How to replace HR-a with a robot?
And how do you like to watch videos or movies? Someone goes to bed, someone sits in a chair with food and tea. Well, and someone, as usual, at the table, well. In any case, I suggest to take your favorite position to view. After all, today I present to you an interview with Dmitry Zavalishin, the founder of the DZ Systems group of companies, with Alexei Kosarev, one of the founders of Robot Vera. From it you will learn about the complex development of the company, problems and their solutions, the technology of the robot itself, and of plans for the future.
Dmitry Zavalishin is a Russian programmer, author of the concept of OS Phantom, organizer and member of the OS Day conference program committee, founder of the DZ Systems group of companies. In 1990–2000 he took an active part in the creation of the Russian segments of the Internet (Relcom) and Fidonet, in particular, ensured the transparent interaction of these networks. In 2000-2004, he was responsible for the design, development and development of the Yandex company portal, created the Yandex.Guru service (hereinafter Yandex.Market). You can read more on the Wiki .
Alexey Kostarev is a non-stop entrepreneur with over 18 years of experience in creating companies. One of the founders of Giftoman is SAAS Service for Retail Sales Directors to involve the store staff in fulfilling sales plans. The representative of the robot faith. The founder of the company, which began to hire people to work with the help of robots.
It sounds scary in fact, because it seems that the day has happened when robots hire people. But by and large, if you look at what you did, if you look at the entire HR process, the robot takes in it, it seems, a very small part. Or I do not fully understand the principle of your business.
Say, if we talk about now, then the robot really takes only a small part. But it occupies absolutely exactly the most routine, most boring part when it is necessary to weed out and understand which of the candidates who were made in the sample is our qualified candidate who should be invited to an interview that has undergone some kind of minimum dropout. At its core, a short-list or short list of candidates. But every day we are moving further and further along this process. And robots, unlike humans, have, of course, a number of their advantages.
They are hard to beat out.
They are hard to beat out. Probably the most interesting findings are related to the following: when a robot conducts an interview with a person, and our company, the first in the world, has already done 1 million interviews with a robot, that is, a million candidates have been interviewed with a robot; now a little more.
For comparison, and your competitors?
Our competitors, the American company Mia, made 40 thousand and the last investment round was closed by $ 30 million. So far we have attracted much less funds, but we are actively developing, almost doubled every month.
The first interesting finding was that when a person answered the questions of the robot and seemed to be already qualified, the robot starts inviting him for an interview, or invites him to talk with the recruiter, and he says: “No, I don't want to.” And this is the moment of truth when the robot asks him the question "why." And we were really surprised that almost 20% of candidates start explaining for a few minutes why they, for example, do not want to work in this company, why they are not satisfied with the conditions or something else.
They probably wouldn't say that to a person.
Man, they would say differently - once. Secondly, it seems to me that they would have said significantly less truth, because when we communicate with a person ...
Emotions are coming.
... we have to deal with emotions. When they say this to a person, you will not listen to all the records, and it is not always clear to the head of the HR service and the company itself, and what happens there. Moreover, the same situation is changing, it is dynamic. That is, for example, when some kind of federal network goes to some region, or suddenly they hired people for this vacancy, but it turned out that people are not hired for the vacancy, or they don’t actively go, or the candidate has become less.
What changed?
At this point, you can take the decryption and see what people are responding, understand what your potential employees are saying on that side.
And let's tell the audience how the service is arranged. Because you know exactly what is there, I tried a little bit, looked at it, but the viewer does not know.
The service is organized as follows: when a recruiter (usually a recruiter does this) made a primary search sample on work sites, he had a number of potential resumes, which he thinks are potentially suitable. Depending on how many candidates you need, that is, the more candidates you need, the more there should be in the sample, which does not always work in different regions or with a particularly rigid sample. At this point, in a traditional (non-digital) story, a recruiter needs to do: 1) when he did samples for several working portals, if one is not enough.
Combine.
He has duplicates. On average, such duplicates - 20-30%, that is, it saves 20-30% of work, because Vera, our Robot, immediately removes these duplicates. Then comes the most interesting process: all those candidates whom he chose potentially interesting, need to call and ask the question "Are you really looking for a job now?". Because the sample is made for a different period of time, and you sometimes need to look.
People have already found a job.
