Want to hire the best programmers? Offer growth

Each week, Triplebyte matches hundreds of programmers with 450+ tech companies. We're somewhat like a dating site for software engineers and the companies that want to hire them. So, much like a dating site would, we ask every engineer on our platform what they would be most excited about in their next job in order to best match them with companies. Among other things, we have candidates select from workplace attributes that we refer to as motivators. These include options like “opportunities for professional growth”, “mentorship”, and “inclusive workplace”.

Motivator data helps us optimize our matches for a good candidate/company fit, but it also gives us useful insight into the values and desires of software engineers. We recently analyzed this data to identify patterns and to better understand how factors like experience level, gender identity, and skill level correlate with these preferences.

It's not hard to imagine how this data might be useful. Many companies, for example, tell us they're interested in increasing the representation of women in their engineering departments. This can be a bit of a guessing game on their part, especially when (almost by definition) the people guessing are often a currently-male-dominated engineering department. Thanks to our very large data set, we don't have to guess: we can say what women engineers in particular are looking for with high statistical confidence. Similarly, some companies are trying to attract competent engineers and some are trying to attract the very best, and we can ask whether those two groups are motivated differently. As we'll see later, it turns out that they are. Outcompeting tech giants is hard , and knowing more about what candidates are looking for can give you a competitive edge.

Engineers Want To Learn, Even If They’re Very Senior

Let’s start with a high-level look at the data. We took a look at every candidate that has passed Triplebyte’s screening process, scrubbed out a few for whom data was incomplete (typically candidates from earlier in Triplebyte’s history, prior to us tracking all of the relevant data), and counted up the selections for every remaining candidate on our platform. Note that because candidates typically select three motivator choices, the percentages in this and the following graphs do not add up to 100%.

“Opportunities for professional growth” is the clear winner, appearing 13 percentage points above even “salary”. As we'll see, this turns out to be a universal interest: in almost every cohort we looked at, professional growth was the most selected motivator. Other motivators move up or down in the rankings between men and women, between junior and senior candidates, and so on, but professional growth is always a priority.

This is probably rational. Software engineering is a fast-moving field, and to stay still is to be left behind.We've written before about how things like an engineer's choice of language can affect their job prospects, and popular choices change quickly over time. In 2002, a very good COBOL or Fortran engineer was probably in a good place as far as their job search was concerned, but today they're unlikely to get much attention from companies that have moved on to more modern languages. Had that engineer focused on growth to keep up with changes in the industry, they might have learned Python or JavaScript and been able to keep up with the market.

It's not uncommon for us to see skilled programmers left on the sidelines for the sin of sticking with PHP or Perl for a little too long. They're stuck in a catch-22, unable to move on to more modern technologies because the companies using them don't want someone rooted in old ways of doing things. For all that software engineering is a lucrative and comfortable career in some respects, a smart and career-aware engineer knows that their long-term employability is more precarious than it might look at any given time.

Still, “professional growth” is a broad term, and it would be dangerous to assume too much about candidates' motivation for selecting it. Perhaps they interpreted “growth” to mean something like “promotion into management”, rather than growth of their technical skills. So we listened to a number of phone calls with candidates and reached out to a few candidates directly to dig deeper about what “professional growth” meant to them. Learning was the common theme in their responses:

• “A job that helps me gain professional growth would include opportunities for me to learn new things and learn how to improve my skills in various technical areas.”
• “[It's important to] feel like you're constantly learning.”
• “I want to be the worst person on a team so I can learn.”
• “I think professional growth means room for improvement in terms of both technical skills and interpersonal skills. I chose this because I want to continue to improve my skills during my work at my workplace instead of just utilizing skills that I already have and sticking to what I am already comfortable with. So a job that helps me gain professional growth would include opportunities for me to learn new things and learn how to improve my skills in various technical areas.”

In short, software engineers of all stripes want, more than anything else, to develop their abilities as engineers . They want it more than pay, more than work/life balance, and more than autonomy. This is probably, at least in part, because they know that every other priority they have will eventually depend on keeping up with the industry.

This remains true even for very experienced candidates. Triplebyte's background-blind process often attracts non-traditional junior candidates who might have trouble getting their foot in the door in a traditional recruitment process. As a result, one might wonder whether this interest in professional growth turns out to be an artifact of having non-traditional junior engineers. This turns out not to be the case.

In fact, preferences do vary between engineers of different seniorities, but the focus on growth does not. Here's the same data shown above, this time broken down by different tiers of seniority: junior (defined here as<1 year of experience), mid-level (1-5 years), and senior (>5 years). All three levels of seniority overwhelmingly prioritize professional growth relative to other options.

Senior engineers actually are marginally less likely to select “opportunities for professional growth”, but the difference is small, and growth remains their most-selected option.

Other motivators, however, do show significant shifts. “Mentorship”, unsurprisingly, plummets as engineers gain more experience. “Autonomy” and flexible work arrangements rise. This is probably in part because more experienced engineers tend to be more skilled, though not by nearly as much as is generally believed (one of many reasons our interviews are background-blind). As we’ll see later, these are motivators that are more common among more capable engineers.

