Mosaic.
Mosaic. (source: Jan Hrdina on Flickr)

Executive Summary

THE 2017 O’REILLY DESIGN SALARY SURVEY explores the landscape of modern design professionals, giving details about their roles and how much they earn. The results are based on data from our online survey that collected 1,085 responses. We pay special attention to variables that correlate with salary, but this report isn’t just about money—we present a range of information, including the popularity of design tools, tasks, and organizational processes.

In what is now our second salary survey, we find some consistency as to what matters in the field of design: that the better-paying design jobs tend to concentrate in tech centers; that experience matters more than age; that knowing more tools, working with more people in a wider variety of roles, and working for larger organizations all correlate with higher wages. And, in a sign that some things in the design world resist change (in some cases, whether we like it or not), we still see women making less than men and that most designers still use pen and paper as their primary design tool.

Note

Designers reporting no process earn the least

Some key findings include:

  • The West Coast (CA, WA, OR) has the highest salaries— salaries are high even relative to those states’ per capita GDP

  • Healthcare, banking, and computers/hardware respondents report the highest salaries

  • Respondents from large companies report higher salaries

  • Agile is the most popular design process; however, those using LeanUX or a hybrid of different processes earn the most

  • Designers reporting no process earn the least

  • Higher earners use a wider selection of tools

  • For prototyping and wireframing, salaries are highest among those that use Sketch

We hope that you will find the information in this report useful. If you can spare 5–10 minutes, please go ahead and take the survey yourself.

Introduction

FOR THE SECOND YEAR RUNNING, we at O’Reilly Media have conducted a survey for designers, gathering information about their compensation and details about their work. This year, 1,085 people from 48 countries took the survey. Respondents are mostly UX, product, and graphic designers, but there are also a fair number of developers and other professionals involved in product design. The survey was conducted online, collecting responses from December 2015 to December 2016.

Note

1,085 people from 48 countries took the survey.

While typical for online surveys, the methodology we used of a self-selecting, uncontrolled respondent pool can lead to less than ideal results. However, the broad range of respondents’ geographies, industries, and company sizes helps mitigate the issues associated with a small, narrow sample.

Throughout the report, we quote median salary statistics for various groups of people, such as those respondents who used a certain tool or came from a particular industry. Since these figures can be misleading if the variable in question correlates with geography or experience, we also sometimes quote a median “adjusted” salary. Technical details are in Appendix A: Adjusted Median Salary.

Note

In the horizontal bar charts throughout this report, we include the interquartile range (IQR) to show the middle 50% of respondents’ answers to questions such as salary. One quarter of the respondents has a salary below the displayed range, and one quarter has a salary above the displayed range.

Salary Overview

THE MEDIAN SALARY OF THE ENTIRE SAMPLE IS $77K, with the middle half earning between $50K and $109K. This “middle half” statistic is called the “interquartile range,” and we show it in many of the graphs to give a sense of the salary spread for various groups of respondents. The spread is very broad, but this isn’t surprising given the diversity of professional backgrounds among respondents and the fact that they come from locations with very different overall wage levels.

Most respondents report some gain in salary over the past three years, and about 10% of the sample saw their salaries double. Another third has wage growth of 30% to 100% (also across the last three years). Respondents who have no wage growth tend to have lower salaries (median of $50K), but otherwise there’s no clear pattern between salary growth (as a percentage) and current salary.

dss2017 1
Figure 3-1.
dss2017 2
Figure 3-2.

Geography

TWO-THIRDS OF THE SAMPLE IS FROM THE UNITED STATES, and 20% is from Europe. After the US, the most well represented countries in the sample are Canada (5%), the United Kingdom (4%), Germany (2%), Russia (2%), and Australia (2%).

For US-based respondents, a disproportionate share comes from the West Coast: 37% of the sample is from California, Washington, and Oregon (states containing only 16% of the US population). New York and Massachusetts also have disproportionately high response rates. The skew in the geographic distribution of respondents likely reflects that the O’Reilly audience and design-related jobs tend to concentrate in tech-centric coastal cities.