And here, on average, about 70% of all calls that a recruiter makes in the traditional world are made in vain. And this is done not only by recruiters in large companies, but also when you need to recruit a driver, accountant, assistant, etc.
The work happens as follows: you call, ask the question "Are you looking for a job?". You need to call, you need to call at a convenient time for a person and ask him basic questions that can not always be taken into account in a sample of a resume or on a work site. How much quality, and how much you understand how to ask these questions (just as we are doing now at the interview), depends on how much you get the qualified candidates. This is a special skill, a special art. Faith does it at the same time and can do it simultaneously with tens of thousands of candidates. We often do this when there are many requests.
Parallel.
She does tens of thousands of interviews in parallel and gives the employer only those candidates who have gone through the entire script (almost all of his questions) and answered them exactly the way the employer wanted. And there may be open-ended questions with different variations, for example, “Why did you quit your job?” Or “Do you have a car?”, “Do you know a foreign language?”. You can immediately switch to a foreign language with a robot. For example, asked “Do you know English?”, If the candidate answered “Yes”, you can continue to continue to trust Faith in English and check if he really knows English. As a matter of fact, there are enough such cases, such cases are an infinite set.
And you can analyze open answers?
At the moment we are able to do the following: we are able to analyze the answers, decipher instantly in the form of text and provide the employer with this in the form of transcripts for each candidate. In principle, from the point of view of the search and such a simple mechanical work, it saves a lot. It saves you from not listening.
The robot does not go into semantics, it simply asks questions, collects answers, and on the basis of these answers, primary filtering is done.
Until recently, we did not go into semantics. The primary task that we solved, we solved the problem of just a short sheet and actually acted according to the script, until recently the robot did not have any special intelligence. When we accumulated a significant number of interviews (a little more than 600 thousand), we decided that the next task should also be solved - besides that the recruiter or the robot asks questions, the candidate may also have questions. Questions may be about salary, about the company, about time to go to work, schedule, etc. That is, the candidates have questions, on the one hand, a lot, on the other hand, their number is of course. We took a certain stack of technologies and trained the neural network on the basis of the interview that we had, added there ...
Typical questions that are usually asked.
Added sample questions there. And as a matter of fact, they worked with semantics. Because the question about the salary can be asked: what about the money and what the salary is. And in any of these options, it is better when the robot recognizes and responds as needed.
Something can say in response.
Because it depends on this parameter whether the candidate reaches the interview.
Come or not, yes.
And the mobility is a significant problem. Now in almost every interview we begin to use the answers to the questions. That is, on the one hand, we interview the candidate, on the other hand, the candidate can ask questions, and here machine learning is being actively used.
Just two questions. The first question is: can he ask this question anywhere in the script, or do we need to wait for the place when the robot clearly tells him: “Now ask me, will I answer something to you”?
We are now moving more forward movement, because, on the one hand, we would like to approach a lively dialogue. But this is such a large and complex technological problem (in fact, no one has solved it in the world yet, and we have not solved it yet either), therefore, as you rightly said, the dialogue is now divided into two parts. On the one hand, Vera asks questions, then she says: “Do you have any questions?”, And here machine training is included. But we accumulate the questions that people ask, and we continue to accumulate them; we give recruiters the opportunity to see these questions that were not answered in their personal account and add answers. What, strictly speaking, benefits both recruiters and us, as owners of the service. Because Faith is getting smarter every day.
I immediately have another question arises. But I’m reporting the previous one: you just described a certain model that clearly lives in the candidate’s market. Today we actually have a significant shortage of labor, and the candidate chooses where he will go, rather than the company freely chooses from a large number of applicants. Or I'm wrong?
It seems to me, rather, not so, because the main problem of the market is the lack of so-called qualified candidates. And this story is not about Russia, it is a story in general about the world.
About the whole world, yes.
There is a funny story in that 54% of large corporate companies declare that they lack qualified candidates. But 75% of candidates respond that they do not receive timely feedback about ...
Not qualified.
No, that they passed the interview or did not pass. Understanding where in terms of this process they are. And here comes the most interesting moment: if we take any retailer, any large federal company, on the one hand, it declares that there are not enough candidates. But even for those companies that are well and publicly known, there are quite a lot of inquiries from candidates. That is, they published, for example, a vacancy on some well-known working site, and then begins a huge number of responses. But to imagine, for example, in some large federal network in the recruitment center, there are only 10-12 people working. And then the next moment arises, actually at the moment of the problem itself: when these requests come one-time, these people ...
