Secret to circular business: listen to customers

Transitioning to circular business models requires not only new circular products, but new ways of collecting and communication information about the resources moving between different actors in a value network.  Materials in a linear economy typically move in one direction— beginning with resource extraction, then to manufacture, distribution, use, and disposal. 

The same is true about information! Manufacturers in a linear economy order uniform components from upstream suppliers so that they can produce the same thing over and over again as cheaply as possible—at economies of scale. In a circular economy, resources and information flow in multiple directions including “backwards” in the value chain. Used clothing, used electronics, and used furniture flow in diverse forms, in various conditions, from countless diverse end-users.

The RISE-led, Vinnova-sponsored, project Digital Decision Support for Refurbishment (2DSR) has tried to address this dilemma by developing digital tools that support decision making for repair and refurbishment in the furniture sector. One case study in the 2DSR project involves a Stockholm-based startup called OOAKI (One of a kind living, ooaki.se). OOAKI sees itself as a “tailor for your sofa”, selling customers new sofa covers that allow them to upgrade their existing sofa rather than purchasing an entire new sofa. 

OOAKI joined the 2DSR project interested in developing a digital tool that could streamline their interactions with customers. While the project envisioned an image recognition tool that could expedite how customers measure and order new sofa components, OOAKI’s journey has taken some unexpected and exciting turns. 

Below is an interview with OOAKI’s Chief Digital Officer, Lennart Svanberg describing what they have learned and how changes in their customer interface have strengthened their circular business.

RISE: OOAKI started as a traditional design company in the luxury sofa segment but later chose to integrate a circular business model. For many companies, the decision to “go circular” involves significant risk. What motivated OOAKI to adopt a circular business model?

The customers. Yes, that’s actually the short and simple answer. Our founder, Andréea Lindberg, had for many years built custom-made furniture for clients. After the furniture had been used for a few years, customers began reaching out with the question: “We love our piece of furniture, but we’d like to change the sofa fabric—can you help us with that?”

Andréea thought that perhaps it wasn’t only her previous customers who wanted to keep their sofa frames, so she launched the website “nysoffkladsel.se” (newsofacover.se) as a test. It took only a few days before the first inquiries came in, and that’s how Ooaki was born.

What challenges did OOAKI face when implementing the new business model?

We are the equivalent of the upholsterers or furniture workshops of the past. That’s where you used to take your “worn-out” furniture to give it new life, for example with new fabric. Tailors can also provide these services, and this model exists all over the world—we know this because we conducted extensive global research.

Scaling up sofa refurbishment to an “industrial” level is something we are unique in, excluding companies that specialize in specific furniture brands. The challenges are many, since every sofa can have a unique combination of width, depth, and height. We therefore need to be meticulous to ensure that the new fabric fits just as well as the old one. In the end, our knowledge of the manufacturing process is something we are likely world-unique in.

When it came to the sales process, that was also challenging at first, as each customer became an individual sales project. We requested photos, measurements, and exchanged conversations about fabric choices. In addition, we send fabric samples to customers—and that will always be the case—because customers naturally want to both see and feel the fabric they ultimately choose.

As we completed more and more sales, we were able to begin standardizing our pricing and provide quotes more quickly. But until the project we initiated with RISE, our sales process required a great deal of manual work and was time-consuming.

Why were you motivated to participate in the 2DSR project? How did you imagine the project could help OOAKI?

I had previously had the privilege of participating in projects together with RISE at the company Lingmill, which was a pioneer in what we now call Natural Language Generation. So when ideas started flowing about how we could streamline OOAKI’s sales process using new technology, it felt natural to contact a RISE contact and discuss our ideas about machine learning. That contact helped us get in touch with Robert Boyer, who was recruiting partners for a the 2DSR project.

After a short period, we were accepted as a project partner, and to our great joy, our application was approved.

The original vision of the 2DSR project was to develop an image recognition tool that could streamline customers’ interaction with OOAKI. For several years, RISE and OOAKI attempted to develop a model capable of identifying the “type” of sofa (three-seater, two-seater, corner sofa, etc.) and estimating its width, so that potential customers could take a single photo of their sofa and receive an immediate price quote—hopefully speeding up the purchasing process. This proved challenging for several reasons. First, we encountered changes in EU legislation that forced us to alter our model development strategy. Then we attempted to develop a model based on data collected from random individuals who took photos of their sofas, but discovered that people generally cannot measure their sofas reliably. We then tested a “Hollywood technique,” where we ourselves took photos of sofas. Eventually, we developed a tool using OpenAI, and at the same time OOAKI decided to follow a slightly new strategy that has proven successful (which I will ask about next). But I am curious whether you have any reflections on AI and machine learning after this experience.

Yes, we learned an enormous amount about what AI is good at—and also what it is not good at. We like to believe in new technology as a “savior,” but generally it is more of an assistant. It is we humans who need to think innovatively in order to ultimately change something fundamental.

There was probably nothing wrong with the original idea—using image recognition to provide faster price quotes. But we learned the hard way that AI-based furniture image recognition still has a long way to go. Even more importantly, this is not actually how the majority of ordinary consumers prefer to proceed. We believed customers would be willing to take photos of their furniture to get a price, but our estimate is that perhaps fewer than 10% of potential customers are willing to do that work. Most want to avoid taking photos and still receive a price.