Calling a signed license or a newly formed company a finished application leads firms and funders to the wrong account of what the university research delivered. University research reaches a real-world application when people outside the university use the result in a product, a service, or a working practice.
University research reaches real-world applications through licensing, spinout companies, industry collaboration, consulting, graduates and researcher mobility, and open publication. The World Intellectual Property Organization lists those paths inside a longer set that also includes presentations, patent documents, staff exchange, and postgraduate projects. A 2019 OECD report splits formal channels, which leave contracts and hiring records, from informal channels such as publication and personal contact. A technology transfer office usually handles licensing and spinouts. A license, a patent, or a company registration is a record of activity. Use is a later record: a product sold to the public, or a process placed into commercial use. The worksheet on this page leaves that use cell blank until the file contains the record.
| Route | What the record is | Dominant failure mode | Record of use |
|---|---|---|---|
| Licensing | Permission to use intellectual property on the terms in the agreement. Ownership stays with the licensor unless a separate assignment transfers it. | A signed license is treated as a product on sale | A product sold to the public, or a process placed into commercial use |
| Spinout | A new company based on technology developed at the university. The rights still require a license or an assignment. | Company formation is treated as ownership of the intellectual property, or as a product | The company sells a product or runs a commercial process on that technology, with the rights agreement still in force |
| Industry collaboration | Joint, sponsored, or contract research. Ownership follows that agreement. | A research contract is treated as the partner’s adoption of a result | The partner applies the result in its own product or process |
| Consulting | Expertise a researcher supplies, where institutional policy allows it | A consulting invoice is treated as a changed process | The client changes a design, a decision, or a process |
| Graduates and mobility | People carry methods into firms and public bodies | A hiring record is treated as use of one method | An employer record that the method is applied |
| Open publication | Other people can read the result | A citation, a download, or a news release is treated as use | Documented use in a product, a practice, a standard, or a policy |
Six routes from the lab to the outside world
Six routes cover the channels in WIPO’s technology-transfer FAQ and in the OECD’s 2019 report on university-industry collaboration. The routes overlap. The arrows in the diagram are possible transfers. They are not successes.
Licensing
A licensing agreement is a contract. The owner gives another party permission to use the intellectual property on terms written in the agreement, such as purpose, territory, and field of use. Ownership stays with the licensor. An assignment is the other instrument: it transfers ownership permanently, and WIPO compares that effect to a sale of a tangible asset. Some licenses carry royalties. Some are royalty-free. WIPO states that a license becomes an actual transfer when the licensor delivers the technology and the knowledge, and the licensee learns to use it, adapt it, and where possible improve it.
Spinouts
WIPO describes an academic spin-off, also called a spin-out, as a newly created company based on technology developed at a university or research institution. On that page, spin-offs are often owners or exclusive licensees of the intellectual property, and the researchers involved often leave the university for the company. WIPO uses a second contract label, a university research-based start-up, for a company built on a university license whose founders are not affiliated with that university and whose money comes from external sponsors. Institutions do not all use those two labels the same way. In both descriptions the rights still come from an agreement. Forming the company moves no rights by itself. WIPO states that the agreement with a start-up has to cover the intellectual property, the financial terms, conflicts of interest, the inventor’s participation, and development milestones.
Industry collaboration
Collaborative research puts two or more parties’ people, facilities, money, and existing intellectual property into one project. They agree who owns the results, who can use them, and who can commercialize them. Sponsored research is different on WIPO’s page: the sponsor funds work in an area relevant to its business and receives preferential access or rights, and the sponsor does not necessarily take part in the research. WIPO states that the university usually owns those results and grants the sponsor a license. Contract research is a hire. The company sets a commercial goal, fully covers the research costs and the intellectual-property protection, bears the research risk, and usually owns the results. A material-transfer agreement moves physical research materials. It defines rights in those materials and in later results. It is not a record that a product is on sale.
Consulting
Consultancy is expertise that professors or researchers provide to an industry partner for payment, often on a personal basis, where the university’s policy allows it. WIPO states that in most cases the company owns the resulting intellectual property, with limited rights for the researcher to publish, and that ownership can also be shared under the institutional policy and the terms of the agreement. When the university joins the negotiation, WIPO states that the university tries to preserve the researcher’s right to publish, while keeping the information confidential for a reasonable period so the company can protect the intellectual property.