Overloaded.
It would seem that they do not have enough candidates, but they don’t have the resources to cope with the flow and qualify these candidates. Then, when the campaign ends, the number of responses becomes less, or the competition increases, or the season, they begin to look again for these candidates. Then, as soon as these candidates are located, and they find a process, there comes a moment that they again cannot process them. Because a person is physically ... How did we start a company: we made a market place 2 years ago for recruiters (such a Uber for recruiters). We were in the paradigm of what, if we take freelance recruiters and take employers, and give the opportunity to freelance recruiters ...
On them to earn money.
On them to earn money, happiness will happen. And that was the original version. We went with her on the market. It turned out that freelance recruiters are on average 20-30% cheaper, but this new market did not create and the quality of the new did not arise. And here an interesting moment arose: we came to Ulmart to one of the HR business partners, he says: “Listen, I don't know. Uber there, I don't know what you're doing there, but you know what, guys? We are opening a warehouse in Domodedovo, we would find 100 storekeepers. ” We say: "Well, we will find." We, as active entrepreneurs, decided to cope with the task. She says: "And I will pay you 2000 rubles for this job for each employed candidate." We are happy: 100 people, 1 vacancy. Counted - 200 thousand.
Little.
We returned to the office, and then an interesting thing happened. We had 700 registered freelance recruiters. To our freelance recruiters, we say: “Guys, we received an order. Let's all urgently look for storekeepers. " All up to one freelance recruiters sent us away for a long time. They said: "Guys, for 2,000 rubles for an employed candidate ...". Therefore, no one worked and did not. We are faced with a situation when we need to solve this problem.
And you have already contracted.
And we have already contracted. At that time there was no Faith. Actually, what to do? We took work sites, began to take resumes from them, found out duplicates. We made more than 1000 calls in the first 11 days. You listened ... And when you are looking for storekeepers, you listen to that. There is such a diverse Russian language. And the main questions when you are looking for storekeepers, for example, "Is there a criminal record or not?". Because if there is a criminal record, then the security service, most likely, will not miss. And such questions are clarified along the way. And since we were tied to placement, where the candidate went to work, we learned to work with the funnel.
Our next iteration was to take three girls, create your own call-center. Girls did not survive long. First, we optimized their work, that is, first they called the numbers that we just gave to Google Docs. Then we began to understand that for a long time, when they take and copy the number and bring it to Google Docs. We wrote a small web-interface that immediately showed them the numbers by which to call. After that, it became clear that this was hard. We connected the IP-telephony and began to dial the number automatically, they only had to answer. Therefore, then we realized that it was necessary to work with the dialer and distribute the load among these girls. But at some point we asked one single question.
And why girls?
Do girls need something? As a matter of fact, at this moment Faith appeared. And at that moment it became clear to us that in history, when we take over the final employment, we do not control most of the process. We do not control and do not manage it, because we cannot influence, for example, the conversion of the security service or what happens to them there.
Well yes. You kind of can, if you ask the right questions, but so complicated, yes.
We can, but this is a very complex scheme, and the scheme is so long. We thought, and for what we could take money to allow us ...
Put your transaction castes into this money.
Yes. As a matter of fact, it was at some point, it was a bifurcation point. It was the moment of birth of the company. We thought, what if we take money for one qualified candidate, that is, for a candidate who has been interviewed with Vera. Because in order to get one of these, you need to make 10 calls. We need to find them and, strictly speaking, the whole story. And it turns out that we can get this one person in a few minutes. We were the first in the world. And, probably, now the companies that would take exactly qualified candidates are starting to appear, but there are fewer of them. And in essence, this is just the main value of the market.
Wasn’t it a problem that, roughly speaking, customers who were used to a different price structure came, and your price structure caused him to have fears or objections?
We probably lived in only two worlds. That is, the first world was the same as in 95% of young companies, when they run around the market, looking for orders. And at the moment when they receive at least one order, they are very happy and begin to work it out. We had such a hungry, interesting time.
