Startup of the Day (December 2017) / Blog of the company Mail.Ru Group / Habrahabr

January 11, 2018
115 Views

Continuing the digest series "Startup of the Day", today I present the most interesting projects for December. If you want to get acquainted with the rest, then ask in my blog. Entries are available in Facebook, ICQ and Telegram.

Privacy.com

The American startup Privacy.com has a very steep and expensive domain, but it does not reflect the essence of the product completely. The startup protects not only the client's private data, but also his money.

The modern Internet always requires the user to pay – subscriptions, goods, new levels in games. The active consumer enters the parameters of the card almost every day, and here there are two problems. Firstly, just scammers. Secondly, smart marketers from venture startups, which turn any service into a subscription, this increases LTV.

Privacy.com allows for each payment to create a special virtual card, one-time (for unsolicited subscriptions) or simply with a limit payment (for questionable stores). The plug-in to the browser automatically recognizes the credit entry form, click-click-and a completely new number for this special case is ready, there will be no unexpected surprises. Here there is, and, in fact, privacy – for a temporary card a fake name with an address comes up, the store does not recognize what it does not need to know. When you buy material goods, you still have to open, but you can go to pornographic sites incognito, wow!

The startup creates its cards on top of bank accounts, not a regular credit card. This is important: in the end it is Privacy.com that receives the commission of the issuer, the order of a percentage of each purchase, and for the user all services are free. The reverse side of the coin is the need to integrate with banks separately, the startup works with the largest of them, but not with everyone.

During three years of life, Privacy.com raised two million dollars of investment, the mass audience has not yet recruited, but the product (in the USA) working.

Kiwi.com

Air tickets are a difficult market for intermediaries. Carriers are few, they have a strong brand, with the exception of directions like Moscow-Peter, each flight is unique in time. As a result, the power of agencies is weak, and their commissions vary from low to zero. The main money OTA earns on hotels or on additional services to an air ticket, remember the same FlySMS.

Kiwi.com found a way out that it sells unique tickets. Ordinary flights with transfers, which are in any search engine, are the result of cooperation between airlines, they negotiate among themselves and sell a single ticket for which they are responsible. If the first plane is late and the passenger does not have time to change, then he will be fed-fed at the airport and taken away by another board at the first opportunity – all at the carrier's expense. In addition to selling such tickets, Kiwi, in contrast to other agencies, also unites the flights itself, at your own peril and risk, without the knowledge of partners, including combinations that are absolutely impossible in other places – loukosterov with premiums, for example. Saratov – Domodedovo arrives at 11:00, Domodedovo – New York leaves at 16:00. OK, we sell Saratov – New York, for five hours the passenger will reach.

If something went wrong, the startup itself compensates for the damage under rules similar to the rules of airlines. Insurance, of course, costs Kiwi money, but he also appoints prices for combined tickets himself, in the margin he is not limited.

The user, except for possible compensation of a non-payment, receives the fact of the chosen route. When flying from Saratov to New York, choosing a transfer is trivial, two tickets are quietly bought in fifteen minutes, but for Moscow-Quebec the cheapest route was found through Milan, London and Toronto. Independently such a person does not find in principle. Of course, when tickets are already found, they can be bought directly, past the aggregator, but the average traveler in life does not realize that it's cheaper, plus the loss of insurance, plus the risk that the second will be sold during the purchase of the first segment.

In all the rest, except for its own dockings, Kiwi is an ordinary online agency. Standard interface, classic promotion through affiliate programs and meta-searchers, everything is like everyone else. The startup did not raise much external investment, it started at the expense of the parent company. Recently, Kiwi loudly hinted that he would not mind selling for half a billion dollars at a gain of one hundred million, a positive EBITDA of nine and plans for a twofold increase next year.

Shazam

In December, Apple bought Shazam for $ 400 million – an excellent occasion to note.

During its life, the startup has gained more than a billion downloads. It is unlikely that any of the readers do not know how it works, but just in case I briefly describe the idea. The user starts the application, presses the main button the size of the half-screen, Shazam recognizes the music playing next to it, tells its author and the name along with the link to the store. The second half of the screen is occupied with advertising;

Shazam works well, music is recognized not only from the official track, but also from the background in the commercials or mixes. The base of records does not cover in general everything created by mankind, but it is very great. For example, the startup knows how to find "Komsomol members-volunteers" in the performance of the ensemble named after Aleksandrov, but does not know "When the dawn, comrade?" Mireille Mathieu, – the actuality border lies somewhere between these two compositions.