Graduates and researcher mobility
The OECD’s formal list includes research mobility, meaning researchers working in industry and the reverse, including temporary assignments, and labour mobility, meaning graduates who join industry. WIPO’s list adds students entering the workforce, postgraduate projects, and staff exchange between the institution and industry. A person trained in a research group can carry a method into an employer with no patent and no license. The OECD states that formal channels leave records that can be traced and counted, such as hiring contracts, and that informal contacts are harder to measure because they carry tacit knowledge.
A course on offer, a graduate hired, and a method used at work are three observations. How to record signup, completion, and an observed work task separately is the sheet for a training program. Use it when the claim is that people carried a skill into a job.
The OECD also lists networking that geographic proximity makes easier, including firms’ laboratories on a campus and science parks beside a university. A campus inside a manufacturing district shows where the buildings are. How manufacturing clusters develop separates that shared address from a documented connection between firms and the institutions around them.
Open publication
WIPO lists publications among the channels, and it states that some knowledge is best released into the public domain, while in many other cases protection and a transfer to a company able to turn an invention into products does more for the public benefit. A paper, a dataset, or a talk lets other people read and test the result. Access is not a record that anyone applied it. A patent document makes a claim public. It is not a product on sale.
The weight of each route changes with the field. On printed page 33 of the 2019 report, the OECD cites studies finding patenting and licensing very important for researchers in materials science and chemical engineering, and considerably less so in computer sciences (Bekkers and Bodas Freitas, 2008). The same paragraph cites work finding contract research, collaborative research, labour mobility, and the flow of students very important in engineering. Those are findings of the studies the OECD reviews. They are not a new measurement in the 2019 report, and they are not a ranking of departments.
What a technology transfer office does
The technology transfer office manages the university’s intellectual property and the transfer of knowledge and technology to industry. From the intellectual-property side, WIPO lists these functions: promotion of intellectual-property awareness, management of disclosure, filing for protection, marketing and licensing, creation of spin-outs, maintenance of the assets, enforcement, revenue sharing, and management of conflicts of interest and commitment. Where income is a goal, some offices also manage consultancy, short courses, contract research, or laboratory services. The office is not always inside the university. WIPO also describes a wholly owned subsidiary, a private service provider, a consortium, and a government office serving that role. Common names on the same page include Technology Licensing Office, Technology Management Office, and Industry Liaisons Office.
Licensing and spin-out creation sit in that function list, which is why the diagram marks them as usually handled by the office. Consultancy and contract research sit in the optional list, which is why collaboration and consulting are marked as sometimes involving it. A graduate taking a job, and a paper that needs no rights review, usually involve none. Whether a manuscript first needs an invention disclosure or a confidentiality check is set by the institution’s policy and by any agreement on the research.
Who holds the rights decides which of those deals can be signed. WIPO states that ownership of a particular research outcome is read from three places: national intellectual-property law, the institution’s policy, and the agreements among the university, the creators, and any sponsors. Countries differ. WIPO’s examples of a professor’s privilege, where the researcher rather than the institution holds the rights, include Italy and Sweden. Its examples of institutional ownership include the United Kingdom and the United States, among others. A national statute can limit what an internal policy is allowed to change.
Students who are not employees are a separate case. WIPO states that most universities recognize a general principle: those students own the intellectual property in work they produce purely from lectures and teaching. Sharing or assignment comes up, on that page, for sponsored students, students on a sponsored project, students working with academic staff on the staff research, and students who make substantial use of university resources. Students are not automatically bound by the policy. WIPO states that the university usually needs the student’s express agreement before the research starts.
In the United States, one statute covers a defined slice of this. The 2024 edition of Title 35, Chapter 18, is popularly known as the Bayh-Dole Act. Section 201(i) defines a nonprofit organization to include universities and other institutions of higher education. Section 201(e) defines a subject invention as an invention of the contractor conceived or first actually reduced to practice in the performance of work under a funding agreement. Section 201(b) defines that agreement as a contract, grant, or cooperative agreement with a federal agency for experimental, developmental, or research work funded in whole or in part by the federal government. Section 202(a) says each nonprofit organization or small business firm may, within a reasonable time after the required disclosure, elect to retain title to any subject invention, unless the funding agreement provides otherwise in the cases the subsection lists. Those cases include a contractor not located in the United States, a contractor without a place of business in the United States, or a contractor controlled by a foreign government; exceptional circumstances the agency determines; specified intelligence activities; and certain Department of Energy naval nuclear facilities. Among the conditions in section 202(c), paragraph (4) requires a nonexclusive, nontransferable, irrevocable, paid-up license for the federal agency to practice the invention, or have it practiced, for or on behalf of the United States throughout the world, and paragraph (7) requires a nonprofit contractor to share royalties with the inventor. This is a rule for U.S. federal funding agreements. It does not assign every university result to the university. Implementing regulations after this 2024 edition were not reviewed for this page.