After we came up with the idea of ​​qualified candidates, and we had the first customer (one of the first customers was MTS, their selection center in Nizhny Novgorod), in fact thanks to the innovativeness of the selection manager, who simply saw the problem and decided to try it. As soon as he tried, he told several other HRGs about it. And from this moment we are in a situation where we have actually created a completely new market for ourselves, and we have such a number of requests that over the past year we have not been in a situation of searching for customers. That is, we were in a situation only in order to properly ship and learn to do so that people could use our tool. We did not look for clients and, probably, we are still not looking for them in Russia. In the international market, we are now in the same situation as in Russia - we are looking at, looking at how to get those first ...
Customers
Customers, our first favorites.
So it turns out that the technological part of the software itself turned out to be very simple and was very easily done. From your story so it sounds.
Here the thing is: we actually created the product alive and from the first version. That is, if the first version of Faith is 3 static questions, then at the moment it is not 3 static questions, in fact, you can create a practical interface of any complexity.
These were 3 questions and 3 answers, which are analyzed in real time. That is, if the first answer is “No”, then it does not continue, but says “Thank you, goodbye”?
Yes, it was three 3 questions and 3 answers, but besides the questions and answers here in the process itself, a lot of things are covered. , , , . , , , , . application tracking system, , . , , , , - , , .
, , , .
Yes. , , , - , . .
, ?
, , . , , , - . , , …
.
.
.
Yes.
, … - . , , .
, , , , blue collars, , .. market place - - , , , , SHR , data miner ..
, .
. , , , - .
– , .
, . . , SHR Bitrix – , .
.
, . – . , , …
skill ?
, . – – , .
, ?
Yes.
...
, , , , , , , . , , , SHR 3 . .
, .
SHR , , , . , . . , , . : « , ( 35+ )… 35+ ? , ». : « ». Checked. , , , , . , .
. , .
. , , , , , . , , , . , . , -, , -, , , , . , , — , , : « . . – ». , .
, , . , « - , ; , ». .
. . , …
, .
, , .
? , .
, . , – , - , ( , ). 3 , , , .
. , , – , HR.
. , , , , - . , , : 1) …
.
, , , , . , , . . – , 10 .
, .
…
, .
, , , . , , , , , , , , , . , , . , , , , , , . .
HR, .
Yes. , , , .
?
. , , . – , . , 3 …
.
, , . ?
, ?
, . , , , , ..
. , ? , ?
Yes. , 1000+ .
...
Of course.
? , , costumer care .
Not. , , - . . – , .
, .
, - , - .. , , — exit , .
. , .
, , . , , , . , , . : , , - . , , , - , - . : , , -…
.
, . , , 3-4 , , …
.
.. , . - , , , . , , , . , , – , , , , .
.
, / , / . , , , — .
, , .
, . , , , - .
? , , , , - , , .
. , , ..
, - : «, - , ».
: . , , , , , - . , , , , , . , …
, .
, . , , , , , , .
? , , , . , ? .
. founder , , : « ! ...». , .
.
, . - , , , , HR (HR TECH WORLD ). . , , , , .
-? … Faith?
Not.
Not? – ?
.
- , , , , .
. , , , .
.
Yes.
, . - . - , , , - , , , , , , ?
Yes. . – -. , , . . , . presale.
, .
Yes. , , .
, : «, ».
– , . – , , …
.
.
, ?
.
?
, , , , , ..
. .
, , . - , - .
, . , .
, , …
, , , , .
, . … , : «, ?». compensation benefits. « ?». , - - .
HR , , .. (!) . , , . , . , , , .
- .
- , . , , , - . , , …
.
, . , , , , …
.
. 100-150 -, .
?
, (, -, ). -- (X5, ..).
?
, , .
?
, . . , . , , . – - , , , , . , , . , , , , - , , , .
.
. : , . , , — .
, ...
. , …
, ?
, . , , , . , , . , . 68 .
.
– . , , , -…
.
. , .
, 68 , , - , . , – .
– . - , . , , , , . …
Yes/no, .
Yes/no.
, .
. , .
. - , hi-tech, , . , ? .
, ( ), : 1) ; - , ; 2) , , - , , , , , , - . : , , , …
.
, …
.
, .
. , ...
.
, , , , , , .
Yes. Thank you very much.
