Central & Eastern Europe Developer Landscape

Central & Eastern Europe Developer Landscape Prepared for Google Developer Relations 12 December 2016

Table of contents ☉ Methodology notes: 4–7 ☉ Developer populations in CEE: 8–14 ☉ Web development in CEE: 15–20 ☉ Mobile development in CEE: 21–27

2

Key findings ☉ There are roughly 1 million professional-grade software developers across Central & Eastern Europe (CEE) ▮ One-quarter of these are in Poland ▮ Across CEE, there are approximately 1.3 developers for every 100 people in the work force ☉ Across CEE, 15% of Web developers are using the Java Web stack ▮ Most popular Web development language in the region is PHP ▮ Java market share of Web developers has been steady over last 4 years ☉ Developer market share for Android is approximately 4% across the region ▮ iOS market share about half as much (2%) ▮ Top 5 cities for Android developer populations are Warsaw, Kiev, Athens, Budapest, and Minsk ▮ Android share has been decaying slightly over last 4 years ☉ Firebase hasn’t “taken off” in CEE or Russia as quickly as it has in the rest of the world — only 0.3% of developers in CEE are actively developing apps using Firebase, vs. 0.5% in Nordic markets and 0.8% in Latin America ☉ While it’s small (0.2%), the percentage of developers in CEE working on Windows Phone apps is about twice as much as the global average.

3

Methodology

Methodology and data sources The bulk of the data in this report are based on developers’ interactions with content on the English-language version of Stack Overflow ☉ Stack Overflow is the world’s largest website for software developers, with more than 40 million unique visitors a month. These visitors come to Stack Overflow to ask and answer technical questions with their peers in our community, which make up 95%+ of developers globally. The average visitor comes seven times a month to either consume or contribute to content. The sum of this activity is more than 20 million page views a day during the average work week. ☉ When developers visit Stack Overflow, our database records their IP and MAC addresses, and their web browser is offered a cookie to track their visitation history. By combining this information, we’re able to build a user history that spans multiple devices and working locations, regardless of whether the user is logged in during a particular visit (or, for that matter, has registered for a login at all). ☉ At the same time, we record which technologies are tagged on the questions each developer is visiting. Our machine learning platform uses the history of tags a developer has interacted with over the last 30 days to assign probabilities that he or she is of a particular “kind” (such as a mobile developer, or systems administrator), and working on a particular stack (such as the .NET stack). ☉ This allows us to understand exactly how many professional-grade software developers are in a given area, where they are based (for home and work, so sometimes also their commute between the two), the types of developers they are, and the technologies they use.

5

We mined Stack Overflow user activity data in four ways ☉ Cross-sectional estimates of most developer “kind” and “stack” populations were generated by our machine learning platform, which classifies users using a multi-step process: first determining an overall kind, e.g. “Web developer,” then a sub-kind, e.g. “front-end Web developer,” and finally a separate classification of the stack(s) the developer is working on. These classification models apply to all visitors to the main Stack Overflow English-language site, regardless of whether they are registered users or logged in. The bulk of the data in this report and the accompanying spreadsheet are outputs from this system. ☉ Cross-sectional estimates of Firebase and Windows Phone developer populations are not available in our machine learning platform, so these were done as “ad hoc” developer kinds. We looked at registered Stack Overflow users (a subset of all users) and classified them as a “Firebase developer” if 5% or more of their traffic in the last several months had been to content with a Firebase-related tag. Similarly, we classified “Windows Phone developers” if 10% or more of their traffic had been to Windows Phone-related content. (The threshold for Windows Phone is higher to account for the fact that traffic to an OS tag tends to be higher than traffic to framework and other supporting tool tags.) ☉ Time-series trends in developer kinds and technical stacks were generated by looking at a random sample of 25% of all traffic over the last five years. Rather than counting “heads” the way the classification models do, this technique aggregated all traffic in a region or country and looked at to what extent it was Java-oriented, Python-oriented, etc. These time series measures were then normalized to the canonical percentages of developers of each kind and stack generated by the machine learning platform. ☉ Time series trends in interest in Firebase and related technologies were generated by looking at the number of questions asked on the English-language Stack Overflow related to each tag, on a by-region, by-quarter basis. (While questions are a different quantity than people, we’ve found in the past that the number of questions asked and the number of people consuming that content tend to be highly correlated.)

