Home PublicationsData Innovators 5 Q’s for Kaarel Holm, CEO of MeetFrank

5 Q’s for Kaarel Holm, CEO of MeetFrank

by Eline Chivot
MeetFrank's founders Kaarel Holm & Anton Narusberg

Editor’s note: The accompanying image is a picture of MeetFrank’s founders Kaarel Holm (left) and Anton Narusberg (right).

The Center for Data Innovation spoke with Kaarel Holm, chief executive officer and co-founder of MeetFrank, a recruitment platform based in Tallinn, Estonia that uses machine learning to match candidates to employers. Holm discussed how MeetFrank’s app can help match employers with job seekers and increase career options for its users.

Eline Chivot: How did you come up with the idea behind MeetFrank? Why did you choose to match candidates and employers based on aspirations rather than on location preferences?

Kaarel Holm: We started MeetFrank two and a half years ago, because we didn’t see many talent-based websites on the market. Human resources as a marketplace tend to heavily focus on solving business-to-business problems, that is, on how companies can hire, but there was a lack of products that could help individuals to be successful on the labor market. There wasn’t a really good way for a candidate to be available on the job market on his or her own terms without being tagged or identified as a job seeker, which can be stressful in case they already are employed.

Even though we spend much of our daily lives at work, we lack a lot of information about what is happening on the larger job market, what is our market value, what are the options at large, and so on. Our basic idea is not for individuals to reach out to a new job, but to be part of the job market, to identify where they stand, what their skills represent globally and not just locally, to build up confidence so that they know what the next possible moves are, and if they are satisfied in their current position or not—while keeping their identity anonymous. They don’t always know where all the options are beyond their ecosystem, and that knowledge is often hard to access.

Our idea to base matches between individuals and companies based on aspirations rather than location came from our data. We realized that a significant number of candidates were trying to relocate within Europe. We also found out that a large share of users would send applications to multiple countries. Our hypothesis or philosophy was that location is not a binary decision point. It is not essential for attracting talent. It is a preference, but for many people, the features, views, and values of a company are more important. If you’re happy to relocate, our goal is to find the perfect position for you—more importantly, the perfect company. But that company can be located anywhere in Europe. In most cases, you don’t find the perfect company because you are biased towards location: You start your search based on location. Talent can be willing to move if the right opportunity comes up, but if search is based on location, that opportunity is less likely to arise.

Chivot: Why is AI, and particularly machine learning, useful in the field of human resources?

Holm: Our view is purely talent-based. We mostly aim to leverage AI and machine learning to build up a recommendation engine that connects the job applicant directly to recruiters. But if we can leverage AI so that you can find the best position for yourself, it indirectly helps recruiters as well. There is a very high volume of possible job openings on the market, but individuals don’t have the resources to work through them. Our idea is to build up this strong recommendation engine that tries to search, identify, and recommend the best possible options on the market, and through this, to lower the workload for individuals. It also helps to overcome a brand-based bias. We can recommend brands that you’re unaware of but that actually are really strong on the market. We try to apply the same notion of other existing solutions to the job market. For example, AI and machine learning can help you to source the best flight ticket, the shortest route, the cheapest route. However, we don’t try to make decisions for applicants. The algorithms are just there to help them make better, more informed decisions.

We are leveraging AI to build up stronger consumer sites. We have seen that if you make the experience much better for the consumer, the volume of the possible job openings on the market will start increasing. You can see this with delivery apps and hotel bookings. On all two-sided marketplaces where consumers participate, if you allow them to make quicker and more-informed decisions, the volume on the marketplace increases. AI is helpful in this, because in our minds, human resources is a consumer-driven marketplace. The biggest data points are from the talent side rather than the company side. It can even out the market, so that a lot of smaller players and companies can compete on more equal terms with larger ones who have a stronger brand recognition on the market and more resources to hire people.

Chivot: How does your AI-powered talent matcher work, and what is it designed to do? In what ways does it help recruiters and applicants?

Holm: When you sign up on MeetFrank, there is a two to three minute-onboarding, which is completely anonymous. We don’t need to know who you are. What we try to find out is what you do. The questions are related to your daily activities in your current workplace—for instance, what are the activities on which you spend 10 hours every week? That varies from different professions. For software engineers, we would go into the languages they use, but for marketing and sales positions, it’s more about soft values. We then have a basic idea of what you do and the type of company you work for: How large is the company, what is its revenue stream, etc. That’s the minimum set of data that we use for matching. Once we know what you do, your experience, career goals, skills, and what you’d like to do next, we can use our engine to build up recommendations for job postings without actually knowing who you are.

