[Open question] Do you actually need to innovate?

TL; DR

Arguably, the most basic question about innovation is whether to do it at all.

But frustratingly, I can't find any good standard to indicate whether, when, where, and to what degree a company must innovate.

And I did look, both over time and as part of my ongoing research into innovation governance. This question is utterly core to the governance topic of what innovation "mandate" or "agenda" to create and how to ensure it's a "must do."

I'll share below what I've found so far.

But be warned: The only standards I have found so far are either bad on their face, nice in theory but hard to use in practice, seriously limited to certain situations, or promising but somehow incomplete.

The best standard I have found so far is your "business's learning curve."

Call to action

Maybe I just haven't looked in the right spots. Maybe there's an easy answer.

Have you found any good standards that might trigger a need for innovation?

If so, let me know!

Caveat: What I mean by innovation here

Compared to other posts, I'm being somewhat generic and inclusive in my use of the term "innovation" here, covering everything from incremental to transformational work.

It's just a matter of "beggars can't be choosers." If we (IMHO) lack good standards overall, then surely the luxury of standards for specific innovation types is equally out of reach. We can come back to those another time.

Reminder: Why the question matters at all

It takes enormous effort to create a business that works. "Innovating" is a way of saying that you are changing that business system. Sure, you might things better. But you also might make things worse, on top of not getting full return on prior effort and investments that led to your current business.

As a result, not every company should innovate–definitely not at every stage of its life.

Even if you might innovate, doing so might not be your best option. Every organization's budget is ultimately limited. Whatever money and effort you might invest in innovation work, you could also invest to improve operations, staffing, culture, the company's financial basis, or more. Or you might create new capabilities that don't fit the more narrow definitions of "innovation," from advanced manufacturing to analytics and AI.

And even beyond all that, there are companies that can't afford to innovate, typically those just scraping by or trying to pull out of a death spiral.

Bottom line: Not everyone can or should innovate. Innovation is not a default choice, even for those who might pursue it.

But if innovation is not for everyone, then we ought to have some sane standard for identifying those companies (and situations) for which innovation work is just what the doctor ordered.

Here are standards that I have seen:

Outcome metrics make for terrible innovation triggers

An all too common reason for initiating innovation work is a problem with outcome metrics ("sales and profit are down!").

Nuances aside, outcome metrics make for terrible reasons to innovate (on their own). Results might be bad for any number of reasons for which innovation is not the solution: bad execution, culture, luck, and more can all contribute.

Beyond that, innovation work can take time to yield results. So even if it's a helpful solution to your problems, it might be too late to pursue it when results are already suffering. Leadership might not have the patience to wait for results.

Operational metrics are messy not available for all innovation types

Operational metrics are in common use to measure innovation in well-defined spaces, where they work decently. For example, in mobility, automotive journalists might compare the fuel efficiency of cars from different manufacturers and use them to judge the need for more innovation in one or another company: If company A's fuel economy improves at a slower rate than it does at company B or C, then more innovation might be called for.

But, sticking with the example of cars for a moment, what happens if a manufacturer only makes incremental improvements for several years, followed by major updates when a model generation reaches the end of its life? In that case, the company already is innovating well. It is just choosing not to show its hand publicly yet. And such batching shows up far and wide. It's not even just a matter of what innovations are released to the public. Breakthroughs also don't happen on a linear schedule. An R&D team might reach insights more quickly at one time and more slowly another time. We can certainly use such operational metrics for their main purpose, namely to fine tune our operations, including innovation operations. But they are not reliable triggers to alert us of a need to innovate.

At other times, there may either not exist good metrics for innovation, or they may take real skill to identify. For example, one of the more unorthodox forms of innovation involved a company here in Minnesota: Medtronic, a world-leading medical device maker, came up with a way to "reverse outsource." In 2014, they moved their official headquarters to Ireland, saving gobs of money in taxes along the way. Very innovative. Very unusual. Very hard to find operational metrics that would have foreshadowed such an opportunity, even to someone like me, who used to work in corporate finance myself.

Operational metrics then also cause problems as clear predictors for who should innovate when.

Strategic metrics may get closest but are hard to measure precisely

Strategic metrics may hold more promise. By their nature, they abstract away from everyday operations and finances/ results/ outcomes. But in exchange, they also offer more simplicity and, hopefully, clarity.