6

Data definitions and notes of caution ☉ What we mean by “developer”: We classify someone as a “developer” for purposes of this analysis if their usage of Stack Overflow is frequent enough to make us think they’re coding for a living or planning to some day. In addition to professional software developers, the data do include but also university (and, in some cases, high school) students and intensive hobbyists. In most cases, however, our classification engine is able to detect the patterns associated with these more casual users and filter them out from our counts. ☉ What we mean by “mobile developer”: We classify someone as a “mobile developer” (sometimes referred to in this report as “developer specializing in mobile apps”) if a large proportion of the content they’re viewing on Stack Overflow is tagged with either a mobile operating system, a mobile SDK, or similar technologies that are meant specifically for building mobile device applications. This spans Android, iOS, and Windows Phone. Developers who work on cloud back ends, databases, or other services that may be called by a mobile app are classified separately (primarily as “back-end Web developers” or “full-stack Web developers,” depending on the extent to which they use front-end technologies). Developers who work on mobile-enabled websites may be classified as designers, Web developers, or mobile developers, depending on their individual content usage. ☉ Someone can be two things: Our machine learning platform doesn’t tag a given individual with a single classification. Rather, it assigns each visitor a probability that they’re a Web developer, a mobile developer, etc. In some cases, a developer may be counted as both a full-stack Web developer and a ☉ Someone can be two places: Many Stack Overflow users login from both their offices and their homes, and we record them as “present” in both places. Some users also VPN in such a way that we think they’re somewhere they’re not. This report presents data at the country and regional level, so these “migrations” don’t have a large effect on the findings. ☉ English only, for now: The data in this report are based solely on the English-language version of Stack Overflow. Data from the recently-launched Russian-language Stack Overflow may be incorporated into future reports. In general, we find that developers in foreign language markets do view content on the English Stack Overflow, but they’re less likely to contribute new questions, answers, and comments than their peers in Anglophone countries.

7

Developer populations in CEE

There are just over 1 million developers in CEE Nearly a quarter (24%) are based in Poland; additional 35% in Ukraine, Romania, and Czech Rep.

Developer population of Central & Eastern Europe (thousands)

☉ Central & Eastern Europe’s developers account for 6% of the global total.

300

254 250

200 166 150 118 100

100 80 64 51

50

47

45

38

34

27

24

20

18

0

Poland Ukraine Romania Czech Hungary Greece Bulgaria Serbia Belarus Slovakia Croatia Lithuania Slovenia Estonia Republic

Latvia

There are 1.3 developers for every 100 people in the labor force across CEE Slovakia and Estonia have especially high densities of developers Developers per 100 people in labor force

☉ Relatively low ratios in Ukraine, Belarus, and Romania may be due to large numbers of developers in those countries preferring to consume content in Russian or another language. (Activity from the Russian-language Stack Overflow is not accounted for in our data.

3.7

3.0

1.9

1.8

1.8

1.8

1.7 1.5

1.5

1.4

1.3

1.2 1.0 0.9 0.7

Slovakia Estonia

Czech Croatia Hungary Lithuania Latvia Republic

Bulgaria Serbia

Poland

Greece Romania Belarus Slovenia Ukraine

10

Relatively high proportions of developers in CEE are in the capital cities of their respective countries By comparison, roughly 7% of developers in the USA are in the Washington DC area Percentage of all developers in country in capital city metro area, CEE 93% 85%

79%

75%

75%

73%

73%

70%

64%

62% 50%

50% 42%

41% 29%

11

There are 68k mobile developers and 164k back-end Web developers in CEE 2.4 back-end specialists per mobile developer in region

Comparison of Mobile Developer and Back-end Web Developer Populations in Central & Eastern Europe (thousands) Mobile developers

☉ Lowest ratio is Ukraine at 1.8:1 ☉ Highest ratio is Estonia at 4.4:1

Back-end Web

45 39

40 35 30

26 25 20 15 10 5

17 15

15

15

12 9 4

5

4

8

8

7 4

3

7 3

5 2

5 2

4 1

4

4 1

1

3 1

0

Poland Ukraine Romania Czech Hungary Greece Belarus Bulgaria Serbia Slovakia Croatia Lithuania Slovenia Estonia Republic

Latvia

12

Highest metro-level population of mobile developers in Kiev and Warsaw Comparison of Mobile Developer and Back End Web Developer population in Eastern European cities Mobile developers

Back-end Web

14,000 12,000 10,000 8,000 6,000 4,000 2,000 0

13

Distribution of Mobile and Web developers remains similar across CEE, mirroring global trends Percentage distribution of Mobile, Web and other kinds of developers in Eastern Europe Mobile developers

Web developers

100% 90% 80%

75%

74%

74%

66%

70%

68%

65%

72%

68%

73%

68%

75%

69%

67%

72%

67%

60% 50% 40% 30% 20% 10%

5%

9%

4%

4%

6%

6%

6%

5%

6%

6%

7%

6%

6%

9%

5%

0%

14

Web development in CEE

PHP is the most popular Web language throughout CEE The exception is Belarus, where Java has taken its place

