The words "exclusive", "non-exclusive" and "sole" wherever they appear.
• The governing law and the party names as written.

Initial each checked entry on the term sheet. A term sheet with no check marks should be treated as unchecked by whoever picks it up.

Tracking amendments and side letters to a scanned license

Track amendments by listing every document that touches the license, in date order, with the clause each one changes. Convert each as its own file and note on the term sheet which document a current term comes from. A term sheet built from the original alone is a record of what the license said on its first day.

A version log for the fictional license:

Side letters are the documents to hunt for. A one-page letter that waives a minimum or consents to a sublicense may be filed with correspondence and never placed in the contract folder. Ask the business contact directly whether any letters or emails changed the deal, and search the converted set for "waive", "consent" and "notwithstanding".

When an amendment replaces a schedule, mark the old workbook as superseded and keep it. A question about 2021 sales starts from the schedule that was in force in 2021.

Check the license for a confidentiality clause covering its own terms. If it has one, read the commitments a conversion service makes before uploading the scans. Schema Law's are listed under security on the home page.