Salaries vary sharply across geography; however, in most cases, country and state variations mirror the local economy. Per capita GDP is a good predictor of salary, although some countries in Western Europe, including Italy, the Netherlands, Spain, and Portugal, have lower than expected salaries—likely reflecting how different recovery rates from the 2008 recession and different tax and social safety net regimes affect wages. In the US, the West Coast states (CA, OR, WA) have higher salaries relative to their per capita GDP, which, combined with their higher response rates, may indicate a relatively high demand for design jobs on the West Coast, helping to push up wages.

Note

In most cases, country and state variations mirror the local economy.

dss2017 3
Figure 4-1.
dss2017 4
Figure 4-2.

Age Versus Years of Experience

THREE-QUARTERS OF THE SAMPLE IS BETWEEN 26 AND 45, with about an eighth younger and an eighth older. Salary generally increases with age until age 50, where we see the 7% of the sample who are over 50 report a slightly lower median salary than respondents in their 40s.

Note

The 7% of the sample who are over 50 report a slightly lower median salary than respondents in their 40s.

Respondents were also asked how much experience they have, and, as expected, salary increases steadily with years of experience, although only up to a point: respondents with 20–25 years of experience earn more (median: $118K) than those with more than 25 years of experience ($101K). After we factor in years of experience, age doesn’t make any difference in salary among respondents aged 31 to 50. Holding experience constant, respondents younger than 30 do make a little less, but the difference isn’t as much as the age medians would suggest.

The lesson of age and years of experience tells us that we should be careful about confounding variables affecting our interpretations: being older (without anything else happening) may not increase your salary (at least not after 30), but having more experience will. While this example is fairly obvious, others are not, and in this report, we make an effort to avoid this same problem in more subtle contexts by occasionally referring to an additional metric we call “adjusted median.”

The adjusted median blocks the effects of geography and experience—creating a metric that estimates what the median would be if the respondents all came from a fixed location and all had the same experience—to make comparing factors more reliable. Appendix A contains additional details on the methodology. To illustrate, we show each five-year age category between 31 and 50 with an adjusted median salary of $80K, those between 26 and 30 with $73K, and those between 51 and 60 with $68K–$69K.

dss2017 5
Figure 5-1.
dss2017 6
Figure 5-2.

Gender

THE SAMPLE IS SPLIT FAIRLY EVENLY BY GENDER. The median salary of women in the sample is $82K, higher than the median salary of the men, which is $74K. However, the adjusted median salary of women is about $4K lower than the adjusted median salary of men. The discrepancy is accounted for by the fact that women in the sample are disproportionately from places with higher wages. In almost every geographic region (adjusting for experience), men are paid more than women on average, even at the same experience levels.

Men and women are equally likely to have received a raise in the last three years, but men are slightly more likely to receive a bonus: 46% of men in the sample received a bonus, while only 41% of women received one.

dss2017 7
Figure 6-1.

Industry, Company Size

RESPONDENTS COME FROM A VARIETY OF INDUSTRIES: software is by far the most common, with about a third of the sample, followed by consulting, advertising/marketing, and retail/ ecommerce. There are variations in salary among industry, although many of the differences diminish once we calculate the adjusted median salaries. When adjusted for experience and geography, healthcare, banking, and computers/hardware all have median salaries of about $89K, compared to an adjusted median of $74K for all other industries. Search/social networking is even higher, with a median salary of $127K (the adjusted median is also high: $96K), although this is based on just 2% of the sample.

Over one-third of the sample come from companies with no more than 100 employees, and another quarter come from mid-sized companies (in the 100–1,000 employee range). Salary does appear to go up with company size: from a median of $63K for 2–100 employee companies, to $96K for companies with over 10,000 employees. Again, these differences shrink slightly once we block out experience and geographical effects, but the 10,000+ group still shows an adjusted median salary of $10K–$15K higher than the other groups.

Note

Healthcare, banking, and computers/hardware all have median salaries of about $89K, compared to an adjusted median of $74K for all other industries.

dss2017 8
Figure 7-1.
dss2017 9
Figure 7-2.

