How we transformed primary validation in mass recruitment
Why we decided to move the prescorring process online
Avito Jobs is an app for searching jobs and ‘blue-collar’ workers.
Working on this product we constantly communicate with drivers, cooks, movers, waiters etc. On the other hand we keep in touch with big employer’s recruiters who specialise in mass recruitment. Both empathise that it’s important not just to find a job or a worker but to do it as soon as possible.
People need to feed their families and recruiters need to implement hiring plans in order not to keep manufacturing idle. If time to hire is crucial to our users, it is crucial to us.
In order to build this metrics it is necessary to create a clear hiring funnel from beginning of job search to starting a job.
For now Avito Jobs provides only part of this funnel: from beginning of job search to responding to a vacancy. Another part exists offline only. We need to move that part online in order to concentrate our future initiatives on reducing time to hire.
From business point of view building of the hiring funnel is a way to change monetization model from generating traffic to cost per action.
We need to build functionality for:
— primary candidate validation
— invitation to interview
— interview
— starting a job
This story is about building primary validation functionality.
What is the process of primary validation
User scenario on Avito Jobs is the following. Candidate finds a suitable vacancy and responds to it via phone or via chat. As a rule in case of a chat response a recruiter asks to provide a phone number. A fair amount of chats between recruiters and applicants contain words ‘phone’, ’number’, ‘call’. So anyway the conversation goes offline and primary validation gets place offline as well.
In order to build the whole process online correctly I decided to go thought it offline by myself. I responded to forty vacancies for ‘Cashier’ position. One day I responded to twenty of them. Recruiters asked me to provide a phone number or to call back.
I called back and all calls were the same. First recruiters described the vacancy once again going deep in such detail as schedule, length of work shift, payout frequency and shift / weekly and monthly rates.
Then I was asked same questions:
— my name and surname
— age
— citizenship
— preferred geo location of my future job
Satisfied with my answers recruiters immediately invited me to interviews.
The second part of my research was a little bit different. This time I sent a brief information about myself immediately after response. I wanted to make sure this information is enough for recruiter not to repeat same questions over the phone but invite me to interview instead. I responded to another twenty vacancies and this time my response looked this way.
I proved my hypotheses. 18 out of 20 recruiters called me and invited to interviews immediately. However, there were two exclusions: Beeline, the Russian mobile operator and KFC, the fast food chain. They probably have very strict hiring process so recruiters don’t step out of it. For these employers I repeated all answers once again.
Primary validation via personal information attached to each response in messenger
We formulated a low-level problem in the following way: recruiters spent significant time on primary validation via phone. Taking into consideration the daily amount of responses (tens to hundreds a day) this turns out to be a problem. So we decided to ask candidates a permission to attach their personal information to each response automatically. This information could be extracted from CVs our users created previously.
Only 15% of active users create CVs. But this amount is enough to prove the usefulness of the product using statistically significant data. If it turns out to be a success we plan to ask all other active users to share their personal information via a chat bot I described previously.
We invited five job seekers to a UX-research and showed them prototypes. The main difficulty turned out to be with text length and contents. First of all the text was too long. People either didn’t read the text at all or read it inattentively so they couldn’t predict further interface behaviour. Above all people didn’t quite understand what exactly personal information will be shared with an employer and when it will happen. So we decided to replace greetings and common words with exact information from a candidate’s CV.
70% of recruiters use desktop web-version of Avito Jobs to process vacancy responses. It’s impossible to call from desktop despite you have a call-tracking system. We didn’t have one at the moment. In order to measure the usefulness of CV attachment functionality on desktop we made a button of phone number request.
The primary validation is strategically important in order to build a time to hire metrics. Without prescorring a recruiter won’t be able to validate whether a candidate suits or not and won’t be able to invite him to interview and offer a job later. So personal information attachment to each response is the first step on this way.
Launch and further estimations
We launchув an AB-test in May 2020 and gathered results within a month period. In 40% of chats with personal data attachments the ‘Show phone number’ button was tapped.
Thus we may see that this functionality is in demand so we may proceed to the next steps along the way of this product development. We plan to contact all applicants who don’t have CVs via our chat bot and collect their personal data. Taking into consideration the performance of the chat bot this initiative will give a significant grow.
40% of those contacted by chatbot create a CV. We aim to contact everyone who responds just once. This is 1M users a month. Thus we will have 400K of chats with personal information attached. 40% of them will get contacts request. That means 160К contacts requests a month. This is enough to start working on a next step of a hiring funnel — invitation to an interview.
