The IT skills market continues to be one of the hottest labor markets in the world, with the Bureau of Labor Statistics projecting a 25.6% increase in IT job growth in the US alone in the next 10 years.
What is clear is that in order to effectively compete in this increasingly competitive area, recruiters and hiring managers need granular data about developers’ IT skills for each position they want to fill.
Our core mission is to give tech recruitment decision-makers the tools and information they need to make the best hires. As part of our commitment to the industry, this is the second year we’re publishing the aggregated data from our platform. You can compare this year’s results with the results in our 2019 report.
As you read the report, you’ll find that some areas have not changed a lot like the IT skills that companies seek out. Other areas have made huge swings since the last year like the US’s position as an international IT skills labor market. To measure the current customer base, the report this year contains a larger dataset from 213,782 coding tests sent to developers in 143 countries by companies in 49 countries.
This year’s report also explores new topics such as the most popular technologies in tech stacks and whether recruiters are using tools to tackle unconscious bias in recruiting, all designed to give you the right information when hiring developers.
Table of contents
& IT skills insights
We looked at the top eight languages tested together this year and found that there is actually been quite a lot of change over last year.
THE TOP 8 LANGUAGES TESTED TOGETHER
React, Spring, ASP.NET, MySQL, HTML, Data Analysis, and Laravel are the most popular technologies in their respective tech stacks*
One of the central tenets of RealLifeTestingTM is that developers should be tested not only in the languages that they’ll use but also in their ability to understand the tools and resources of the tech stack that they will need to work with. Because our tests are created with the tech stack in mind, we can see what the most popular IT skills, resources, and technologies are in the most popular tech stacks.
|3. Spring Boot||16.02%|
|3. Entity Framework||27.74%|
|4. .NET Core||18.07%|
|5. SQL Server||11.96%|
|2. SQL Server||35.99%|
|2. WCAG 2.0||25.53%|
|3. HTML 5||17.37%|
|1. Data Analysis||29.76%|
Spring is used in 47% of Java tests
ASP.NET is the most popular .NET/C# technology, used in 55% of tests!
ASP.NET, the popular web app framework is the most used technology in the .NET tech stack. ASP.NET’s position shows the importance of web development to the .NET/C# tech stack. MVC architecture and Entity Framework are used in 42% and 28% of the tests respectively. Coming in fourth is .NET Core, .NET’s open-source, portable version which is gaining popularity. Rounding out the top 5 is SQL Server, .NET’s server technology.
MySQL leads the SQL stack with 37.3% of the tests
In the database field, there’s a pretty even split. MySQL is used in almost 40% of SQL tests while SQL Server is used in about a third. PostgreSQL is used in about 10%, about the same as Oracle. HSQLDB comes in fifth with 8%. This last entry means that recruiters are testing general SQL knowledge that is not related to a specific server.
CSS and HTML go together with HTML tested in 63% of tests
Once again, it is not surprising that HTML figures strongly in the CSS tech stack, as CSS and HTML are closely intertwined. Probably the most interesting thing is the fact that HTML is still tested as a separate entity. This proves that while it might be easy to learn, it requires specialist skills to master. After that, the accessibility standard WCAG 2.0 is used in about a quarter of all CSS tests. While not a technology, accessibility is an increasingly important consideration in web development. HTML5, HTML’s more media-focused version comes in at 18%, while CSS framework Less and alternative style sheet language Sass come in at positions 4 and 5 respectively.
Data analysis is used in 30% of Python tests
Data science is a major application for Python so it is not surprising that Data Analysis is the most common IT skill in that stack. Python’s other main application as a server-side web development technology is the second most popular with Django appearing in roughly one out of every 5 coding tests. NumPy, Pytools, and Pandas come in third, fourth, and fifth respectively.
PHP is not dominated by one resource with Laravel used in only 28.3% of tests
In the PHP stack, Laravel is the strongest resource. Interestingly, there are not as many overwhelming technologies in this stack with Doctrine, CodeIgniter, Symphony, and PDO only used in relatively small percentages of all PHP tests.
*Updated on January 20, 2020
Candidates wait on average 2.27 days to take a coding test
The average time a candidate waits between retrieving a coding test invitation and taking the test is 2.27 days. This is down from 2.88 days in 2018. The drop in time suggests that candidates are taking coding tests more seriously as a part of the hiring process. Rather than putting coding tests off, developer candidates are acting on them more proactively. This could be in part an effort on the candidate’s part to shorten the hiring process and field multiple offers at once.
