The Center for Data Innovation spoke with Sara Green Brodersen, chief executive officer and co-founder of Deemly, a Danish social reputation website based in London. Sara Green Brodersen discussed the added value of platform independent services and online trust systems for both companies and users.
The interview has been edited.
Eline Chivot: You founded Deemly in 2016, after doing some research on the sharing economy and interviewing European platforms and startups. You found out that trust was a key challenge for these companies and their user communities. Can you explain why?
Sara Green Brodersen: Trust is a driver at the heart of the sharing economy. Yet our research showed that it was a major barrier for both the users interacting online and the companies establishing themselves in this industry. Through many of their “sharing transactions,” users are interacting with people they don’t know. In addition, this area is not regulated. I have defined trust in the sharing economy based on a three-tier model: for a user to go into the sharing economy and take steps such as a transaction, there are several trust layers that need to be fulfilled.
Let’s say I’m going to Brussels and I’m staying in an Airbnb. First, why would I do that rather than staying in a hotel? To choose Airbnb as an option, I would first have to accept the premise of the sharing economy. Meaning: is that something my friends and family are doing? Is it even legal? This level is about the general societal concept of trust. Second, I would need to trust a company to facilitate the transaction—which in this case would be Airbnb. Do they have a 24-hour support? What happens if something goes wrong? Do they have and provide insurance? The third level of trust is about trusting my peers. Who am I staying with? Are they an axe murderer or just a bit annoying?
The trust level that we’re focusing on with Deemly is the latter, which is about identifying who you are, and who you are “sharing” with. Based on our research, we know that although many platforms include ID verification and background checks—which is very good and important—many of the services are expensive and there will be different checks for every country. That makes it sometimes difficult to scale the platform across borders. Also, it might provide information such as a user having a clean record, but not about how the user you’re interacting with may be as a person, if you have anything in common, or if he or she will leave your house is a good state. Hence users’ reputation and credibility require a different layer of data showcasing that as well. This is what we are trying to build across all these platforms.
Chivot: How does your technology provide this missing link of trust? And what happens when bad actors post, for example, fake reviews that can undermine trust?
Brodersen: Our software currently works as a freely customizable tool. Every platform that installs it can showcase it in any way they want: we know that they are all very different from each other, and that their users pay attention to different elements. When we started, we did use a “trust score,” which is still functional, but we no longer promote it as much as it turned out not to be useful in breaking down trust barriers. It is in e-commerce and the travel industry, but in our case, it is hard to quantify trust into one score. What we’re working with now instead provides an overview based on the analysis of written reviews through natural language processing. For example, it includes information about the words that are being used often and mostly generated about a particular user, are they more negative, more positive, etc. This helps platforms learn about what it is that individuals are discussing, for example what they tend to complain and be concerned about when writing these reviews.
Regarding fake reviews, there are several aspects to consider. First, we aggregate from all platforms, so the actual reviews are not generated with us. This means we are depending on them detecting fake reviews. Second, there are fewer fake reviews in the sharing economy compared to what we see with e-commerce platforms. To create a fake review as a user, you would have to go through a transaction within a platform’s system, which is quite cumbersome. You’d actually have to book through the platform, and your review would only show up if you were to get a review from another person as well: it’s a two-way, blind review. As a result, this is not as common a problem as in other industries. On e-commerce platforms, you can just go to a site and say that you had dinner at this place or bought something at this company—and there are just not that many checks to prevent you from doing that even if it’s not true.
Chivot: Companies from other industries could be interested in the sharing economy as a business model. Which other applications do you see down the line for your system?
Brodersen: We have not ventured into any of those industries yet ourselves: Deemly is a startup, with limited resources. There are some legal and ethical barriers that need to be defined and refined as well before we would want to go forward. But we have been contacted by every possible industry as companies are indeed trying to figure out how they could be using our data.
One is the financial industry, including banks and businesses for credit scoring. They are looking for ways to collect other kinds of data to assess users’ credibility and trustworthiness. For example, some of their customers or potential customers may be millennials who travel, who have lived in different countries, and who have not stayed in a country long enough. This makes it hard to build credit worthiness. Insurance is another interesting sector of course: as there isn’t a lot of data that can be used for insurance policy at this stage, they are looking at including individuals’ trustworthiness as a risk factor.
