When a company splits in two, one side usually ends up renting what it used to own. Which side, and on what terms, is decided in the separation workstream that gets the least board attention: the allocation of patents, brands, data and know-how. That allocation then sets the limits of what either company can become.

A BCG analysis published recently makes the business case for treating a carve-out as a redesign of both companies, not just the one being sold. Its data says carve-outs paired with a full transformation programme outperformed conventional separations by 20.3 percentage points of total shareholder return over two years. Its warning is that the continuing company usually gets a clean-up exercise while the new company gets the strategy. The line to keep: “Rationalization optimizes the inherited model; it does not determine whether that model remains the right one.”

The authors note that transitional services agreement design can constrain future systems choices. The same is true, with more force, for the IP split, which the article does not discuss.

The carve-out IP ownership test

Every shared asset in a separation gets one of three fates: it goes with the new company, it stays with the continuing company, or it is licensed across. The third option is the easy one at the deal table and the costly one afterwards. Three consequences follow.

– A licence fixes the strategy of the licensee. A new company that holds its core technology under licence from its former parent cannot enter fields, territories or product lines the licence excludes, and cannot sell or raise capital on a platform it does not own. The parent’s constraints become its own. Who Owns the Design? Why IP Ownership Is often a Condition of Winning the Contract shows how a licence signed years earlier decides which contracts a company may bid for.

– A licence fixes the economics of the licensor. A continuing company that keeps the patents but licenses them to a business that now competes with it has priced its own competitor’s inputs. Those terms will be tested the first time either side changes direction, which is the point made in Who sets the price of your IP? Why the licence decides, not the patent.

– Data and know-how do not split cleanly. Customer data, process know-how and trained models sit in shared systems, and the TSA that keeps them running keeps them shared. Whichever company is meant to own them needs ownership documented, access controlled and confidentiality running the other way before Day 1, or the trade secret is lost in the handover.

The practical move is to allocate IP with the strategy, not after it. For each business, decide what it must own outright to pursue its planned position, and assign those rights before the TSA and the systems choices lock the arrangement in. The question in Does your AI strategy move you, or just add a feature? The IP test for platform partnerships applies without change: after this move, what will each company control that it did not control before.

A carve-out that leaves one company dependent on the other has not created two businesses. It has created a licensor and a licensee, and the transformation each is promised will stop at the edge of the licence.

Read the article: Better Apart: When a Carve-Out Transforms Both Companies, Jared Feiger, Simon Weinstein, Georg Keienburg, Nidhi Meppadan and Dominik Degen, BCG.