Coding tests sent on Wednesdays get the fastest response
Coding tests sent on Wednesday get the fastest responses. This is a sharp reversal from last year when coding tests sent out on Wednesday had the slowest response. What we do see though in section 5 is a shortening of the average time it takes to send back a completed coding test. From this, we can assume that candidates are most likely to take the test at the same time of the week (i.e. Friday evening) but are acting on them more quickly.
THE AVERAGE WAIT IN HOURS BASED ON THE DAY OF THE WEEK THE INVITE IS SENT
Less experienced tech recruiters are taking extra time to develop themselves
Devskiller’s Tech Recruitment Certification course is proving to be sought after by even more tech recruiters than last year with roughly double the attendance. An interesting outcome, though, is that tech recruiters have been needing more time to pass the course and get their shiny LinkedIn badge.
THE TIME IN DAYS IT TAKES RECRUITERS TO BECOME TECH RECRUITMENT CERTIFIED DEPENDING ON THE NUMBER OF TRIES THEY NEED OT PASS
THE NUMBER OF TRIES IT TAKES RECRUITERS TO BECOME TECH RECRUITMENT CERTIFIED
Over half of the people who pass the course now do so after two tries compared to one try last year. They do so over a longer period of time, almost 60 days. In fact, the average time recruiters take to pass the course in two tries is about a week longer than it takes those who pass the course on three tries. This shows that when recruiters don’t pass the first time, instead of taking the test over and over, they take the time to look at the material again.
Recruiters are slow to take up bias reducing technologies
Interviewer bias is an important issue in tech as candidates from non-traditional backgrounds are often eliminated for unconscious reasons. Interestingly, our data finds that tools that can be used to eliminate this bias are still not being taken up by most tech recruiters.
Devskiller’s feature to anonymize candidate reports for hiring managers was only used in 6% of the reports generated.
6% of reports
This means that most tech recruiters are missing a huge opportunity to objectively evaluate candidates on their IT skills without taking irrelevant personal details into account. This is one metric that we would like to see an increase in next year’s report as it will benefit employers, candidates, and the industry as a whole.
The geography of
IT skills recruitment
The US is doing more overseas technical recruitment while the UK and Germany are both doing less
International hiring for tech workers continues to be a hot trend for companies all over the world. Many countries have seen their companies looking further outside their borders for tech talent than last year.
The US, France, and New Zealand have all seen the percentage of their international technical hiring go up. Other countries dealing with possibly slowing economies and political uncertainty like the UK and Germany have both seen their international technical recruitment rate decrease.
THE PERCENTAGE OF INTERNATIONAL TECHNICAL HIRING BY COUNTRY
The US recruits a quarter of all overseas developer candidates
The US and Poland still remain the top two drivers of international technical hiring, but their overall percentage has gone down compared to last year as we widen our reach to companies in more countries.
The UK, which was previously the third-largest driver of international tech recruitment, has dropped to 6th place. This is possibly because of the uncertainty in their access to an international labor pool caused by Brexit. In their place, Saudi Arabia has risen to the number three spot, possibly based on the new focus on technology in that country.
THE PERCENTAGE OF TOTAL OVERSEAS CANDIDATES TESTED ON OUR PLATFORM
The US is the largest international recruiter but is only the 4th largest labor market for overseas recruitment
In our report last year, we established that certain countries like the US do a lot of external technical hiring while also being major labor pools for other countries. This year, we see that position changing noticeably. The US continues to be the largest recruiter of overseas talent in the world, but it is no longer the second-largest pool of talent. This is less a testament to the desirability of US developers and more to the increasing desirability of other countries outside the US. India, Brazil, and Egypt are becoming increasingly larger sources of overseas candidates, with the US now coming in fourth.
In fact, when you look at the top countries that the US recruits from, predictably the top three are India, Brazil, and Egypt, followed by Mexico, and Argentina. What we see is Brazil and Egypt emerging as major sources of outsourced talent. Additionally, the timezone proximity of Brazil, Mexico, and Argentina clearly makes developers based there more attractive to US companies.
What we see is in addition to an efficient allocation of IT skills among developed economies, there is a growing importance of outsourcing-focused countries.
Click on a country to find out where they recruit from internationally and who recruits their local developers
Latvian developers score the highest (54.65%) on coding tests
Latvian developers score the highest on coding tests with Latvian developers scoring on average 54.65%, followed closely by developers from the Netherlands and Italy.
