A Berlin-based startup offers its users a steady stream of job adverts, automatically selected to match a subscriber's interests. The initial focus is on IT and marketing jobs. Sabine Kinkartz reports from Berlin.
Spending hour upon hour searching online job adverts, looking for something that fits one's interests and expertise, is something most of us have experienced at least once. 'Jobspotting' proposes to make job-searching easier and more precise - using automated search algorithms that learn a user's preferences.
The company is based in Berlin's Mitte district, in an attic office filled with old furniture taken over from the last tenant. It's near a crowded intersection called Rosenthaler Platz, where hipsters, backpackers and immigrants flow through a jumble of cafés offering everything from Turkish fast food to French cuisine.
Founding COO Robin Haak, 28, is the business guy; his three co-founders are long-time ex-Google coders, specialized developers with deep knowledge of search algorithms.
The team is on a mission to revolutionize Internet job-searching. The goal: to do for job-searchers what Spotify, Pandora or Netflix did for the music and film markets - namely to automate the process of searching through job-offer databases by storing job-searchers' profiles and learning their preferences over time, in order to present them only with job ads that are likely to be of interest to them.
Initial focus: IT and marketing jobs
Jobspotting's website has been up and running since the beginning of 2015. Its meta-search engine is focused, for now, on browsing through around 200,000 IT and marketing job listings in Germany and the UK on behalf of Jobspotting's users.
If that works out well, then other professional job categories will be added. Civil and mechanical engineering jobs will be added next, and perhaps finance jobs after that. The geographic scope will eventually be expanded to cover the rest of Europe and the USA.
Jobspotting has collaborations with established job portals like Monster, Xing and Stepstone, where firms looking for new employees post job adverts. The startup's meta-search engine reads the ad texts, classifies them, and then compares them with the profiles job-searchers have posted on Jobspotting.
A number of other companies have tried to offer automated job-searching in a similar way, but all of them failed. "You have to use semantic search methods to read and assess whole texts, and that's technically very demanding," Haak explains.
30,000 skills to choose from
Feedback from Jobspotting's users is key to the system. They're asked to provide as precise a profile as possible, and to help with that, they can choose from 30,000 linked and overlapping skills.
"It's like a huge skills map, and the system has to know that, for example, marketing is closer to online marketing than to computer programming skills like Java or CSS," Haak says.
The system also asks users to enter other relevant data, for example, job experience, location preferences, and so on.
The program then learns and improves its custom targeting of job adverts to individual users based on their ongoing use of the system. It depends on whether the user accepts or rejects a particular ad as relevant, but also on how the user sets boundaries on his skill set.
Jobspotting have registered a patent application on the search algorithm, and the developer team is continuing to improve it. Data confidentiality isn't an issue, Robin Haak says: "We only have users' email addresses. We don't know where they live or what they do. You could almost say that all our users' profiles are anonymized."
Only growth counts
Asked if the Jobspotting team is worried that some other company might copy their approach, Haak just smiles indulgently. "If five really good developers got to work trying to do what we're doing, they'd need at least 24 months - assuming they knew how the algorithm works and had a roadmap for building it - just to build what we already have now. It would be very, very difficult."
Two years, that's half an eternity in the world of Internet startups. It's a world that follows other rules than that of normal enterprises. One of the differences is that nobody cares whether Jobspotting is making a profit or not, at this early stage. The only thing that matters is growing fast and getting well-known.
The startup brings in some money by means of a 25 cent per click reward whenever a user goes from Jobspotting's site to a site like monster.de. But Jobspotting only earns a couple of euros per user, Haak estimates: "But if I have a hundred thousand users, then after a couple of months, I'll have brought in maybe six hundred thousand euros."
The team is the key
Jobspotting signed up 20,000 users in the first two months. That's not yet enough to pay the bills. The nine-person employee team's salaries are being paid using $400,000 in seed capital, put up by a couple of angel investors. The four co-founders are living mostly on savings, which are gradually dwindling.
"At the next financing round, I'd like to net another 500,000 euros or so, that would be cool," says Haak, and laughs. "But as a founder, you don't really pay yourself until the B series round at the earliest, when around $5 or $0 million in investment capital come in."
The next financing round is scheduled for the coming summer; it'll aim to raise around 2 million. "Early on in a tech startup, you have to focus all your available money in the technical team, the employees. It's crucial to have the best people."
Those employees aren't motivated alone by pay. Equally important, Haak says, is a friendly atmosphere at work, and being given interesting technical challenges to solve.
The developer team is international; the language in the office is English, not German. Two employees are Polish, another is Spanish. One of the co-founders was raised in Sweden, with an Iranian father. He cooks for the whole office every Monday. Wednesdays, the whole team goes to a nearby restaurant, at company expense.
A shared breakfast sitting on benches is par for the course, as are flexible working hours.
"Here everyone can come and go as they please. I don't have to see people at their desks here, I don't care about that," Haak says. Employees can work from home if they like, and after about 6 p.m., there's rarely anyone in the office. "When we're here, we work in a totally structured and efficient way."
It isn't only the flexible working hours that are different from standard corporate employment conditions. There's little in the way of a hierarchy. Each individual is a part of the team, and everyone is working on a self-directed basis for the team's common success. "What a person does or how they want to do it is up to them," Haak says.
"If people want to spend some time with me and use me as a sounding board, I'm here for that. If they don't, then they don't."
Selling the company isn't on the table - yet
Every three months, the team resets its goals. That doesn't refer to goals like more money, Haak explains: "We work like they do at Google, which means we always try to set goals which really can't be achieved - you set ambitious objectives for yourself, and then you try to deal with it."
Someone who only sets achievable goals, Haak says, isn't challenging himself enough. At the same time, he admits that it isn't good to fall short of every target.
One goal that isn't on the table for now is selling the company - which is a standard 'exit strategy' for many startups. Haak admits that if an "unbelievably high price" were offered, then the founders would consider it. But so far, they've only received an offer to buy into the company by a would-be investor who wanted a say in its management. And that, Haak emphasizes, is out of the question.