Dogfooding is an English slang term for software companies using their own products.
Today, we'd like to share some of the data changes that ShowMebug's recruiting team uses to use their own products.
ShowMebug was officially released in September this year with a technical evaluation version with a job model, and our showmebug team has been using it internally since August. Our products using showmebug will be divided into two scenarios: one is to add a technical assessment link during the initial screening of resumes. The HR recruiter communicates with the candidate first, and then sends a technical assessment link, and after the candidate completes the assessment, the candidate who meets the assessment results will be interviewed by the interviewer. Another scenario is to use showmebug during the interview process. We don't require candidates to be on-site, and we even encourage interviews with Showmebug's interview feature. In this scenario, we generally have 3 rounds of interviews: the first round of technical interviews, the second round of project interviews, and the third round of CEO interviews (there are also technical components). Thanks to the help of showmebug coding and architecture drawing, the online interview is better than the offline one, and the programming ability can be more examined. In the past few months, we have focused on recruiting J**A engineers (PaaS direction), React front-end engineers, and general junior engineers (skills assessment direction). Let's take a look at the before and after changes in the technical assessment during the initial screening. In the first stage, before the introduction of ShowMebug, combined with the early experience of our CEO Li Yafei (who previously served as a CTO technical partner of several companies), HR recommended an average of 5-6 resumes, and the technical colleagues would arrange an interview. Often, many of them have a wet resume, and the actual interview may take 10-20 times before an offer is made. In other words, it takes an average of hundreds of referrals to make an offer。In the second stage, before the release of the showmebug job model, we used the ** written test function of showmebug, and the assessment questions were grouped in the form of internal questions + public question bank, and the initial screening was carried out in accordance with the above method, which can reduce the number of invalid interviews by about half. But there was a very big pain point at the time:When the candidate's result comes out, it is a score, and the criteria for evaluation are not very clear, and you can only look at experience. In the third stage, after the technical assessment version of showmebug based on the job model was released and launched, we threw the previous assessment questions directly into the trashUse showmebug's Smart Roll feature directlyAccording to the characteristics of our job model, we fine-tuned and selected the Linux operating system with higher capability requirements, J**A capability requirements with medium capability requirements, and Spring Framework capability requirements with medium capability requirements on the standard model. Here's what we did after about 2 months of data. 
Let's take the J**a position as an example to interpret it:
When recruiting for J**A back-end positions, we had a total of 64 valid resumes and preliminary communications, and arranged for the issuance of papers, of which 16 were passed, with a pass rate of 25%. The first part is all done by HR and does not require the involvement of the interviewer.
Before arranging one, HR will send resumes, evaluation reports, and reasons for recommendation to one interviewer, 9 J**A posts have one side, and 5 have two sides through one side. To be honest, we were shocked when we got the data, this data is so valuable!Originally, under the traditional technical recruitment method, the conversion rate of resumes to one side was frighteningly low, often no more than 20%, and even less through the "effective side" of one to two sides, only about 20%, and the combination of the two was 20% * 20% = 4%. In front of the screen, you can think about whether it is in line with the current situation of your business. And now the data shows thatThe pass rate of the assessment is 25%.(The latter two positions have reached more than 30%)What's even better is that after passing the assessment, the quality of talent has obviously increased significantly, more than 50% of the interviews have been arranged, and the pass rate to the second side has also exceeded 50%. In the end, through accurate talent screening, the recruitment effect can be increased by 4-5 times. To sum it up,A 30% pass rate for technical assessments is a particularly healthy signIt relieves the problem that HR often worries about the difficulty of recruiting people. At the same time, the data shows that the quality of talent after passing the technical assessment has increased significantly, which in turn greatly accelerates the final recruitment effect. After the assessment, showmebug will also generate an assessment result report with a rating, which fully demonstrates the candidate's real technical ability, displays each score from the two aspects of skill points and ability dimensions according to the candidate's answer results, and automatically generates the evaluator's job rating according to the evaluator's performance. Enterprises can clearly understand the strengths and weaknesses of the evaluators according to the assessment results report, and then carry out personalized employee training, and at the same time, the job rating given can also be an objective reference basis for their job classification. 
Assessment results with grading.
Seeing this, everyone will definitely be curious, how does showmebug's job model work?What job models are currently supported?How did you achieve such a high level of reliability and validity?
The ShowmeBug job model has been accumulated for a year and a half, and the positions that have reached full confidence in reliability and validity are: J**a engineer, Golang back-end engineer, React front-end engineer, Vue front-end engineer, as well as test engineer and operation and maintenance engineer also have a certain degree of reliability and validity support. In addition, more than 30 other positions have completed modeling and internal preliminary verification. The underlying of the showmebug post model isIt is composed of a massive question bank supported by skill trees and knowledge points。The question bank is built by a team of more than 7 technical evaluation experts, with the "practical programming" question type as the core, combining theory and actual combat, restoring business scenarios, accurately evaluating real strength, and carrying out multiple verifications with strict quality standards. At the same time, we also hire external industry expert consultants, and cooperate with well-known vendors such as JiHu (Gitlab) to jointly explore and polish the proposition methods and other content for specific positions. If your enterprise technology recruitment is facing high costs, high recruitment decision-making risks, low initial interview rate and other problems, please contact our senior consultants to arrange for you
It's completely free
POC (Product Verification). Improve your technical recruitment, from:NowBegin.