Series of Digital Transformation articles
Technological articles:
1. Start .
2. Blockchain in the bank .
3. We learn the car to understand human genes .
4. Machine learning and chocolates .
5. Loading ...
')
A series of interviews with Dmitry Zavalishin on DZ Online :
1. Alexander Lozhechkin from Microsoft: Do we need developers in the future?
2. Alexey Kostarev from “Robot Vera”: How to replace HR with a robot?
3. Fedor Ovchinnikov from Dodo Pizza: How to replace the restaurant director with a robot?
4. Andrei Golub from ELSE Corp Srl: How to stop spending a lot of time on shopping trips?
Who is the interview with?
Dmitry Zavalishin is a Russian programmer, author of the concept of OS Phantom, organizer and member of the OS Day conference program committee, founder of the DZ Systems group of companies. In 1990–2000 he took an active part in the creation of the Russian segments of the Internet (Relcom) and Fidonet, in particular, ensured the transparent interaction of these networks. In 2000-2004, he was responsible for the design, development and development of the Yandex company portal, created the Yandex.Guru service (hereinafter Yandex.Market). You can read more on the Wiki . Alexey Kostarev is a non-stop entrepreneur with over 18 years of experience in creating companies. One of the founders of Giftoman is SAAS Service for Retail Sales Directors to involve the store staff in fulfilling sales plans. The representative of the robot faith. The founder of the company, which began to hire people to work with the help of robots.Interview
It sounds scary in fact, because it seems that the day has happened when robots hire people. But by and large, if you look at what you did, if you look at the entire HR process, the robot takes in it, it seems, a very small part. Or I do not fully understand the principle of your business.
Say, if we talk about now, then the robot really takes only a small part. But it occupies absolutely exactly the most routine, most boring part when it is necessary to weed out and understand which of the candidates who were made in the sample is our qualified candidate who should be invited to an interview that has undergone some kind of minimum dropout. At its core, a short-list or short list of candidates. But every day we are moving further and further along this process. And robots, unlike humans, have, of course, a number of their advantages.
They are hard to beat out.
They are hard to beat out. Probably the most interesting findings are related to the following: when a robot conducts an interview with a person, and our company, the first in the world, has already done 1 million interviews with a robot, that is, a million candidates have been interviewed with a robot; now a little more.
For comparison, and your competitors?
Our competitors, the American company Mia, made 40 thousand and the last investment round was closed by $ 30 million. So far we have attracted much less funds, but we are actively developing, almost doubled every month.
The first interesting finding was that when a person answered the questions of the robot and seemed to be already qualified, the robot starts inviting him for an interview, or invites him to talk with the recruiter, and he says: “No, I don't want to.” And this is the moment of truth when the robot asks him the question "why." And we were really surprised that almost 20% of candidates start explaining for a few minutes why they, for example, do not want to work in this company, why they are not satisfied with the conditions or something else.
They probably wouldn't say that to a person.
Man, they would say differently - once. Secondly, it seems to me that they would have said significantly less truth, because when we communicate with a person ...
Emotions are coming.
... we have to deal with emotions. When they say this to a person, you will not listen to all the records, and it is not always clear to the head of the HR service and the company itself, and what happens there. Moreover, the same situation is changing, it is dynamic. That is, for example, when some kind of federal network goes to some region, or suddenly they hired people for this vacancy, but it turned out that people are not hired for the vacancy, or they don’t actively go, or the candidate has become less.
What changed?
At this point, you can take the decryption and see what people are responding, understand what your potential employees are saying on that side.
And let's tell the audience how the service is arranged. Because you know exactly what is there, I tried a little bit, looked at it, but the viewer does not know.
The service is organized as follows: when a recruiter (usually a recruiter does this) made a primary search sample on work sites, he had a number of potential resumes, which he thinks are potentially suitable. Depending on how many candidates you need, that is, the more candidates you need, the more there should be in the sample, which does not always work in different regions or with a particularly rigid sample. At this point, in a traditional (non-digital) story, a recruiter needs to do: 1) when he did samples for several working portals, if one is not enough.
Combine.
He has duplicates. On average, such duplicates - 20-30%, that is, it saves 20-30% of work, because Vera, our Robot, immediately removes these duplicates. Then comes the most interesting process: all those candidates whom he chose potentially interesting, need to call and ask the question "Are you really looking for a job now?". Because the sample is made for a different period of time, and you sometimes need to look.
