Innovation.

The framework

1. 01

   Inbound is the make-or-buy call with a search in front of it.

   Bringing outside knowledge in — licensing, partnering, buying a component instead of writing one — is the flow every company means when it calls itself open, and its hard part, what to make and what to buy, is the decision you worked on Day 83. What that decision rarely gets is a search in front of it. Ralph Katz and Thomas Allen showed why in 1982: across fifty R&D groups, performance rose and held for the first few years, then declined once average team tenure passed about five years, tracking a fall in how much those groups talked to anyone outside themselves. So make the search procedural — above a size threshold you set, no build starts without one day, one named person and a paragraph on who else has solved this.

2. 02

   What are you holding that you're never going to ship?

   Run the audit once, properly: the shelved project, the internal tool three other firms would pay for, the dataset, the process, the brand you stopped using, the patent renewing quietly every year, the research that got as far as a working prototype. Write them all down in one list without arguing about any of them yet. Most businesses have never made this list, which is why they've no idea what it's worth — and why the things on it keep drawing renewal fees and management attention while returning nothing. The list is the whole exercise. The decisions are easy once it exists.

3. 03

   Two questions decide every line, and only one combination pays.

   Against each item, answer two things. Will we ever commercialise this ourselves, honestly, by a date we'd be willing to write down? And is it worth something to a business built differently from ours — different scale, different market, different cost base? Where both answers are yes, get on with it. Where both are no, delete the line and stop paying to store it. The combination that pays is no and yes: something you'll never take to market that somebody else would. That isn't a dead asset, it's an unsold one, and Xerox's spin-outs were exactly that — value its own model couldn't route to a customer.

4. 04

   Pick the route out, and put a date on one line.

   There are four ways a line leaves the building: license it and keep the asset, sell it outright, spin it out with somebody who'll run it, or publish it because a standard you shaped is worth more to you than a secret nobody's using. Which one fits depends on how much of your business goes with it and how much attention you've got left to give it. What kills these deals isn't valuation, it's that nobody owns them — so put one name and one date against a single line this quarter and let the rest wait. One completed exit teaches you more about what the shelf is worth than a year of listing it.