One strategic metric I encountered recently feels particularly promising; both specific and comprehensive. But I warn you up-front that it's someone fuzzy, not precise.

Give me a second to set it up and focus on the true reason for innovation.

If we simplify things to the conceptual level I have used before, there are only three things with companies:

• We can create organizations (startups).
• We can run them (operations).
• Or we can change them (that’s us, in innovation land).

Now let's apply those categories here: We don't need to innovate while we are already creating something new. We don't need to innovate (much) while the business we just created works well and we just need to run this business. We only need to innovate when it's time to change a business system.

The key question then becomes: When do we need to change a business system?

Simplistically, it's when the business has run its course. When it can't grow much more and is more likely to stall or shrink.

A useful trigger for innovation then is (some variation of) the question:

How close to the end of its growth and profitability potential is our current business?

If we look at it that way, there is still a grey zone. Some companies might want to start innovating when the business has reached 70% of its way to the end of its potential, others at 80%, and others even later. But the issue becomes graspable.

And of course, as I already warned you, a business's "potential" is a completely hypothetical concept. Measuring it more practically might be beastly! Maybe it's just me. But I have only found very crude approximations for such "remaining business potential."

One way of approximating the trigger for innovation

As already mentioned, I don't have a great answer here. That's why I'm sharing this quandary with you after all, in the hopes that I might inspire you to find one for your own needs. Who knows. You might already have one!

But I'm not a total lost cause. I can at least offer an initial approximation that narrows down the issue a bit.

One of the world's most effective innovators is the U.S. government's DARPA (Defense Advanced Research Projects Agency), originally known just as ARPA. Following its long list of success since 1958, e.g., including foundational work that led to weather satellites, GPS, drones, stealth technology, voice interfaces, the personal computer, the internet, and some COVID-19 vaccines, the American government in 2009 created a civilian equivalent in the Department of Energy, known as ARPA-E.

Finally getting back to the point, one of ARPA-E's recent funding announcements points us to a neat construct that can more specifically point out the need for innovation. It's still somewhat abstract, but slightly less so.

The team at ARPA-E points to the importance of "learning curves" as a litmus test in innovation.

Applying the idea to us first before describing it in a bit more detail, we can basically watch the "learning curve" on which a business operates. If we observe the metrics that ARPA-E suggests, we can infer that the curve is starting to flatten out–hence the need for substantial new innovation.

It's the observability of those metrics as proxies for the learning curve itself that make me excited about this as one way of showing the need for innovation.

Let's look at what ARPA-E suggests. I will re-word and structure the quote a bit, to focus on our needs and abstract away from the specific criteria for their funding opportunity in the energy space:

"Existing ... technologies generally progress on established “learning curves” where refinements to a technology and the economies of scale that accrue as manufacturing and distribution develop drive improvements to the cost/performance metric in a gradual fashion.

This [represents the] continual improvement of a technology[,] is important to its increased commercial deployment and is appropriately the focus of the private sector ....

In contrast, ... transformative research that has the potential to create fundamentally new learning curves. [Those] projects typically start with cost/performance estimates well above the level of an incumbent technology. Given the high risk inherent in these projects, many will fail to progress, but some may succeed in generating a new learning curve with a projected cost/performance metric that is significantly better than that of the incumbent technology.

[Some transformative research has] the potential to be disruptive in the marketplace. The mere creation of a new learning curve does not ensure market penetration. Rather, the ultimate value of a technology is determined by the marketplace, and impactful technologies ultimately become disruptive – that is, they are widely adopted and displace existing technologies from the marketplace or create entirely new markets. ... [D]efinitive proof of market disruption takes time .... [Those solutions with] clear disruptive potential, e.g., [demonstrate it by their] capability for manufacturing at competitive cost and deployment at scale."

Let's boil it down to just the key metrics that they call out:

As we approach the end of a learning curve, we may see:

• Technological refinements
• Economies of scale in manufacturing and distribution
• Gradual improvements to the cost/ performance metric.

By contrast, the beginning of new learning curves sees metrics like:

• Longer timeframes
• Approach toward value of the innovation by the marketplace (what we'd call "product/ user fit" and "product/ market fit")
• Approach toward a capability for manufacturing at competitive cost
• Approach toward a capability for deployment at scale

Of course, the team at ARPA-E writes for a specific space. So these metrics are more illustrative. Only the idea of the "learning curve" carries over universally. You may need to find other operational, financial, or strategic metrics yourself that are appropriate in your space for identifying the beginning and end of learning curves.