People have already found a job.
And here, on average, about 70% of all calls that a recruiter makes in the traditional world are made in vain. And this is done not only by recruiters in large companies, but also when you need to recruit a driver, accountant, assistant, etc.
The work happens as follows: you call, ask the question "Are you looking for a job?". You need to call, you need to call at a convenient time for a person and ask him basic questions that can not always be taken into account in a sample of a resume or on a work site. How much quality, and how much you understand how to ask these questions (just as we are doing now at the interview), depends on how much you get the qualified candidates. This is a special skill, a special art. Faith does it at the same time and can do it simultaneously with tens of thousands of candidates. We often do this when there are many requests.
Parallel.
She does tens of thousands of interviews in parallel and gives the employer only those candidates who have gone through the entire script (almost all of his questions) and answered them exactly the way the employer wanted. And there may be open-ended questions with different variations, for example, “Why did you quit your job?” Or “Do you have a car?”, “Do you know a foreign language?”. You can immediately switch to a foreign language with a robot. For example, asked “Do you know English?”, If the candidate answered “Yes”, you can continue to continue to trust Faith in English and check if he really knows English. As a matter of fact, there are enough such cases, such cases are an infinite set.
And you can analyze open answers?
At the moment we are able to do the following: we are able to analyze the answers, decipher instantly in the form of text and provide the employer with this in the form of transcripts for each candidate. In principle, from the point of view of the search and such a simple mechanical work, it saves a lot. It saves you from not listening.
The robot does not go into semantics, it simply asks questions, collects answers, and on the basis of these answers, primary filtering is done.
Until recently, we did not go into semantics. The primary task that we solved, we solved the problem of just a short sheet and actually acted according to the script, until recently the robot did not have any special intelligence. When we accumulated a significant number of interviews (a little more than 600 thousand), we decided that the next task should also be solved - besides that the recruiter or the robot asks questions, the candidate may also have questions. Questions may be about salary, about the company, about time to go to work, schedule, etc. That is, the candidates have questions, on the one hand, a lot, on the other hand, their number is of course. We took a certain stack of technologies and trained the neural network on the basis of the interview that we had, added there ...
Typical questions that are usually asked.
Added sample questions there. And as a matter of fact, they worked with semantics. Because the question about the salary can be asked: what about the money and what the salary is. And in any of these options, it is better when the robot recognizes and responds as needed.
Something can say in response.
Because it depends on this parameter whether the candidate reaches the interview.
Come or not, yes.
And the mobility is a significant problem. Now in almost every interview we begin to use the answers to the questions. That is, on the one hand, we interview the candidate, on the other hand, the candidate can ask questions, and here machine learning is being actively used.
Just two questions. The first question is: can he ask this question anywhere in the script, or do we need to wait for the place when the robot clearly tells him: “Now ask me, will I answer something to you”?
We are now moving more forward movement, because, on the one hand, we would like to approach a lively dialogue. But this is such a large and complex technological problem (in fact, no one has solved it in the world yet, and we have not solved it yet either), therefore, as you rightly said, the dialogue is now divided into two parts. On the one hand, Vera asks questions, then she says: “Do you have any questions?”, And here machine training is included. But we accumulate the questions that people ask, and we continue to accumulate them; we give recruiters the opportunity to see these questions that were not answered in their personal account and add answers. What, strictly speaking, benefits both recruiters and us, as owners of the service. Because Faith is getting smarter every day.
I immediately have another question arises. But I’m reporting the previous one: you just described a certain model that clearly lives in the candidate’s market. Today we actually have a significant shortage of labor, and the candidate chooses where he will go, rather than the company freely chooses from a large number of applicants. Or I'm wrong?
It seems to me, rather, not so, because the main problem of the market is the lack of so-called qualified candidates. And this story is not about Russia, it is a story in general about the world.
About the whole world, yes.
There is a funny story in that 54% of large corporate companies declare that they lack qualified candidates. But 75% of candidates respond that they do not receive timely feedback about ...
Not qualified.
No, that they passed the interview or did not pass. Understanding where in terms of this process they are. And here comes the most interesting moment: if we take any retailer, any large federal company, on the one hand, it declares that there are not enough candidates. But even for those companies that are well and publicly known, there are quite a lot of inquiries from candidates. That is, they published, for example, a vacancy on some well-known working site, and then begins a huge number of responses. But to imagine, for example, in some large federal network in the recruitment center, there are only 10-12 people working. And then the next moment arises, actually at the moment of the problem itself: when these requests come one-time, these people ...
