Back in 2014, retail labs were all the rage. It was hard to look around and find a retailer without a lab. New York Fashion Tech Lab, for example, a collaboration between the Partnership Fund for New York City, Springboard Enterprises and major fashion retailers offered selected startups mentorships with Ralph Lauren, J.Crew, Kate Spade and the Estée Lauder Company.

Fast forward to 2017, and all of these companies bar Estée Lauder are in trouble. They are marred by the leadership turmoil, sales and foot traffic decline, store closings and employee layoffs. Their troubles are easy to trace to the lack of innovation. Herein lies the conundrum of the retail lab, forcing upon us the question of innovation as the radical transformation of business models versus incremental improvement in product and services. The former is a strategic mandate, the latter a PR pitch.

The most successful retail labs are companies that didn’t exist give years ago. They swiftly introduced and tested new business models, new distribution and supply management practices, a novel go-to-market strategy and audience building tactics.

Sure, they do not need to deal with the muscle memory of a large, legacy retail organization. But they do not focus either on making small tweaks to protect and preserve the retail process that exist in their market.

Neiman Marcus’ most successful “retail” innovation is to introduce in-store smartphone charging stations. While desirable, feasible and viable, it won’t help Neiman Marcus reverse its shrinking footprint. Walmart Innovation Lab’s site hasn’t been updated since the hackaton in 2016. Instead, Walmart went on a shopping spree among retail startups.

Modern retail companies start from the self-disruption premise. They know that if they themselves do not constantly cannibalize their own business, someone else will.

This cannibalize-versus-protect model represents a paradigm shift in retail. It is a fundamentally divergent starting point that accounts for dissonance between the players in the modern retail market. It is a difference between asking “what does the store of the future look like?” and “do we even need a store?”

Here is the reason why most retail innovation labs fail. They are organized, resourced and governed to protect incumbents’ existing business models. The use of data, technology, prototyping, consumer insight or distribution are all filtered through the premise that their job is to make a faster horse — and not an entire new way to move.

Nordstrom folded its innovation lab into Customer Excellence Center (CEC) back in 2015. Prior to that, however, it created an iPad app for sunglass shopping. The team did everything that a lab does: went through the sprints, rapidly prototyped and collected feedback. The project flopped because the critical link in the in-store shopping process — sales managers — didn’t offer it to customers. It would have made them obsolete.

In contrast, Everlane and Ministry of Supply are set up as R&D departments, meant to continuously challenge these companies’ business models. What they lack in scale, they make up in having small groups of fans who beta-test their experimental products. Incumbents like Nike, Adidas and Lululemon are also known for their strong legacy R&D departments built around experimentation with new business models as an ongoing strategic capability — and not as a matter of tactical execution.

“It’s easier for companies to come up with new ideas than to let of of old ones,” noted Peter Drucker. Retail landscape of 2017 is different than it was in the golden age of the innovation labs. It will change even further. To survive, legacy retailers need to understand that the best mentorship comes from their former mentees.

This piece was originally published in Advertising Age on June 27th, 2017.