A lot of dating sites have contacted us as well. Unsurprisingly, perhaps, as trust remains a huge issue in this space. Users are facing problems related to meeting up for dates with strangers. And usually, they don’t tell many people that they are going on a date.
Systems like ours can really support stakeholders in the field of human resources and employment. Many platforms are freelance-based. Deemly has been working on a Swedish governmental project that helps every user that has been active on a platform to gather ratings, reviews, as well as all the activities they have done through this platform, and to consolidate all that into an online CV. They can use it as a format for an application in the public sector. This makes sense as many students are, for example, walking dogs, or putting up shelves, and that should also count as experience. Furthermore, many immigrants coming to Sweden are driving Ubers as their first job and making a living, which should also be taken into account by the system.
Chivot: What are the benefits for bigger platforms to open their data to smaller structures like Deemly?
Brodersen: Big platforms work with us either as customers using our technology, or as “data partners” giving us access to their data if their users allow them to. When we talk to data partners, their motivation is two-fold.
First, the European Union or the Federal Trade Commission in the United States are pushing them to share more about their data, so as to help legislators understand the industry.
Second, platforms are also trying to understand their own data better: perhaps is there an innovation they haven’t really seen yet themselves? We can help them create a commercial benefit with the data they’re sitting on.
I should add that over the last year, some developments have pushed the discussion between us and our data partners. One is the GDPR coming into force, and particularly the right to data portability. Companies have to oblige if users ask them for their data to be taken away or transferred to a different platform. This way, we can help platforms not only to comply with legislation, but also to create commercial benefits. Another development has been this series of data leaks and breaches. Companies are becoming increasingly aware that it’s not really their data they’re sitting on, it’s the users’ data. As a result, they are looking for ways to help users to not only keep the data secure and safe but also to put it into their own hands, so they can leverage it in any way they want. For example, if users want to use their data to apply for a job, they should be able to.
Chivot: What do you expect the sharing economy will evolve into? What are the conditions and the environment that could support or prevent its development?
Brodersen: The big discussion in 2018 was blockchain, especially from a data perspective. The idea was to help break down barriers to trust, by keeping data secure on the blockchain. In theory, I think this is a really good idea. But what we’ve heard and seen is that first, this system is not really being requested by end users because they don’t really understand what it is. Second, the platforms think it is too difficult to work with. And third, to build and develop it, a company like ours would have to raise ten times more in funding and investment. Right now, this system is not mature enough and no one really dares to do it full scale yet. At the same time, technological advances have a way to happen so fast that blockchain could still be a big topic this year again.
Another trend I see is that many more large companies are joining the sharing economy, either by setting up their own sharing initiative or by acquiring existing initiatives. This is mostly where opportunities could happen in terms of adoption at scale—perhaps because a well-known brand behind a platform may be a strong driver of trust among users.
What I hear is the biggest problem for platforms is to be able to scale across borders—especially in the case of Europe, where there isn’t just one legislation in place for all countries. Legislators were too slow to deal with the sharing economy, so platforms had to move forward and create their own self-regulating mechanisms. Systems such as ratings and reviews have proven helpful in supporting efficient self-regulation. But if we could at least agree on something at a European level that would be reflected in each country, that would be a helpful start to support the sharing economy.
Another obstacle to keep in mind is that there is no such thing as one law applying to the sharing economy. It may be coined as one single industry, but it includes many different pillars: transportation is completely different from shared accommodation or freelance. Many different legislations—one for example being employment laws—would have to be updated to “fit” the sharing economy.
Finally, circling back to trust, our Western societies may seem very homogeneous, but there are considerable differences in terms of how we assess who should own our data and the trust we place in whoever handles it. I once had a discussion with a Swedish entrepreneur who runs a sharing economy platform. He thought that what we’re doing at Deemly should really be handled by a government—that governments should gather all the data. And I just came back from a conference in the United States, where I heard the exact opposite opinion: the less government has knowledge of your data, the better.