Overloaded.
It would seem that they do not have enough candidates, but they don’t have the resources to cope with the flow and qualify these candidates. Then, when the campaign ends, the number of responses becomes less, or the competition increases, or the season, they begin to look again for these candidates. Then, as soon as these candidates are located, and they find a process, there comes a moment that they again cannot process them. Because a person is physically ... How did we start a company: we made a market place 2 years ago for recruiters (such a Uber for recruiters). We were in the paradigm of what, if we take freelance recruiters and take employers, and give the opportunity to freelance recruiters ...
On them to earn money.
On them to earn money, happiness will happen. And that was the original version. We went with her on the market. It turned out that freelance recruiters are on average 20-30% cheaper, but this new market did not create and the quality of the new did not arise. And here an interesting moment arose: we came to Ulmart to one of the HR business partners, he says: “Listen, I don't know. Uber there, I don't know what you're doing there, but you know what, guys? We are opening a warehouse in Domodedovo, we would find 100 storekeepers. ” We say: "Well, we will find." We, as active entrepreneurs, decided to cope with the task. She says: "And I will pay you 2000 rubles for this job for each employed candidate." We are happy: 100 people, 1 vacancy. Counted - 200 thousand.
Little.
We returned to the office, and then an interesting thing happened. We had 700 registered freelance recruiters. To our freelance recruiters, we say: “Guys, we received an order. Let's all urgently look for storekeepers. " All up to one freelance recruiters sent us away for a long time. They said: "Guys, for 2,000 rubles for an employed candidate ...". Therefore, no one worked and did not. We are faced with a situation when we need to solve this problem.
And you have already contracted.
And we have already contracted. At that time there was no Faith. Actually, what to do? We took work sites, began to take resumes from them, found out duplicates. We made more than 1000 calls in the first 11 days. You listened ... And when you are looking for storekeepers, you listen to that. There is such a diverse Russian language. And the main questions when you are looking for storekeepers, for example, "Is there a criminal record or not?". Because if there is a criminal record, then the security service, most likely, will not miss. And such questions are clarified along the way. And since we were tied to placement, where the candidate went to work, we learned to work with the funnel.
Our next iteration was to take three girls, create your own call-center. Girls did not survive long. First, we optimized their work, that is, first they called the numbers that we just gave to Google Docs. Then we began to understand that for a long time, when they take and copy the number and bring it to Google Docs. We wrote a small web-interface that immediately showed them the numbers by which to call. After that, it became clear that this was hard. We connected the IP-telephony and began to dial the number automatically, they only had to answer. Therefore, then we realized that it was necessary to work with the dialer and distribute the load among these girls. But at some point we asked one single question.
And why girls?
Do girls need something? As a matter of fact, at this moment Faith appeared. And at that moment it became clear to us that in history, when we take over the final employment, we do not control most of the process. We do not control and do not manage it, because we cannot influence, for example, the conversion of the security service or what happens to them there.
Well yes. You kind of can, if you ask the right questions, but so complicated, yes.
We can, but this is a very complex scheme, and the scheme is so long. We thought, and for what we could take money to allow us ...
Put your transaction castes into this money.
Yes. As a matter of fact, it was at some point, it was a bifurcation point. It was the moment of birth of the company. We thought, what if we take money for one qualified candidate, that is, for a candidate who has been interviewed with Vera. Because in order to get one of these, you need to make 10 calls. We need to find them and, strictly speaking, the whole story. And it turns out that we can get this one person in a few minutes. We were the first in the world. And, probably, now the companies that would take exactly qualified candidates are starting to appear, but there are fewer of them. And in essence, this is just the main value of the market.
Wasn’t it a problem that, roughly speaking, customers who were used to a different price structure came, and your price structure caused him to have fears or objections?
We probably lived in only two worlds. That is, the first world was the same as in 95% of young companies, when they run around the market, looking for orders. And at the moment when they receive at least one order, they are very happy and begin to work it out. We had such a hungry, interesting time.