Coding Time, Programming Languages

ONLY 43% OF THE SAMPLE reports that programming plays some role in their work. The other 57%, in fact, earn more, with a median salary of $83K (coders earn a median of $69K). However, this difference all but disappears when we adjust the salaries, since the distribution of respondents who code vary greatly over geography. For example, 58%–59% of respondents from Europe and Asia say that they spend at least some time coding, while only 30% of respondents from California code. Combined with a review of the 2016 Design Salary Survey, we don’t see code having more than a noisy impact on salary.

Note

58%–59% of respondents from Europe and Asia say that they spend at least some time coding, while only 30% of respondents from California code.

Among those that do code, most report spending at most 8 hours/week on the task, while only 9% of the sample report that they code more than 20 hours/week. The group with the highest adjusted median salary is the one that spends 1–3 hours/week, although the salary differences are fairly minor.

As for language choice, by far the most popular language is JavaScript (31% of the sample), with Python and Java coming next, with 8% and 7% shares of the sample, respectively. Swift users tend to have high salaries: their median of $95K is far above the sample-wide median, and this margin is decreased but still significant after geographical adjustment.

dss2017 10
Figure 8-1.
dss2017 11
Figure 8-2.

Tasks

A SET OF QUESTIONS ON THE SURVEY asks respondents whether they engage in certain tasks, either with “major” or “minor” involvement. Some tasks are nearly universal, such as brainstorming (74% major involvement, 21% minor involvement) and user interface design (64% major, 25% minor), and others are relevant to a much smaller subset of the sample, such as managing people (25% major, 35% minor) and data analytics (13% major, 46% minor).

We found three highly correlated tasks (wireframing, prototyping, and sketching), meaning that respondents who do one are more likely to do another. These are among the most common tasks, each with 85%–-87% of the sample (major or minor involvement), while 93% are involved in at least one of the three. The 40% of the sample that has major involvement in all three tend to earn aboveaverage wages (median: $87K). After adjusting the medians for experience and geography, this discrepancy holds.

A second set of tasks also correlates with one another: pitching, presenting, requirements gathering, leading design critiques, managing products, and managing people. For each of these tasks except product management, respondents who have major involvement earn more than those that do not. The differences in adjusted median salaries are similarly significant, around $10K for each.

While product management does correlate with these other tasks (in particular, with managing people), it doesn’t correspond to a boost in salary. While the median salary of those who manage products is higher than those that do not ($81K versus $76K), this difference disappears when we adjust the salaries for experience and geography. Furthermore, respondents who have major involvement in product management but do not have major involvement in managing people, have below-average salaries ($9K difference in adjusted median salary).

Note

Respondents who have major involvement in product management but do not have major involvement in managing people have below-average salaries.

dss2017 12
Figure 5-1.
dss2017 13
Figure 5-2.

Meetings

MEETINGS ARE A PART OF MOST DESIGN PROFESSIONALS’ WORK WEEK: 94% of the sample spends between 1 and 20 hours per week in meetings. As we’ve seen in past salary surveys, those who attend more meetings earn more. An effect we see even after adjusting for experience.

Respondents from the US tend to spend more time in meetings than those outside of the US. Among US-based respondents, 33% spend at least 9 hours per week in meetings, while this figure is only 20% for non-US respondents.

As we would expect, meeting times vary dramatically with job title. VPs and directors spend the most time in meetings, followed by project/product managers. UX designers and product designers each spend about seven hours in meetings per week, on average, while software developers/engineers and graphic designers only spend about four hours on average.1

dss2017 25
Figure 10-1.

1Note that these are rough average figures, since the original survey data was collected in binned ranges.

Working with Other People

MOST RESPONDENTS REPORT THAT THEY WORK WITH PEOPLE IN A VARIETY OF ROLES. Only 3% of the sample say they only work with (other) designers, and this group has a median adjusted salary of $56K, far below the sample-wide $77K. With five answer choices to pick from (designers, programmers, product managers, salespeople, and industrial designers), most respondents chose three or four.

The minority of respondents (7%) who work with industrial designers earn high salaries (median: $101K; adjusted median: $82K), which is likely related to the high wages in computers/hardware. Aside from this, no single answer stands out as having an effect on salary. However, it does appear that interacting with a wider variety of roles correlates positively with higher incomes: the adjusted median salaries of respondents who interact with one, two, three, and four of the listed roles are $60K, $68K, $77K, and $83K, respectively.