Outside work has its own rules. WIPO describes a conflict of commitment as outside activity, paid or unpaid, that interferes with the person’s duty to the institution, and a conflict of interest as a personal stake that can affect professional judgment. The examples on that page include a founder negotiating the license with the university, and a person who consults for a company deciding whether to assign an invention to the university or to that company. The policy, on WIPO’s account, discloses and manages those situations, and it identifies situations that are prohibited.
Why a license or spinout is not yet a product
An agreement moves rights, or it sets who will try to develop the result. It does not place a product on sale. WIPO states that many university results are no more than proofs of concept or laboratory-scale prototypes, and that they require further research and development before they can be commercialized. The same FAQ states that a good intellectual-property policy is not enough. At the market level the page lists three conditions no policy supplies: enough firms able to invest the money, time, and effort to turn the intellectual property into marketable products; a market ready to buy the product once it is developed; and absorption capacity to take in new knowledge and apply it.
The U.S. National Science Foundation’s I-Corps program is a seven-week experiential course. The program page describes it as training that prepares scientists and engineers to look beyond the laboratory at NSF-funded and other basic research that is ready to move toward commercialization. The national-teams page states that, through those seven weeks, teams engage prospective customers, partners, and others, and that a national team has to complete a minimum of 100 potential-customer interviews. The interviews test commercial potential. They are not a record that a product was sold.
If the result is a production method, commercial use still has to show up as work someone runs. What a job-linked record has to show on the factory floor is that later question. A license does not answer it.
A university news release is an announcement. A trade-press item about a funding round is an announcement. Neither one is a product in use. The library search for this page returned one such funding item. It is not used as evidence here.
How to tell whether research reached real-world use
Keep four layers separate. NIST Special Publication 1234, the April 2019 Return on Investment Green Paper, is a discussion document about federal laboratories. NIST states that the paper was not cleared through a formal interagency process and that its views are NIST’s. On printed page 107 it says that evaluating technology-transfer performance requires four examinations: the operational processes of the offices, the outputs of internal actions, the outcomes from external adoption of those outputs, and the broad impacts on industry, the economy, national security, and society. Using that chain for universities is an editorial adaptation. The paper does not set a university score.
WIPO’s typical benchmarks for a transfer office are counts of intellectual-property disclosures, applications, and grants, then licensing contracts, licensing revenue, start-up companies, and joint ventures. Those are output counts. AUTM’s FY2025 definitions, which participating institutions use when they report their own figures, split the same territory into narrower lines:
- An amendment is a written change to an existing license, such as duration, royalties, or added property. The instructions say an amendment does not count as a new license.
- A licensed technology is available when it is sold as a product to the public or placed into commercial use, as opposed to developmental use. The instructions point to royalties generated for the first time, or to licensee diligence reporting, as evidence that this happened in the survey year.
- License income received includes issue fees, option payments, annual minimums, running royalties, termination payments, cashed-in equity, and software or biological-material end-user fees of $1,000 or more. It excludes research funding, patent-expense reimbursement, a valuation of equity that was not cashed in, end-user fees under $1,000, and trademark royalties on university insignia.
- Adjusted gross income, in the survey, is license income received minus license income paid to other institutions. That subtraction avoids double-counting. It does not subtract costs.
- Legal-fee expenditures are external legal fees for patents and copyrights: prosecution, maintenance, interference, and minor litigation. Significant litigation, and the salaries of the office, are outside that line.
- A startup, in this survey, is a new company that depended on licensing the institution’s technology for its formation, and that was formed to develop the technology licensed in the reporting year. A company formed to develop a different technology is counted as a small company, not a startup. The definition does not require the company to have been created in the survey year.
AUTM’s public FY2025 survey page, under the U.S. key findings, states that nearly 2,000 license amendments were reported in 2025. The page does not print the exact integer or the number of responding institutions. An amendment count is an output. It is not a count of products in use. The same list prints percentage changes for disclosures, licenses, options, startup licenses, equity agreements, and patent filings. Those percentages are left unprinted here because the page does not print the counts underneath them.