COUNTRIES WHOSE DEVELOPERS SCORE THE HIGHEST ON CODING TESTS
While this is a great result for the skill and quality of Latvian developers, it also suggests that coding tests are being treated differently by recruiters. Instead of trying to make coding tests incredibly difficult, recruiters are using coding tests which are a better reflection of the IT skill level required for the job. As a result, developers are scoring higher on them.
Italian companies have the highest scoring candidates (57.14%)
The average score of the candidates tested by Italian companies is 57.14%.
These results bolster the same thesis about test difficulty level we’ve mentioned in section 12. Italian companies are adjusting tests to be more reflective of the skill level required for the job rather than giving developers excessively challenging tests. At the same time, the country’s relative wealth and small talent pool allows them to seek the best developers from around the world.
Companies from Israel are the most selective
As we’ve established in sections 12 and 13, companies are now targeting their tests more closely at the skill level of the positions that they are hiring for. At the same time, they are becoming much more selective. Israeli companies are the most selective this year, accepting for consideration only 12.26% of the developers they test.
That is compared to nearly 20% by Singaporean companies, who were the most selective last year. Luxembourgian companies, who attract the highest-scoring candidates, are the second most selective. They accept for consideration only 16% of their candidates.
What this shows is that tests that are targeted at a more appropriate level are attracting many more candidates. The ability to test a wider spread of candidates has given companies the opportunity to be a lot more selective in who they accept for consideration.
THE PERCENTAGE OF CANDIDATES ACCEPTED BY THE RECRUITER BY COUNTRY
Companies are sending out coding tests to a wider section of candidates than last year
As in-stack coding tests using the RealLifeTestingTM methodology becomes more and more appreciated, we see more recruiters sending them to a wider range of candidates. In some countries, almost every candidate takes the test they are sent but the overall test-taking rate has gone down compared to 2019.
This reduction in the rate candidates are taking coding tests is because recruiters are becoming more comfortable with moving and automating the tech screen to the beginning of their tech recruitment process. Where last year we would have seen recruiters do a phone screen or resume screen before a coding test, more recruiters this year are becoming confident enough with the results of a Devskiller test to make it one of the first interactions with the candidates.
THE COUNTRIES WITH THE TOP UPTAKE RATES
Since the candidates are less invested at this point, it makes sense that fewer would take the coding test sent to them. As a result, we’ve seen a drop from 73% of candidates taking the test has sent them to only 41% taking them.
Still, certain countries like Estonia still have very high rates of candidates taking the tests they are sent.
93% of developers globally finished the coding test sent to them for recruitment purposes, up from 91.9% last year
This means that when challenged with meaningful in-stack coding tests, developers are willing to devote the time it takes to complete the challenge. In-stack coding tests, like the ones found on Devskiller, are no longer treated with the same level of suspicion developers have historically had for algorithmic tests.
THE PERCENTAGE OF DEVELOPERS WHO FINISH THEIR CODING TEST
A full 93% of developer candidates completed the coding test they started, compared to 91.9% last year. Are there some variation among countries, though. The lowest rate came with Madagascar (76%), which is probably more down to the state of their internet infrastructure than the displeasure that local developers have for taking these tests.
By comparison, 97% or more of the candidates from Ecuador, Bulgaria, Azerbaijan, Ghana, Kazakhstan, Uganda, Jordan, Hungary, Bosnia and Herzegovina, Latvia, The Netherlands, Ireland, Belarus, and North Macedonia completed the tests they started.
The methodology we used for this study
The data we used for this study came from a 365-day snapshot of users on our platform between December 1st, 2018 through December 1st, 2019. The insights are based on 213,782 tests taken through the Devskiller platform by candidates in 143 countries. All data presented here is generic aggregated demographic information. It is not linked to any specific information regarding certain candidates or companies.
*In sections 1 and 2 the percentages don’t add up to 100%. Why is that?
A Devskiller test can include multiple technologies. For instance, you could have a test in Java and a test in Java+SQL. In this sample, 100% of the tests test Java and 50% of the tests test SQL. In the same way, the percentage in the chart refers to when the technology is tested in any test.
*In sections 12 and 13, why isn’t China on the maps?
We don’t have a lot of data about mainland China so while we included the countries that send invites to mainland Chinese developers in the interactive map, we eliminated China and other countries where we had a small sample size from our maps in sections 12 and 13.
Did you find this report valuable?
Share it with your friends and colleagues.