Penetration of popular Web stacks in CEE 30%

Full Stack Web Java 25%

20%

Full Stack .Net Full Stack Node.JS Full Stack PHP

15%

10%

Full Stack Python Full Stack Ruby Full Stack Wordpress

5%

0%

16

In terms of Web stack developer shares, CEE resembles SE Asia and LatAm more closely than the rest of Europe Percentage developer share for most common Web stacks, all regions 30%

25%

20%

Java .NET

15%

Node.JS PHP

10%

Python

☉ While 16% of developers globally use Java for Web development, this ranges on a regional basis from 12% in Sub-Saharan Africa to 19% in North America. ☉ Though the Java stack is the #1 most commonly used globally, it isn’t the #1 language of choice in any given region — rather, it “wins” globally by being the #2 choice in the largest regions (North America, Western Europe, and SE Asia). ☉ More developed markets (e.g. North America, Nordics) make greater use of Python for Web development, while many emerging markets (e.g. LatAm, SE Asia, Sub-Saharan Africa) are using PHP more heavily,

Ruby On Rails 5%

0%

17

.NET losing developers to Node, Python in CEE

Developer market shares of common Web development stacks in CEE, 2012–16 25%

PHP

20%

.NET Java

15%

Node.JS Ruby 10%

Python

5%

0%

2012

2013

2014

2015

2016

18

Node.JS gaining developer share throughout CEE Fastest gains and highest penetrations in Ukraine, Belarus 21.0%

Ukraine Belarus

19.0%

Bulgaria Latvia Estonia

17.0%

Serbia Poland

15.0%

Lithuania Slovakia

13.0%

Croatia Hungary

11.0%

Slovenia Czech Republic Greece

9.0%

7.0%

2012

2013

2014

2015

2016

19

Python usage for Web development growing throughout CEE Strongest adoption growth in Estonia, Slovakia

Percentage of developers who use Python for Web development Central & Eastern Europe, 2012–16 21.0%

Estonia Slovakia Croatia

17.0%

Greece Czech Republic Poland Slovenia

13.0%

Hungary Ukraine Belarus 9.0%

Latvia Serbia Lithuania Bulgaria

5.0%

2012

2013

2014

2015

2016

20

Mobile development in CEE

Roughly twice as many Android developers as iOS developers in CEE Both Android share and relative share vs. iOS in line with global average

Percentage of developers who work on Android vs. iOS apps, all regions Android

iOS

8%

7%

7%

7%

6% 5%

5% 5%

5%

2% 1%

4%

4%

4% 3%

6%

5%

3%

4%

4%

2%

2%

3%

3% 2%

2% 2%

2%

2%

2%

2%

3%

2%

1%

0%

22

Kiev and Warsaw have the highest metro-level populations of Android developers iOS and Android developer populations in Eastern European cities Android developers

iOS developers

3,500

3,000

2,500

2,000

1,500

1,000

500

0

23

Share of developers who specialize in Android declining slightly across CEE Percentage of developers in CEE who specialize in Android apps, 2012–16 6.0%

5.5%

Belarus Ukraine

5.0%

Serbia Greece

4.5%

Croatia Slovakia Poland

4.0%

Slovenia Hungary

3.5%

Latvia Bulgaria

3.0%

Lithuania Czech Republic

2.5%

2.0%

Estonia

2012

2013

2014

2015

2016

24

Share of developers specializing in iOS apps steady over last 5 years Significant decline in Belarus from 2012 to 2013; steady afterwards

Percentage of developers who specialize in iOS apps CEE, 2012–16 Belarus

5.0%

Ukraine 4.5%

Serbia Slovenia

4.0%

Croatia Latvia

3.5%

Bulgaria 3.0%

Slovakia Hungary

2.5%

Poland Greece

2.0%

Estonia 1.5% 1.0%

Czech Republic Lithuania 2012

2013

2014

2015

2016

25

Firebase adoption in CEE significantly lower than every other market except Russia LatAm Firebase share about 2X higher than CEE’s Percentage of registered Stack Overflow users who are developing apps using Firebase, by region Error bars show 95% credibility intervals 1.20%

1.00%

0.80%

0.60%

0.40%

0.20%

0.00%

Sub-Saharan Latin America Africa

MENAT

Other APAC

Southeast Asia

Nordics

North America

Western Europe

UK/IE

Central & Eastern Europe

Russia

26

Significantly higher share of CEE developers work on Windows Phone vs. rest of Europe Percentage of registered Stack Overflow users who develop Windows Phone apps, by region Error bars indicate 95% credibility intervals 0.40% 0.35% 0.30% 0.25% 0.20% 0.15% 0.10% 0.05% 0.00%

27

Thanks! Any questions? John Muskett, Senior Consultant [email protected] Kevin Troy, Director [email protected]