We then proceed with asynchronized data collection. What I mean by “asynchronized” is that we collect data on demand. We accumulate more and more data based on the actions that users take on the platform. If you look at the conventional human resources platforms and processes, they are mainly built on CVs, collecting as much data as possible beforehand from CVs, and then using the same amount of data for every application. If we’re building a new application profile, we collect a new set of data that is important for the position, and we accumulate data over time that helps us make better matches for users. The basic idea is to ask for the minimum set of data we need to provide value, and build on this by collecting data from every transaction or interaction point between users and the platform.

The application works like a chatbot. Frank is the app’s character, who interacts directly with the user, and by examining his or her needs, recommends suitable offers and opportunities from companies. Frank acts as a mediator between the company and the job seeker, who cannot be approached directly by the recruiter. If interested, the user can decide to start a private conversation to engage and find out more about the company. Individuals can use the app to find out who wants to hire them, but they can also choose to let employers apply to them.

On the recruiter side, companies post detailed job ads, with salary offers, and contrary to users, they are not anonymous. Recruiters inform the algorithm about the expertise they are looking for. They find out how many people in our user base are qualified, and how they can make their offer more engaging.

Using our data, we can distinguish between active and passive job seekers. When a job seeker is active, he or she builds multiple applications in a short period of time—one day or one week. This shows the person has made up his or her mind and that it’s time for him or her to switch. But what makes us different from conventional platforms is that 80 percent of our user base are passive job seekers. A very large pool of our users makes on average 1.2 applications annually: They are not active on the job market. They only act and apply if they find something that really interests them. They don’t apply because of a need to switch positions, but when they find something really inspiring, something that matches their aspirations. It is this category of job seekers, the “passive” ones, which are difficult for companies to reach by using traditional match-making. We help companies connect with that talent pool more easily.

Chivot: How effective and useful are data anonymization techniques? Why can anonymity be an advantage in the recruitment process?

Holm: Anonymizing the data in human resources is a trend, especially to get away from biased decision-making. But we haven’t seen this work 100 percent because in the end, you do need to meet the person and that human bias is hard to remove. But we want to treat all people on the platform equally. The premise is for you to get information about the market, so that you can make informed decisions. So we use anonymization to even out the market data so that you know what are the salary levels and expectations, for instance. And anonymity helps passive job seekers keep an eye out on the labor market or weigh their options and take concrete steps towards a new position, without running the risk of their current employer finding out about their job search. It also helps them to be more open about their aspirations.

In addition, we don’t make offers or target users based on gender, age, or origin. This can help recruiters aim for greater diversity. Companies often miss out on talent because they lack the necessary tools to know how to find it, while the competition for talent is becoming fiercer. Recruiters are also biased towards location when acquiring talent. Using MeetFrank, they can find people across locations and functions.

Every position on the job market is unique. Whether qualifications, skills, experience, or the personality should be prioritized in recruitment really depends on the position as well as on the company, its culture, and its stage of development. Some companies offer time and resources and have the processes in place to nurture people’s skills and help them grow, but others need to bring in someone who is equipped with the know-how the current team currently lacks. There isn’t one single answer. Every aspect is correct, but it depends on the situation.

Chivot: What makes Estonia a successful hub for disruptive businesses?

Holm: Estonia has a generation of founders and extreme success stories, from Skype to TransferWise, Bolt, etc. There is an environment that is enabling, facilitating, and supportive of the creation of companies. It’s in our genes to be entrepreneurial and there aren’t many obstacles to start your own business in Estonia. But also, and quite importantly, our country and population are small. So when you start a company, from day one, your only objective is to expand and export. The only way to survive and build something bigger is to plan for global expansion from the start. This makes us different from French or UK-based companies where, for the first three to five years, you can focus on your own market before expanding. In larger countries, the market penetration required for success and to gain traction can be rather small. In Estonia, in most cases, you do start from your home market, but because the market is so small you need to be really efficient from the very start. The total customer pool size is so small, so the conversion rate to larger markets is higher from day one because you don’t have other options.

You may also like

Show Buttons
Hide Buttons