We also asked how many designers and programmers work at the respondents’ organizations. Most respondents work at companies with at least 5 designers and 20 programmers. Generally, the more programmers and designers at a company, the greater the salary, although part of this gradient may be attributable to company size, since there are higher salaries at larger companies, and larger companies tend to have more designers and employees.

However, even among subsets of respondents partitioned by company size, this pattern remains, at least for designers. For example, among respondents from companies with more than 1,000 and fewer than 10,000 employees, the adjusted median salaries of respondents who work with no more than 10 designers is $68K, while that of respondents who work with over 10 designers was $81K. Similar patterns are present with other company sizes.

Note

The minority of respondents (7%) who work with industrial designers earn high salaries

dss2017 14
Figure 11-1.
dss2017 15
Figure 11-2.

Types of Products, Products or Services

MOST OF THE SAMPLE WORKS ON BOTH PRODUCTS AND SERVICES, with 31% working on products only and 14% on services only. Respondents who work in products only have the highest salary (median: $87K; adjusted: $77K), and those who work in services only have the lowest (median: $61K; adjusted: $70K).

As for types of products, most respondents work on web products (83%) and mobile products (65%). Few respondents (3%) work only on mobile products. Many more work on web products only (22%), but these respondents tend to earn less (median: $70K; adjusted: $72K) than those who work on web products and a different type of product (median: $84K; adjusted: $78K).

Two other product categories (wearables and other connected devices) are less common, with 12% and 18% of the sample, respectively. However, respondents who work on one or both of these product types earn higher salaries than those that don’t, with a median salary of $90K (adjusted: $82K).

dss2017 16
Figure 12-1.

Design Process

THE TOP DESIGN PROCESS IS AGILE, with 45% of the sample. Design sprints are a distant second (17%), followed by waterfall (10%) and lean UX (8%). Practitioners of Agile tend to earn above-average salaries (median: $81K; adjusted: $79K), but not as much as those that practice lean UX (median: $89K; adjusted: $83K) or the 7% of the sample who use a hybrid or combination of design processes (median: $98K; adjusted: $83K).

A small subset (8%) report no design process, and these respondents tend to earn rather low salaries: a median of $48K, rising slightly to $53K after adjusting for geography and experience. These respondents are more likely to come from smaller companies with fewer designers, but this tendency isn’t absolute: about 5% of respondents from large companies (>1,000 employees) with over 50 designers say that they have no design process.

dss2017 17
Figure 13-1.

Tools

We asked about a variety of tool categories—for the most part, software—from prototyping and wireframing to project management and user research. The answer choices included 95 tools, but another 668 were entered in “other” fields: clearly there’s a lot of variety from which to chose from in design tools.

Respondents use an average of 12 tools, and salary generally rises with the number of tools used. Users of 7 or fewer tools have a median salary of just $58K, while those who use more than 7 but less than 14 have a median of $78K, and those who used 14 or more have $89K. The corresponding adjusted medians preserve this upward pattern: $66K, $73K, $81K. Even among designers with similar levels of experience, those that have a larger set of tools earn more.

Some of the most ubiquitous tools are project management/ collaboration tools, such as Google Drive, Dropbox, and Slack. Both the need to have consistent platforms across organizations and these applications’ ease of use are likely contributors to their higher usage rates. The variations in salary among users of project management tools are not significant.

dss2017 18
Figure 14-1.

Tools: Wireframing and Prototyping

THE VAST MAJORITY OF RESPONDENTS use at least wireframing or prototyping tools (93%), and about half use five or more. The most popular wireframing and prototyping tools are not software, but pen/ pencil and paper. These are used by respondents of all ages and experience: there is no indication that they will be replaced by software anytime soon. CSS/HTML is also frequently selected, especially for prototyping (44%), less for wireframing (22%).

