Editor’s note: This is an edited version of a post that previously appeared on Finiculture, a blog on fintech.

Let’s paint some broad strokes: a startup throws mud at the wall to find out what sticks, with a 0.2% success rate or less. A corporation makes more planned, but slower, moves and hits around a 12.5% success rate, according to a Harvard Business Review article.

The danger of large companies trying to innovate like a startup is how their collaborations represent the worst of both worlds: throwing mud at the wall, slowly and expensively.

As corporations reinvent their innovation processes, they have lots to learn from startups. However, corporations like large banks must realize that the lean startup model — a method for getting out a minimum-feature product quickly in order to test its reception — wasn’t designed for their business context.

Airbnb's decision to hire professional photographers to shoot listings was sparked after the cofounders had visited rentals in person to discover the need for better photo quality. But such a process of experimentation may not work for banks' innovation efforts. Bloomberg News

Corporations have different goals (serve a mass market, rather than a niche), different risks (a reputable brand, rather than the flexibility to start over), different constraints (they answer to regulators and owners) and of course, an “unfair advantage” in that they have resources to put to use — none of which is suited to a lean startup model. A corporation’s institutional nature — which makes decision-making slow and risk aversion high — busts the idea of lean.

We need to re-address how corporations are using the startup model for their own innovation efforts. As the last few years have seen inadequate responses from different types of accelerators (an outside entity that assists the corporate’s startup efforts) and incubators or innovation labs (that are owned by the corporate), we are re-entering testing mode with corporate innovation once again. Now is the time to create an optimal new approach.

As with each in-vogue wave, there is preaching and swallowing without really understanding what something is, when to use it and how to really do it. In the application of lean startup, the emphasis has been on “build, measure, learn,” and this mantra has been used as justification to start with building an un-validated hypothesis (read: gut-feeling ideas not based on any customer insight, aka mud). The focus is on speed while customer research and learning is often lost from sight.

But this approach does not ensure success in the least.

When we start with build, we start with what we already know, and refine from there — but this approach restricts us to a narrow set of opportunities that are not thought through. It’s the reason there are thousands of startups that help startups start up — they solve the problems in front of their nose. It’s the reason that the CES tradeshow is filled with drones, robots and wearable devices — fascinating technologies, but they lack indication of what real problem they are solving. The trouble is compounded by the view that developers and entrepreneurs think in a certain way and incorrectly assume that their users will think and feel like them.

So how can we augment lean startup to overcome this shortcoming? Immersive research is the best way to uncover non-obvious, highly important customer needs. One of the classic lean startup success stories is Airbnb’s experiment to hire professional photographers to shoot each listing, resulting in a twofold to threefold increase in bookings. But what precipitated the experiment? During Airbnb’s infancy, the technology company’s cofounders travelled to New York City, met every single host, lived with them and wrote their first reviews. The cofounders’ aha moment was seeing the mismatch of grainy photos compared to the real home. This insight gave life to what was, on paper, a mad experiment to run. But improving photo quality helped. As a human-centered designer, Airbnb cofounder Brian Chesky lives this philosophy today (with guests on his couch every night). It’s this customer immersion that informs the lean experiments that have made Airbnb so different and successful.

Here’s where large corporations – with their institutional approach – can get in trouble trying to use a lean startup-like model on it's own.

Lean startup works just fine on its own as an ecosystem. As long as you have thousands and thousands of startups throwing mud at the wall cheaply, a tiny fraction of them will be able to pivot their way into a non-obvious, important solution. But isolating that effort to one company, blindly testing hypotheses, is not efficient. For a corporation, throwing out new propositions in the dark is a very expensive way to innovate. Darwinism doesn’t work for their creationism. For a lean startup to be effective on a corporate scale, a large company initially needs to optimize its understanding of what a customer wants and can use, before it develops a product.

A startup’s customer research can leave a lot to be desired, giving corporations an opening to claim customer testing as an advantage.

Startups will typically find a customer insight is too fragile. Taking stock of this fragility can take a while. First, you interview friends and family, or perhaps a wider but limited universe in your immediate ecosystem, and everyone (who is just like you) congratulates you on your great idea. The investors love the idea because they’re in the same bubble too. The first sales look good because they’re done in the same way — you start by selling to the people you know, and the people they know.

Indeed, it can take a while for someone to realize how fragile this process is. We are now seeing signs this fragility is being recognized. More than a fair share of initial fintech startups failed to reach escape velocity, for example. Many smart home applications are built for tech enthusiasts, but fail to solve a real need at scale. Grounding and grounded customer insights are crucial and required. We need to not just get out of the building, we need to get out of the bubble.

The way the startup ecosystem is innovating has made it very difficult to design a corporate-startup collaboration optimized for success. In one’s echo chamber, it’s hard to tell if a startup is a “success” that one wants to collaborate with, and if that “success” is a false positive or a false negative.

Corporations (particularly banks) are scrutinizing the agility of the startup world with envy. They want to mash startups into their business units to overcome their corporation’s inertia. Under pressure, unit heads in need of a hero can snag the latest fintech poster child, while said poster child is confident of its impending success based on the validation derived from the trust its investors have thrusted onto it.

Startups approach the challenge without enough context — but instead of building context via insight, they jump right into lean development — one of their raison d’etre. They believe, as a startup, their job is to build at pace. While that might be okay in a startup ecosystem, it is an expensive and risky approach for a company with an existing customer base, history and brand.

Corporations can take advantage of the huge advances in lean innovation – without innovating like a startup. To build a new process for tomorrow’s innovation, we must consider how the whole innovation value chain is being reinvented.

Based on many years of experience building products and services with corporates and startups from across industries, we believe this new process will marry the best of lean startup thinking with the best of customer insights to produce a different kind of venture building framework. Building products and services needs to start from insights. We need to get out of the bubble. We need to build in agile teams with subject matter and industry experts. We need to hardcode customer centricity into the innovation and venture building teams from the start.

Pascal Bouvier

Pascal Bouvier

Pascal Bouvier is a venture partner with Santander Innoventures. He writes the FiniCulture blog. Contact him on Twitter @pascalbouvier.

BankThink submission guidelines

BankThink is American Banker's platform for informed opinion about the ideas, trends and events reshaping financial services. View our detailed submission criteria and instructions.
Aldo de Jong

Aldo de Jong

Aldo de Jong is co-founder of the innovation firm Claro Partners. He can be reached on Twitter @aldodj.

BankThink submission guidelines

BankThink is American Banker's platform for informed opinion about the ideas, trends and events reshaping financial services. View our detailed submission criteria and instructions.
Harry Wilson

Harry Wilson

Harry Wilson is a senior associate at the innovation firm Claro Partners. He can be reached on Twitter @_harrywilson.

BankThink submission guidelines

BankThink is American Banker's platform for informed opinion about the ideas, trends and events reshaping financial services. View our detailed submission criteria and instructions.