Nonetheless, it feels like a start.

My overall, point-in-time take

Acknowledging that this is merely a work-in-progress solution and that you well may have better ones, I'll propose this as a thesis for conversation:

Why is that not enough?

Unfortunately, a focus on "learning curves" and the operational metrics that tell us where we are is not enough.

• It's not specific enough, as already mentioned: At just what point should those end-of-learning-curve metrics trigger a need for innovation? And how might some poor team member practically measure the state of those metrics? It feels intellectually right but practically not actionable yet.
• It includes no forcing function: As you have likely seen in your own org, just because an org should innovate doesn't mean that it will do so. That's what my whole emphasis of "must-do" purposes is about after all. My hunch is that such a forcing function must come from outside the org, where even your CEO and Board can't just deny it or decree it to be unimportant. Given the human bias for loss aversion, such external forcing functions are more likely to work if they show how we are about to lose business rather than showing how much more we can gain. But I already said that financial metrics like sales and profit aren't the best. So the question is what kind of metric could serve this role.
  My hunch for a likely candidate is that we might use the concept of "unbundling __." But that's just a hunch so far. I can't point to concrete research proving it. What I mean by "unbundling __" is that market research firms have created many visuals that show how startups are increasingly able to do what certain incumbents do today. Those visuals are typically called "Unbundling [company x]." (E.g., see here for "Unbundling Meta" by CB Insights.) I believe that such a concept might work because it bursts the bubble that we as incumbents are somehow "better" or "differentiated." It points to us being at direct risk of losing all our business. Showing such "possible" loss might be even better than showing "actualized" loss of sales and profit because it lets people imagine a worst-case scenario. Their own imagination will add more urgency to the problem.
  Might one consider such an approach "manipulative?" Maybe, but I don't think so. All "change management" works with human tendencies and fallacies. But if the research and recommendations are honest, then one is just making a good, persuasive case. The only thing we must watch out for is that our case is actually as good as we think and don't fall prey to overestimating the quality of our work ourselves.

So an updated recommendation might look as follows, again with the caveat that this update is merely a hypothesis, not confirmed certainty:

Just a reminder: Help me out!

I'd love to know your take.

As mentioned above, it'd be awesome to hear:

• Found standards that might trigger a need for innovation? Please LMK!
• Got ideas on how to improve what I have found so far? Please send them my way too.
• Want to be part of finding answers? Please share!

Related post-publication updates

Ash Maurya has pointed out that "must-do" purposes typically follow a "switching trigger:

"Instead of guessing at a bunch of attributes, start with the one distinguishing trait that all early adopters have.

Can you guess what that is?

Answer: A switching trigger.

All early adopters should have taken action toward solving a problem or getting a job done. That's their tell.

A switching trigger (or inciting action) changes something in their world (causing an old way to break) and prompts them to look for a new way.

This can be caused by

a bad experience with an existing alternative,

a change in circumstance,

an awareness event.

In the mattress example above, “2 am when people can’t sleep” is a bad experience switching trigger."

-- Ash Maurya

Dan Toma has pointed out that "must-do" purposes are inherently bounded by/ relative to "how much we are willing to invest," not un-bound.

"Experience has shown us that, when it comes to setting goals for innovation, leaders are typically interested in finding a concrete answer to one of two questions:

How much do they need to invest given what they want to accomplish with innovation?

What is a realistic goal for their innovation efforts given how much they are prepared to invest?

We usually refer to the first question as the Future-to-Present way of setting innovation goals, and to the second question as the Present-to-Future way. One way is not superior to the other and every leadership team picks the one that best suits their company’s context. "

-- Dan Toma

[Somewhere that I can't find right now, it was pointed out that "must-do" problems are not binary but occur on a range.]

T.I.S.C.

Further reading

ARPA-E overall description of the SPARKS initiative (Accessed Feb. 27, 2024)

Specific ARPA-E concept papers for the SPARKS initiative, which include the quoted section above (Accessed Feb. 27, 2024)

ARPA-E Wikipedia page

DARPA Wikipedia page