Other popular tools include Invision (41%), Axure (26%), and Keynote (19%) for prototyping; and Sketch (40%), Illustrator (40%), Balsamiq (17%), and InDesign (17%) for wireframing. We find only two patterns of strong co-usage between the set of tools in the survey: users of Adobe’s Illustrator and InDesign, and between Sketch and Invision users. Sketch/Invision users tend to earn slightly above-average salaries, while Illustrator/InDesign users tend to earn slightly below average. In particular, the 6% of the sample that use Illustrator and InDesign but not either Sketch or Invision have an especially low median salary of $54K (median adjusted salary: $55K).

Note

Sketch/Invision users tend to earn slightly above-average salaries, while Illustrator/InDesign users tend to earn slightly below average.

dss2017 19
Figure 15-1.
dss2017 20
Figure 15-2.
dss2017 21
Figure 15-3.
dss2017 22
Figure 15-4.

Tools: Information Organization / Architecture

THE NEXT TOOL CATEGORY IS INFORMATION ORGANIZATION/ARCHITECTURE, including software for card sorting and mind mapping. 45% of respondents use at least one tool in this category, but most that did use just one. No single information organization/ architecture tool is used by more than 10% of the sample, in stark contrast with some of the other tool categories, such as wireframing or project management.

The most commonly used information organization/ architecture tools are OptimalSort, Google Drawings, Simple Card Sort, and XMind. Users of each of these four tools have above-average salaries. (For XMind, median salary was only $70K, but median adjusted salary was $86K. The shift is due to XMind being much more popular outside of the US than within the US.) More generally, respondents who use any information organization/ architecture tool earn more than those that don’t: the difference in median adjusted salary is $9K.

Note

Respondents who use any information organization/ architecture tool earn more than those that don’t:

dss2017 26
Figure 16-1.

Tools: User Research and Testing

The category of user research and testing tools is notable for its variety: 21 tools are used by at least 1% of the sample, far more than any other category. About two-thirds of the sample use one or more of these tools, although only 30% of the sample use more than two.

The top tools in this category are SurveyMonkey, Skype, GotoMeeting, User Testing, Google Hangouts, and Webex. Differences in salaries among users of various user research tools are not significant, although, as with information organization/ architecture tools, respondents who use at least one of these tools tend to have higher salaries than those that don’t, again by a margin of $9K.

dss2017 23
Figure 17-1.
dss2017 24
Figure 17-2.

Conclusion

THE FIELD OF DESIGN HAS SEEN PROFOUND CHANGES IN THE LAST DECADE—new mediums, new tools, new frameworks for thinking about design and how and where design principles should be applied. Keeping up with the evolving tool ecosystem can have an impact on one’s career development. We see that those adopting particular tools and techniques, e.g., Sketch and Invision, Agile, information management, and testing, all correlating with higher salaries. We also see that those who know more tools, work in larger organizations, interact with a wider variety of roles, and work on multiple platforms earn more—information that designers can use to help expand their career horizons.

As design principles and design thinking move beyond the world of designers, we see these findings as relevant well beyond the world of design. Software developers, data scientists, or anyone who does design work or works closely with designers can benefit from understanding what designers use and how they work.

When we quote statistics about salary, for example, that users of this tool make this much more than users of that tool, it’s important to remember that learning the “high salary” tool is not guaranteed to give you a raise. This survey data is observational, and we can’t assume cause and effect. On the other hand, knowing that particularly well-paid designers frequently use some tool might be a potential sign that this tool is especially efficient or powerful, and that alone could be enough justification for trying out a new tool.

This research is an ongoing project, and it depends on your participation. If you’ve found this report useful, please consider taking 5 to 10 minutes to complete the 2018 survey yourself for next year’s report: http://www.oreilly.com/design/2018-design-salary-survey. Thank you!

Appendix A: Adjusted Median Salary

GEOGRAPHY AND EXPERIENCE CLEARLY MAKE A DIFFERENCE IN SALARY, and this is fully expected. However, since geography and experience can correlate with other variables, unless we analyze all three variables together, it can be hard to tell whether variations in salary are due to these variables or to geography/experience.

For example, age correlates with experience and with salary, but if we consider groups of respondents with equal experience, then age no longer correlates with salary (at least, strongly or monotonically). This is what we mean when we say that age and salary don’t correlate when we “block” years of experience.

