First blog post: moving from idea to project layout

Hi Habr! This is the first post in the corporate blog BankFiltra and I will try to tell you what BankFilter is and why it is called that way, as well as how I came to the idea of ​​doing what I’m doing.

Writing a post before the release of the product may not be very correct, but we would like the readers of this blog to understand what happened before the birth of the project, and also to try on the principles and rules laid down in it on themselves, personally observing its appearance.
')

Market situation

This story began in 2012, when I first heard about such a lucrative web industry as financial traffic arbitration. In those years, banks only went out with their offers to affiliate programs, built up Internet promotion staffs, and IT banks like TCS were gaining their peak.


Growth of the query “Credit Cards Online” in Yandex

Like you, I saw dozens of websites laying for affiliate programs, which were created with the sole purpose: to resell financial traffic as profitably as possible. At that time, web analytics of banks was lame and only the lazy of those who worked in the industry did not earn money on fraud bids, “left” traffic and a lot of other things. But the banks, of course, were not so stupid and gradually began to fight for the purity of the traffic, and the partner programs are harder and harder to drive out the frank g ** but sites.

The “Holiday of Life” was passing by; I was then working as an ordinary project manager in a rather typical and dull organization in my own way. Traffic frauds never attracted me, creating useful and convenient services — that’s what’s really interesting, to arbitrate from one channel to another — no. That is why I immediately realized for myself that the whole crowd that surrounds the financial traffic is not about me.

Another year passed, but I didn’t do anything, because I knew: I lack experience.

Where did the wind come from

I always liked Yandex.Market. A large number of properly working filters allows you to find exactly what you need. Working as a manager, I designed the work and implemented similar solutions in online stores and other projects with a large number of parameters. The successful projects of CreditCardsOnline or Banks.Ru at that time did not give users the ability to fully, conveniently and competently filter out such complex and difficult to understand products like credit cards and loans.


Choosing cards on CreditCards Online, Banks.Ru, Compare.Ru and other portals

From words to deeds

After a preliminary analysis, I realized that there could be several filterings and sorts:

The first is based on user data. This filtering allows you to weed out those products that are clearly not suitable for any of us in such simple parameters as:

• Age: everything is simple, we show only what suits
• Gender: cut off products by gender
• Activity: we filter products under what the user is engaged in
• Proof of income: we filter products by the most problematic side of getting a loan
• Territoriality: display only what is in the region / city of the user

After the first filtering based on the profile, the user can no longer face the human factor, for example, not falling into the age requirements, being in a situation when a credit card is issued in another region and other similar ones. This is an important part of human error can be hidden behind the simple and familiar registration for us: logging into the service, you already see only what suits you. In addition, parameters like Activity and Income Confirmation already allow the service to present some User Credit Score and not to show premium products to retirees and students, for example.

The second is custom filtering. This item is the most obvious and clear, for the user, the simplest and most understandable filtering by amount, bonuses, payment system, card status and other product chips remains.

According to all the criteria, mobile development was suitable: it solved the issue of geo-determination especially easily, so it was decided to make a mobile application, and not a website. The name also appeared somehow by itself: we are filtering banks, which means BankFilter. But my story would be incomplete, and the service is absolutely completely empty if there were not another most important parameter that should help any user in choosing a loan product.

Math will save the world

As you know, in their advertising campaigns, banks usually indicate the lowest interest rate and the highest credit limit. The real interest rate that a borrower gets on his hands can be very different from the advertising one, and it is impossible to recognize it in other ways except by contacting the bank. The same with the credit limit. Typical example:


In fact, TCS Bank issues an average of 3 to 15 thousand rubles for the regions and slightly larger sums for Moscow and the region, at 35-40% per annum.

This was the main goal - to make the impossible possible. For this task, I had to write together with fellow mathematicians an algorithm that could most accurately determine the interest rate that the bank actually sets. But one thing is to write a formula, another is for it to work, so it had to be checked for something. To do this, I had to kick all possible acquaintances in the banks so that they leaked information about the real interest rates issued by the banks in their hands. So we saw the difference between advertising and fact, and also saw that our formula has small errors.

This ordinary mathematical formula passed the preliminary tests, showing a deviation of no more than 2% upwards or downwards, which suited me completely. Of course, the algorithm remains a secret and a principle of the project, as well as being a direct participant in sorting and user chips: products in the application sorted by exactly this “average rate”. Thus, anyone entering the application immediately sees the most profitable products with the lowest average rate, and the list itself is sorted by the ascending average rate.

So I understood what users need and how to make a convenient selection of products. The other part of the project should have been other market participants - Banks, because they will be my employers and they also need services and data, but this is a completely different story ...

Stay Tuned.