The Shazam story – the history of two technological revolutions. First startup investment was received back in 2001, six years before the release of iPhone, a year before commercial 3G networks appeared. The product was impossible not only to massively use, but also to effectively monetize – the presentation of the iTunes Store took place on April 28, 2003, Spotify was born in 2006. It took ten years for all necessary infrastructure to appear and Shazam became really in demand. The audience grew, together with it the revenue grew, the company attracted new investments for an ever higher appreciation. In 2015, the startup reached an estimate of one billion dollars, the prospects were the most rosy.

And then a revolution of deep training occurred, a breakthrough occurred in the development of neural networks. Before our eyes, the key technology of Shazam from the semi-mystical achievement of engineering thought is rapidly turning into a good diploma work of the student of the Technosphere. And at the same time, prospects are lost: in the new reality, the product is relatively easy and cheap to repeat, there is no competitive barrier to it, the user does not care whether to use the original or the clone.

At the same time, maybe not today, Shazam wants someone to sell: the funds want to return their investments, and the founders finally have something to gain from a project that has been running for almost twenty years. The fair selling price is determined by the current revenue: $ 50 million per year – and this is definitely not a billion. Apple is from ideal buyers: it has a lot of money, it's a huge interest in music, it's hard to come up with a more logical deal.

Earnin

The monstrous 300 % per annum is at the same time pathetic 0.4% per day. This mathematical illusion suffices for the lives of many MFIs, but people are gradually learning, and regulators are struggling with ultra-high interest rates. The American startup Earnin makes the business even more curious: a miserable five with a $ 100 loan is a predatory 5% in two weeks.

The startup gives out loans at zero percent, but offers to voluntarily leave a tip – after all, his service is no worse than a restaurant or a hairdresser. In addition, he asks quite a bit, because see the main mathematical illusion: five dollars for help in a difficult situation is perceived much better than 250% per annum.

In addition to this focus, Earnin is indistinguishable from any modern MFI: mobile application, instant delivery of money, all like everyone else. For scoring, it requires access to an online bank and issues only amounts that the borrower will return from the nearest salary, and to guarantee a refund, he automatically writes off the money as soon as they appear. However, it seems that nothing prevents crossing the technology with tips and more risky scoring.

Unicorn Earnin on this model obviously will not, but he found his niche. The economy of the start-up among generous Americans is excellent, the company has raised 65 million dollars, in 2017, managed to get two rounds.

Pymetrics

: interviews do not work, interviewers are subjective and select candidates according to incorrect criteria. Not only does the effectiveness of business suffer, but so does discrimination: the correct ratio of sexes and races is not observed, white men get to the best positions.

The mission of the startup Pymetrics is to correct these shortcomings. They developed a psychological-intellectual test, made in the form of games. The candidate answers questions like "you were given 10 dollars, and the partner was not given anything, how much you give to him", solves mathematical problems and for speed clicks on the space. Testing, for my taste, is pretty fun, a couple of games are just fascinating.

The results are ready in 40 minutes. The candidate receives a sorted list, how much it corresponds to each profession in the personality warehouse: "investment banker – 5%, project manager – 10%, teacher – 90%". Startup claims that behind this career guidance are not just Sly Algorithms, but High Science, and she needs to believe without reasoning. A posteriori confirmation of the accuracy of the tests Pymetrics considers his immortality in demography: men and women, Asians and Latinos receive the same estimates. In case studies, the startup pushes cases when with its help the clients have balanced the flows of accepted candidates. Whether this confirms the effectiveness or senselessness of the algorithm is up to you. On my example there was a clear nonsense: judging by IT technologies, which appeared in one of the last places.

On the other hand, Pymterics decided that I ideally suited the work of investment banker, and my name and link to the profile on Facebook went all interested in such specialists companies. It's hard to imagine that this mechanism really helps hiring: the recruiter gets semi-anonymous candidates without specifying professional skills and licenses, conversions should be zero.

Two other monetization options look realistic. Pymetrics offers to screen tests of already existing candidates: there are 1000 people wishing to become programmers, we will run them through toys, we will leave 500 of the most suitable "in science" for the next stages. If the competition is large, even an absolutely random selection will not be too harmful, "we do not need losers," and the results of Pymetrics correlate with the true estimate. Yes, and an alternative – interviews, they are more expensive and even worse. The second approach is the assessment of employees already working, let the technology prompt which side can be moved and from which testers to raise future managers. The final decision is still for the people, the machine will not hurt much, and again, even at random, sometimes it will offer good solutions. And of course, diversity is always respected.

The start-up investment received $ 16 million.

Leave a Comment

Your email address will not be published.