After we came up with the idea of ​​qualified candidates, and we had the first customer (one of the first customers was MTS, their selection center in Nizhny Novgorod), in fact thanks to the innovativeness of the selection manager, who simply saw the problem and decided to try it. As soon as he tried, he told several other HRGs about it. And from this moment we are in a situation where we have actually created a completely new market for ourselves, and we have such a number of requests that over the past year we have not been in a situation of searching for customers. That is, we were in a situation only in order to properly ship and learn to do so that people could use our tool. We did not look for clients and, probably, we are still not looking for them in Russia. In the international market, we are now in the same situation as in Russia - we are looking at, looking at how to get those first ...
Customers
Customers, our first favorites.
So it turns out that the technological part of the software itself turned out to be very simple and was very easily done. From your story so it sounds.
Here the thing is: we actually created the product alive and from the first version. That is, if the first version of Faith is 3 static questions, then at the moment it is not 3 static questions, in fact, you can create a practical interface of any complexity.
These were 3 questions and 3 answers, which are analyzed in real time. That is, if the first answer is “No”, then it does not continue, but says “Thank you, goodbye”?
Yes, it was three 3 questions and 3 answers, but besides the questions and answers here in the process itself, a lot of things are covered. , , , . , , , , . application tracking system, , . , , , , - , , .
, , , .
Yes. , , , - , . .
, ?
, , . , , , - . , , …
.
.
.
Yes.
, … - . , , .
, , , , blue collars, , .. market place - - , , , , SHR , data miner ..
, .
. , , , - .
– , .
, . . , SHR Bitrix – , .
.
, . – . , , …
skill ?
, . – – , .
, ?
Yes.
...
, , , , , , , . , , , SHR 3 . .
, .
SHR , , , . , . . , , . : « , ( 35+ )… 35+ ? , ». : « ». Checked. , , , , . , .
. , .
. , , , , , . , , , . , . , -, , -, , , , . , , — , , : « . . – ». , .
, , . , « - , ; , ». .
. . , …
, .
, , .
? , .
, . , – , - , ( , ). 3 , , , .
. , , – , HR.
. , , , , - . , , : 1) …
.
, , , , . , , . . – , 10 .
, .
…
, .
, , , . , , , , , , , , , . , , . , , , , , , . .
HR, .
Yes. , , , .
?
. , , . – , . , 3 …
.
, , . ?
, ?
, . , , , , ..
. , ? , ?
Yes. , 1000+ .
...
Of course.
? , , costumer care .
Not. , , - . . – , .
, .
, - , - .. , , — exit , .
. , .
, , . , , , . , , . : , , - . , , , - , - . : , , -…
.
, . , , 3-4 , , …
.
.. , . - , , , . , , , . , , – , , , , .
.
, / , / . , , , — .
, , .
, . , , , - .
? , , , , - , , .
. , , ..
, - : «, - , ».
: . , , , , , - . , , , , , . , …
, .
, . , , , , , , .
? , , , . , ? .
. founder , , : « ! ...». , .
.
, . - , , , , HR (HR TECH WORLD ). . , , , , .
-? … Faith?
Not.
Not? – ?
.
- , , , , .
. , , , .
.
Yes.
, . - . - , , , - , , , , , , ?
Yes. . – -. , , . . , . presale.
, .
Yes. , , .
, : «, ».
– , . – , , …
.
.
, ?
.
?
, , , , , ..
. .
, , . - , - .
, . , .
, , …
, , , , .
, . … , : «, ?». compensation benefits. « ?». , - - .
HR , , .. (!) . , , . , . , , , .
- .
- , . , , , - . , , …
.
, . , , , , …
.
. 100-150 -, .
?
, (, -, ). -- (X5, ..).
?
, , .
?
, . . , . , , . – - , , , , . , , . , , , , - , , , .
.
. : , . , , — .
, ...
. , …
, ?
, . , , , . , , . , . 68 .
.
– . , , , -…
.
. , .
, 68 , , - , . , – .
– . - , . , , , , . …
Yes/no, .
Yes/no.
, .
. , .
. - , hi-tech, , . , ? .
, ( ), : 1) ; - , ; 2) , , - , , , , , , - . : , , , …
.
, …
.
, .
. , ...
.
, , , , , , .
Yes. Thank you very much.
Source: https://habr.com/ru/post/345662/
All Articles