The income identity in the definitions is:
Adjusted gross income = license income received - income paid to other institutions
Worked example: the public survey page does not print either dollar amount, so the cells stay blank. Filling them from a news release would invent a total.
| Line | What the FY2025 instructions put on the line | Amount |
|---|---|---|
| License income received | Fees, running royalties, and cashed-in equity, with the exclusions above | |
| Income paid to other institutions | Payments under inter-institutional agreements | |
| Adjusted gross income | License income received minus income paid to other institutions | |
| External legal fees | Patent and copyright fees as defined for question Q7 | |
| Office running costs | Not part of adjusted gross income, and not part of the legal-fee line |
A blank cell means the public page used for this article does not print the figure. It does not mean the figure is zero.
Copy the check below for one result. Leave the last column blank until the document is in the file.
| Question | What to write | Evidence and date |
|---|---|---|
| Route | One of the six routes, or unknown | |
| Rights holder | The party named by the law, the policy, or the agreement | |
| Agreement | Type and date | |
| Development after the agreement | Who is doing the next engineering, testing, or funding | |
| Available | Product sold to the public, or process in commercial use | |
| Impact | What changed, for whom, against what baseline, over what period |
Here is a hypothetical file built for this page. It holds a university news release about a license and a copy of a signed license. The release is an announcement. The license is an output under the table above. The available row and the impact row stay blank. Nothing in that file shows a sale or a commercial process.
Dividing one of these counts by another does not produce a success rate. AUTM’s instructions count licenses, amendments, available technologies, and startups as different observations. One technology can sit under more than one license. Outputs reported in one survey year can come from research done earlier. On printed page 33, the OECD states that contributions from basic research can take a long time to appear, which makes a single-year ratio a poor account of that work.
Rule of thumb: If the file holds a license, a new company, or a university announcement, and it holds no record of a sale or of commercial use, leave the use cell blank.
Where this does not hold
The six routes describe ways a result can travel. They do not state how often it arrives.
- Early results. When the result is a proof of concept or a laboratory-scale prototype, WIPO states that further research and development is still required. An agreement does not finish that work.
- No firm or no buyer. Without firms able to invest the money, time, and effort, or without a market ready to buy, WIPO’s market-level conditions are unmet. Licenses and spinouts can stall there.
- Field. A patent count is a weak proxy in fields where the OECD’s cited studies found other channels carrying more of the transfer, including computer sciences.
- Attribution. Informal channels are hard to count, and a hiring record or a citation does not by itself show that the university caused a later gain.
- Jurisdiction. Ownership, inventor rights, and government rights differ by country. The Bayh-Dole election described above is a U.S. funding-agreement rule.
- Disclosure before a patent filing. WIPO states that a public disclosure of the key features before filing can bar a patent in countries that apply a universal novelty requirement. The examples on that page include a journal, a poster, a talk, a website, and a conversation with someone who is not under a confidentiality duty. WIPO also states that many countries allow a grace period, usually between 6 and 12 months, and that the types of disclosure covered differ by country. This page does not give a filing date.
How we researched this
The sources are WIPO’s FAQ, agreements, and organizations pages; the OECD’s 2019 report, with the channel list and field comparison on printed pages 32 and 33; NIST Special Publication 1234, April 2019, printed page 107 for the four-part evaluation; AUTM’s FY2025 survey instructions and the public FY2025 survey page; NSF’s I-Corps page and the national-teams page; and the 2024 edition of Title 35, Chapter 18. Each page was fetched once on 18 October 2026. Library searches for technology-transfer documents, including searches limited to wipo.int, oecd.org, and autm.net, returned no usable text from those publishers. A general search returned unrelated pages, including packaging equipment and one trade-press item about a spin-out funding round. That item is not evidence of a product in use. A cached search-result list for “technology transfer” was empty. No new search-result data was purchased.
The year range WIPO gives for a transfer office covering its operating costs from its own income is unprinted here. That page cites no study and no sample for the range. Dollar totals for license income and legal fees are unprinted because the public survey page does not give them. No conversion rate, success rate, or time from license to product is stated. No fieldwork was done for this page.
The laboratory photograph and the three diagrams were already on the page. The captions describe those images: a laboratory bench with a prototype, the six-route diagram, the illustrative rights path, and the four-layer measurement diagram. Further explainers in this section are on the Progress page. Corrections and source practice are in the editorial policy.
