June 6, 2013 at 10:59HeadHunter Salary Expectations Statistics
Any topics related to salary are of great interest. My topic is " How to understand how much you are worth, or salary clusters in action!" »About how to independently collect statistics on salaries, scored nearly a hundred comments and received a positive rating from readers.
Today I will tell you how we automated the visualization of such statistics by creating a new service, “Salary Expectations Statistics”. It is available only to employers so far, and we will start the service for applicants (with statistics on salary offers) later.
Under the cut, details about the feature and how we developed it.
The issue of salary is one of the most difficult when hiring employees. By not paying, you can’t find yourself a suitable person or find an employee with too low competencies. The other side of the coin - you can overpay by spending, in fact, extra money, which on an annualized basis, including taxes, will amount to a considerable amount.
To establish the very “tasty”, but at the same time adequate salary, you can use statistical information and salary reviews from consulting companies, recruitment agencies and job sites. Unfortunately, most of the instruments existing on the market have one serious drawback: they are not oriented to a specific vacancy. Roughly speaking, they measure the average temperature in a hospital. Our service does not have this drawback.
The team approached the task with the understanding that we are sitting on a mountain of information. Still, 10 million resumes are a fairly large amount of data for analysis by various sections. We implemented this analysis in real time on the form of creating a vacancy:
Statistics are built on the basis of the expected salaries indicated by applicants in the resume. When the employer, when publishing a vacancy, enters the proposed salary, a graph is displayed to the right of the input field with the distribution of applicants by expected salary. It highlights the range of job seekers who could potentially be interested in a job with such a salary (job seekers who indicated an expected salary similar to that offered in the job). When updating data, the chart is rebuilt using unobtrusive animations that accentuate the user's attention at the time of recounting on the chart.
We started developing this feature on two fronts at the same time: designers worked on the appearance, and developers on the backend and test visualization of the displayed data.
From the point of view of development, the main risk was the possibility of a lack of data, because quite often we need to read statistics on rare specialties, and even in small cities. As a test, we tried to make a visualization for the position “manager in a shoe store”:
There was enough data, but there was an understanding that such “fences” are not very suitable for analysis - you need to either smooth out or present the data in the form of a histogram. Our designers, meanwhile, have already made the first sketches of the future interface (we do not pay attention to the currency - this is just a “fish”):
We realized that on the histogram it is very difficult to display the range of salaries so that it is understandable to the user. Therefore, another option appeared, which turned out to be closer to the final one: It
remained to concentrate the user's attention on the fields with salary and remove information noise:
we did the frontend part using the Raphaёl JS library , which allowed us to support even fairly old browsers (Firefox 3.0+, Safari 3.0+ , Chrome 5.0+, Opera 9.5+ and Internet Explorer 6.0+) and to avoid troubles with the implementation of SVG and VML for Internet Explorer.
In the process of demos, we realized that it would be nice to give the user a text prompt about the salary range that he indicated:
These messages can solve the problems that I described at the beginning of the article.
I want to say right away that statistics on salary expectations for employers when creating a vacancy is only the first step towards standardizing the salary market. We have many ideas where such statistics would be useful (including for applicants), so stay tuned!
PS Your constructive suggestions and comments can be written in PM or on my mail b.volfson (dog) hh.ru, well, you can troll in the comments :)
Today I will tell you how we automated the visualization of such statistics by creating a new service, “Salary Expectations Statistics”. It is available only to employers so far, and we will start the service for applicants (with statistics on salary offers) later.
Under the cut, details about the feature and how we developed it.
What problem do we solve
The issue of salary is one of the most difficult when hiring employees. By not paying, you can’t find yourself a suitable person or find an employee with too low competencies. The other side of the coin - you can overpay by spending, in fact, extra money, which on an annualized basis, including taxes, will amount to a considerable amount.
To establish the very “tasty”, but at the same time adequate salary, you can use statistical information and salary reviews from consulting companies, recruitment agencies and job sites. Unfortunately, most of the instruments existing on the market have one serious drawback: they are not oriented to a specific vacancy. Roughly speaking, they measure the average temperature in a hospital. Our service does not have this drawback.
The team approached the task with the understanding that we are sitting on a mountain of information. Still, 10 million resumes are a fairly large amount of data for analysis by various sections. We implemented this analysis in real time on the form of creating a vacancy:
Statistics are built on the basis of the expected salaries indicated by applicants in the resume. When the employer, when publishing a vacancy, enters the proposed salary, a graph is displayed to the right of the input field with the distribution of applicants by expected salary. It highlights the range of job seekers who could potentially be interested in a job with such a salary (job seekers who indicated an expected salary similar to that offered in the job). When updating data, the chart is rebuilt using unobtrusive animations that accentuate the user's attention at the time of recounting on the chart.
Process of creation
We started developing this feature on two fronts at the same time: designers worked on the appearance, and developers on the backend and test visualization of the displayed data.
From the point of view of development, the main risk was the possibility of a lack of data, because quite often we need to read statistics on rare specialties, and even in small cities. As a test, we tried to make a visualization for the position “manager in a shoe store”:
There was enough data, but there was an understanding that such “fences” are not very suitable for analysis - you need to either smooth out or present the data in the form of a histogram. Our designers, meanwhile, have already made the first sketches of the future interface (we do not pay attention to the currency - this is just a “fish”):
We realized that on the histogram it is very difficult to display the range of salaries so that it is understandable to the user. Therefore, another option appeared, which turned out to be closer to the final one: It
remained to concentrate the user's attention on the fields with salary and remove information noise:
we did the frontend part using the Raphaёl JS library , which allowed us to support even fairly old browsers (Firefox 3.0+, Safari 3.0+ , Chrome 5.0+, Opera 9.5+ and Internet Explorer 6.0+) and to avoid troubles with the implementation of SVG and VML for Internet Explorer.
Making salaries even smarter
In the process of demos, we realized that it would be nice to give the user a text prompt about the salary range that he indicated:
These messages can solve the problems that I described at the beginning of the article.
To be done
I want to say right away that statistics on salary expectations for employers when creating a vacancy is only the first step towards standardizing the salary market. We have many ideas where such statistics would be useful (including for applicants), so stay tuned!
PS Your constructive suggestions and comments can be written in PM or on my mail b.volfson (dog) hh.ru, well, you can troll in the comments :)