Managing Innovation: Pipeline Productivity and the False Security of the Portfolio Effect

“Put all your eggs in one basket, and watch that basket!”

?????????????? - Andrew Carnegie

I’ve been (re)thinking about productivity and product development lately and decided to revisit something I mused about prior here.? A basic equation for looking at new product pipeline is shown below.

Basically, to manage a portfolio of new product development you need to measure the number of projects in pipeline at any stage (WIP), multiply by historical attrition rate (90%? See below) multiplied by the expected average value of product targeted.? This can be tricky but a standard NPV like approach can be taken.? [A caution here is to think connected adjacencies; a $B market rarely appears spontaneously. More often they are a chain of small seemingly unconnected markets linked by a shared unmet need]. ?In any case this idea/concept/product pipeline is then divided by your cycle time from ideation to implementation multiplied by your average cost to do so to get to P.

One of the key aspects about Open Innovation is that you should embrace and include lots of external partnerships and investments to increase the diversity and number of your WIP.? All things being equal, this should increase your P.? However, this does so at the expense of not just money but precious management time and talent that is now spread over an increasing WIP portfolio. ?No surprise then that we see companies investing in Innovation Management.? So WIP needs to increase faster than CT x C, and while p(TS) goes down due to increasing volume and risk being taken, the hope is V goes up to compensate.

But does this actually work?? If so, how? I propose we take a look at Venture Capital as our analogue. The rationale will be made clear down below.

?A 2012 WSJ article first claimed that the average venture capital (VC) investment portfolio has a 75% failure rate.? Failure here is defined as failing to return capital back to investors.?

Another, later, theoretical argument [see here], calculates that failure rate to be as high as 95%.? This is interesting in that it highlights V, you need one really big win to justify all your failures.? Which in turn explains the chase for unicorns.

However, this all begs the question; if at least 75% of a VC portfolio fails, and perhaps as much as 95%, what is the failure rate of the VC system itself, which is a portfolio of portfolios?? And how does this apply to corporate venture capital (CVC)?? The traditional CVC playbook is to co-invest with traditional VC on behalf of the parent corporate- either for R&D and new product development learnings, and/or to create a financial return for the corporate parent.? Could CVC also suffer from the same rates of failure?

So how might we look at this with real data?

Let’s first look at data for new U.S. VC firms raising money for the first time.

We observe a range of 190-436 new firms raising first rounds per year with a high of 436 in 2021, a mean of 360 over 10 years (skewed by 2021-2022), and a median of 240 excluding the partial 2024 count.?

How does this help us understand the performance of this portfolio of portfolios? Actual yearly numbers of failed or shuttered VC firms are hard to come by outside of a paywall.? Not even my AI enhanced search can find a single source of truth.? But perhaps we can estimate and create a proxy.

This is not a total count of VC funds but a representative cohort of those trying to raise money each year. ?Graph shows over the past two-three years that even experienced firms [defined as more than 4 funds launched] can struggle at the same rate as emerging firms [defined as <4 funds launched per NVCA-Pitchbook].? The run rate forecasted for 2024 seems to indicate a return to the mean.

We see a decline from 843 to 457 at end of 2023 with projection to 199 by end of 2024 for a 77% projected peak to trough decline through 2024.? Of course, many funds could have been flush with nondeployed capital from prior years, or representative of a post-Covid gold rush, but also, one needs to have a successful portfolio vintage to even contemplate entering into new fundraising.?

?

Graph of NTI vs CVC from Table above.? Count of investments by NTI and CVC by year.? We see a 45% decline in CVC investments projected by end 2024 vs the post Covid high in 2022.

According to the NVCA, at end of 2023 there were 3417 VC firms in the U.S. Below is a rough graph of estimated US VC firms by year. ?

?

?In the period between 1995 and 2020, there was an average of 708 U.S. firms in existence at any given time.? If per our above estimations 240 firms are raising money for the first time, then perhaps an average 20-30% of VC firms are churning.? That is, assuming a usual landscape of 700 U.S. VC firms perhaps 20% first time fundraisers are replacing those that are exiting [not in a good way]. As we see, the post-Covid period saw an extraordinary rise in VC firm creation.? But what we should expect to see then is a massive wave of firm shuttering in the next 7-8 years as the market returns to its mean, and along with them, their investment portfolios. ?

What is a corporate VC and Innovation person to do? If the average VC investment portfolio is only 10-25% successful in returning sufficient capital to limited partners to warrant another fund raise (and continued survival), and an average 10 year vintage fund is failing at a 20-30% annual rate; then following the dismal results of this portfolio of portfolios and applying it to your own singular portfolio is more fraught with potential failure than most realize.

The key I think, for heads of R&D, Strategy, and Innovation to think about, is to go back to our equation and focus on 3 things:

1.??????? Manage your innovation process such that decisions are made more quickly, robustly.? Kill early, kill often, as they say.? Of course, this is why stage-gating in innovation funnels exist. But those of us who work with these know that the gates get fuzzy over time; elongating CT and increasing C.

2.??????? The counterbalance to this is to focus on V, the value of any idea.? Key to establishing V is the range of potential scenarios your new idea can survive and thrive in plus the number of plausible potential adjacencies.?

3.??????? The big one to focus on in my mind is p(TS).? To find ways to increase probability of success given technology and market uncertainties.? One way I believe is to create a systems-based way of thinking about future scenarios that both unlocks alternative possibilities and derisks the knowns by enumerating and quantifying where possible all assumptions. Building a multi-lens scenario generation machine is one way to both create more robust ideations and the means to more quantitatively and systematically create and select ideas and to evaluate them as they become projects, selectively pruning early and often to select for those with greatest potential V.? This is important with long development cycles as technologies, markets, and consumers change. This also has the benefit of collapsing the funnel and many stage-gate processes with their sequence and cohort bias and enables far more more rapid portfolio evaluation with the same objective lenses.

A prototypical innovation funnel. Instead of stage-gating, a created evaluation engine is used to ideate, create, and evaluate ideas and product concepts as they evolve and develop-combining the virtual with the physical.??

Fewer shots on goal, greater concentration of time and talent, and constant testing of scenario assumptions is the better way to select for V and raise your P.