To give another example, this time with geography: the median salary of the 9% of respondents who say they code over 20 hours/week is $50K, while the rest of the sample (those who spend less time coding, if any at all) is $80K. However, the difference is attributable to the fact that most of the people who code over 20 hours/week happen to come from places that have lower salaries in general. For example, 36% of respondents from India, Russia, and Ukraine say that they code over 20 hours/week, while only 1% of California respondents do. This probably shouldn’t be taken to mean that CA design professionals don’t code: correlations like this will appear frequently on such surveys; namely, when there is little control over the sampling. This correlation is likely just noise that we should try to filter out.

The solution we use in this report is to provide, when appropriate, an additional metric, the adjusted mean salary. The first step in computing this metric: is to create a simple model to predict salary using country/state and experience. After trying a few economic metrics to quantify geography, we found that per capita GDP gave the best results.2] Using this survey’s data, no complicated modeling technique or transformation made a big improvement over a simple linear model, so we stuck with the latter. The model is:

predicted_salary = 1.95 x years_of_experience + 1.26 x per_capita_GDP – 1.29

where monetary values are in thousands of USD, and years of experience is capped at 20 (for someone with more than 20 years of experience, the value inserted into the model is 20). For example, the predicted salary of someone with 7 years of experience from Australia (where the per capita GDP is $51K) is $76.5K. This model explains about half of salary variance in this sample.

We use this model to create the aforementioned “adjusted median salary” statistic. This works by recalculating salaries as if the respondents who received them were from a single, fixed place and had the same amount of experience. The actual fixed values are somewhat arbitrary, and we pick values close to the sample averages: seven years of experience and $51K for the per capita GDP, which is roughly the per capita GDP of Australia, Denmark, Singapore, Ohio, North Carolina, and Wisconsin. To adjust someone’s salary, we simply subtract an amount that the model attributes to their experience and geography, and then add a fixed amount for seven years of experience and a per capita GDP of $51K.

Perhaps a simpler way of understanding the calculation is to consider the residual, the difference between the observed (reported) salary and the predicted salary. If someone earns much more than we would expect given their experience and location, their salary residual is high. We calculate the residual and then add it to $76.7K, the predicted salary for someone with seven years of experience living in a place with a per capita GDP of $51K.

For example, suppose someone from New York with five years of experience earns $120K. According to the model, this person is expected to earn $100.3K, so they “outperform” the expectation by $19.7K (this is the residual). If we add this $19.7K to the fixed $76.7K, we arrive at the adjusted salary, $96.4K. It is worth noting that a single adjusted value in isolation doesn’t have much relevance; the real purpose of presenting these adjusted values is comparison. In a sense, we are really just including residuals, and the conversion from residual to adjusted salary (i.e., the operation of adding $76.7K) is performed to convert the number to something less abstract and so that we don’t have to introduce technical language (“residual”) into the text.

It is likely that with more data, a more complicated model (i.e., one that still just takes in experience and GDP, but is not a simple linear model) would provide better results. For example, it seems likely that not every incremental year of experience is the same (e.g., 3 to 4 years versus 13 to 14 years) or that experience has the same relation with salary in every place (e.g., one year of experience adds as much in Switzerland as it does in Poland).

However, the simple linear model above performed just as well as a few others we tried that attempted to explain a more complicated relationship, and furthermore, the simplicity has the major advantage that it is easier for you, the reader, to plug in your own numbers. Unlike the models in other O’Reilly salary surveys, including last year’s Design report, this model only takes two variables, so insofar as there are other relevant variables that affect salary, this model will miss them. More fundamentally, the variance of salaries for any given experience and per capita GDP is quite high: the predicted salary that the model outputs is an average, and any particular salary may fall a ways from it. For that reason, the real value of this model is not to predict someone’s salary, but to allow us to compare groups of salaries in a way that the comparison is minimally impacted by significant differences in experience and geography.

2Both country and state numbers were taken from Wikipedia. Country figures are from the IMF column. http://bit.ly/2iCUvcD.[

Article image: Mosaic. (source: Jan Hrdina on Flickr).