Kill criteria for an innovation portfolio

The portfolio review has twelve lines and eleven of them are still green. The one that slipped has a new name and the same sponsor. Nobody in the room is asked what would make the company stop.

Rita McGrath puts that stop at the centre of whether a corporate venturing programme returns any money. She argues that one of the most important factors in the financial returns is the ability to stop investing in projects that will not work out at all, and in projects that would be inconsequential to the parent even if they succeeded. The mistake she describes is a familiar one. Organisations take the logic of a well-understood, repeatable business and lay it on an uncertain new one. Once an idea has survived the approval process, its champions assume the work now runs through to a commercial launch.

I would break that assumption in the review. A project that sounded right on a napkin can still have been useful, and the direction can still be the wrong one to fund. Useful means the team learned something a later bet can use. The wrong direction means more money and more people will not turn it into a business the parent should own. McGrath offers the first six of a longer set of questions as a way to tell those apart, before next quarter's funding becomes a habit.

When the original goal has moved

Her first question is whether the team has learned that the original goals will be much harder or much more expensive to reach. She cites the old venture-capital joke that a win takes twice the time and double the investment you expected, and then she asks the serious version. Given what the work has taught you, are the original objectives still attainable? She offers the Drinkworks platform as a possible case, and she is careful about it. The parent companies have not said why it was discontinued. Cost and difficulty may have been the issue. The public reason can stay unknown. The review still needs the internal one in writing: what we now believe it will cost, set against what we approved.

The second question is time. With an uncertain project, she writes, it is almost impossible to forecast accurately what will happen when. Time passing can also make an idea that made sense at the start look worse. She uses Ikos, a startup that connected landlords and real estate owners. It grew quickly. Then the pandemic cut revenue, other funding fell through, and the company ran out of time. Co-founder Steve Welles said inventory was extremely tight, so revenue growth was limited. Continuing would not have been the business they had originally envisioned. They looked at options and decided to wrap it up. The sentence I would keep, for a bet inside a large company, is that one. If continuing means the thing in market is no longer the thing we approved, the review has to say so in the minute.

A portfolio that cannot answer those two questions is funding the original slide. Put the approved cost, the approved date and the current estimate on the same row. When either gap is large, continuing needs a new decision and a new number. Leaving the line green is a renewal nobody signed.

When the work grew and the resources did not

The third question is scope. The project has become much larger than anyone planned, and the resources on it have not changed. McGrath sees this often in digital work and in advisory offerings. People add features and ideas. Without a firm hand, the result is a large programme that is short of people and set up to fail. She treats scope creep, which she also calls feature bloat, as a reason major systems projects fail or land short of what they could have been.

Windows Vista is the case she uses. In trying to build a next generation of the operating system, Microsoft tried to please a vast ecosystem. Ben Fathi, an insider she quotes, called Windows a beast. So many people touched the project that the result was bound to be a long series of compromises, leaving no one particularly happy. The same observer described a different choice at Apple. Apple killed its server and storage efforts, dropped support for earlier PowerPC machines, and built iOS as its own system, separate from MacOS, when the form factor changed. In his account, Windows tried to be all things to all people.

I would ask the sponsor to name one thing the bet will no longer do. If they have no answer, the scope question is already settled, and the resources will stay where they are. Dropping a feature is a smaller stop than stopping the bet, and it is still a stop. A review that only adds scope has skipped the question.

When the champion outruns the people who know

The fourth question is the one a portfolio review dodges. People with expertise in the organisation have misgivings, and the champion is determined to press on. McGrath's point is sharp. Senior support, which a venture needs, can also keep a dead project alive long after the people close to the work can see it. Iridium is the case she takes from Syd Finkelstein. Robert Galvin, then Motorola's chairman, gave the project manager, Bertiger, approval to proceed. Galvin, and later his son and successor Christopher Galvin, treated the low-earth-orbit satellite network as a symbol of Motorola's technical prowess. The engineering challenge motivated the team. Service began in 1998, at a total cost of over $5 billion. She names Amazon's Fire phone and Zillow's house-flipping venture as further examples of the same pattern: a champion, and a bet that kept its backing.

The question for a corporate portfolio is who may put the misgiving on the page. If the only person who can move the status is the sponsor, expertise in the room is decoration. I would give the review a column for who disagrees and what they know, and I would read it before the champion presents. A bet can survive disagreement. It should have to survive disagreement that has been written down.

When the market or the partners moved

The fifth question is about competitors who were not in the original picture. Sometimes, with no particular failure of effort, another solution makes yours less convenient, more expensive or impractical. SAP's Business ByDesign, aimed at the middle market, ran into on-demand products from Salesforce and Workday and could not compete on the terms SAP had set. The portfolio question is whether a rival's offer has changed the reason this bet existed. If a customer can already get the job done another way, more build time is a way of postponing the stop.

The sixth question is the ecosystem around the offer. McGrath notes that a complete offer for a customer almost always depends on partners, a point she credits to Ron Adner's book Winning the Right Game. American Express backed Plenti, a coalition of merchants that would share points and data. A shopper might buy clothes at Macy's and put the points toward fuel at Exxon and Mobil. The programme lacked common goals and cooperation among the members. Defections followed, and the concept failed. A corporate bet that assumes a partner will show up, integrate and stay is carrying an assumption the review should test every quarter. If the partner list still matches the approval slide, and none of those partners are live, the ecosystem has not evolved. That is a condition for stopping, including in the week when the demo still looks fine.

There is a prior cut that sits above the six questions, and it comes from her opening. Some projects would be inconsequential to the parent even if they worked. A small win that cannot move a number the company already reports is a different failure from a project that collapsed. I would ask, before the six questions, what changes in the parent if this bet succeeds on its own terms. If the honest answer is very little, the courageous stop is available now, while the team is small and the learning can move.

How a review uses the questions

I would print the six questions as the agenda, in this order, and I would keep status colours off the page until each bet has an answer.

1. Are the original goals now much harder or much more expensive, and are they still attainable with what we have learned?
2. Will this take much more time, and does the extra time change the reason we started?
3. Has the scope grown while the people and the money have stayed the same?
4. Are experts in the building unconvinced, while the champion is still pushing?
5. Has a competitor appeared who makes this offer less convenient, more costly or impractical?
6. Have the partners this offer needs failed to show up, or failed to stay?

A single yes still leaves a decision to make. McGrath's standard is whether you are about to throw good money and good people after a pending disaster, and she calls the decision to stop, in that case, a courageous move. A yes means the bet needs a fresh decision, with the new fact in the minute, and with a date by which you will stop if the fact has not changed. Two yeses, or a yes the sponsor cannot answer, and I would stop the funding that month. The people can move to a bet that still matches what was approved.

The returns she is talking about depend on those stops. A page of green lights is a page that has not asked. Innovation leaders can run the questions on the live bets this month, with no more bets in the room than the meeting can actually discuss. Write each answer in a sentence, with the name of the person who holds the evidence. Mark each bet continue, continue only if a named fact changes by a named date, or stop. Move the people off a stopped bet in the same week, so the stop is real and the learning stays with the team that did the work.

A lean startup and product sprint can put one of those bets through the questions with the sponsor and the operator together. The review after that week still belongs to the person who can turn the funding off.