Women Engineers Want Growth, Too (But Also Comfort and Inclusivity)

Going into this analysis, we felt fairly sure that “inclusive workplace” would turn out to be a strong differentiator between male and female engineers. We weren't sure whether or not other motivators would show significant differences, but one of the valuable things about data analysis is that you sometimes find things you're not looking for.

Our first step was to simply break the data down by gender.

(Both here and throughout this article, 'gender' and related terms refer to a candidate's self-reported choice of preferred pronouns. Although a few candidates chose gender-neutral pronouns or did not select a pronoun at all, the number who did so is small enough that we can't draw any meaningful statistical conclusions about them. For the purposes of this article, we'll focus on candidates who chose binary pronouns.)

“Opportunities for professional growth” remains at the top. In fact, it’s even more of a priority for women engineers than it is for engineers in general.

As expected, “inclusive workplace” shows the largest relative difference between men and women engineers. Female-identified candidates chose it three times as often, a far larger relative difference than any of the other common motivators. Although some of the reasoning is probably obvious, we reached out to a few women to ask why they prioritized inclusivity:

• As a female software engineer with an international worker status, I recognize that I may be minorities in multiple different dimensions. So I definitely wanted to be in a workplace that would at least put some emphasis on creating inclusive workspace so that I can feel comfortable in the environment.

One response was especially striking:

• I picked “inclusive workplace” and I didn't want to put “mentorship” down, even though it is important to me. There's a sense that women are less independent, and I feel like saying you want “mentorship” gets you told that “we don't do hand-holding here”.

Some of the other apparent differences are more surprising. Women in our data set were much more likely to choose “mentorship” and much less likely to choose “flexible work arrangements” than men were, for example. But here, we have to be careful with a naive analysis, because our samples of men and of women are not directly comparable. As gender-imbalanced as tech is today, it was even more so in the past. As a result, women in our data set tended to have fewer years of experience than the men in our dataset did, and thus motivators more common to junior engineers would (spuriously) appear to be more common among women. As an example, only about a third of our sample of women had more than five years of experience, while around three-fifths of our sample of men did. Since “flexible work arrangements” is chosen much more often by senior engineers than by junior engineers, this could easily introduce an apparent difference between men and women engineers that turns out to be a masked difference in seniority rather than a difference in their actual priorities.

To correct for this, we scaled our sample of women engineers to a seniority breakdown that matched our male sample in order to make apples-to-apples comparisons. In other words, we split our data set by both gender and seniority, then weighted the seniority tiers in our sample of women to match the proportions of the corresponding tier among men.

The resulting data looks like this:

Some of the differences disappear. “Mentorship”, for example, no longer shows any difference between male- and female-identified candidates (19% of men and an adjusted 20% of women chose “mentorship” as a motivator).

'Opportunities for professional growth' remains the dominant motivator and, in fact, becomes even more dominant among women. After adjusting for seniority, a whopping 62% of women chose professional growth as a motivator.

“Work-life balance” is a more interesting case. Among men, “work-life balance” doesn't vary much by seniority, so one might reasonably expect that adjusting our sample of women for seniority wouldn't affect its prevalence very much. However, senior women engineers pick it so often (almost 50% of the time) that it rises into second place, selected almost half again as often by our adjusted sample of women (43%) as by men (31%).

Most of these differences are highly statistically significant, but this is a case where we might be more interested in effect size than in statistical power. Here’s the same seniority-adjusted data, this time sorted by the relative differences between men’s and women’s selections. For example, “open communication” is selected 75% more often by men. The highlighted entries are statistically significant.

Unsurprisingly, women selected “inclusive workplace” 171% more often than men did. There’s no contrarian conclusion here: inclusivity matters to women engineers. That being said, even this very large relative difference doesn’t override the fact that women engineers, like engineers in general, still prioritize professional growth over all else.

More interesting is the fact that many other motivators without obvious links to gender or to gendered experience show differences. In addition to inclusivity, women were more likely to select “high quality codebase”, “work-life balance”, and (much more weakly) “culture of transparency”. These motivators, for the most part, fit a “comfort cluster” of environmental motivators weakly correlated with one another. In other words, after growth and inclusivity, women in our data set tended to disproportionately value comfort in both their workplace and their codebase.

Men, on the other hand, were statistically more likely to select “autonomy”, “salary”, “fast-paced environment”, “flat organization”, “product-driven”, and “open communication”. Some of these fit into traditional gender norms in ways that probably don't need much further explanation. Some candidates may believe in these roles themselves; others may feel social pressure to conform to them. In either case they apparently prioritize them in their job search.

Obvious norms don't account for all the differences, though. “Flat organization” in particular stands out - not a single woman in our data set selected it! ( Flat organization is a rare selection in general, but the difference is highly statistically significant to the point that it is unlikely to be noise.) It's harder to get a signal on why a candidate might avoid selecting one of these motivators, but one speculative possibility is that women engineers may see ostensibly flat organizations as fertile ground for implicit biases, and thus find them less comfortable than their male counterparts.

Great Engineers Are Different From The Best Engineers

One advantage to Triplebyte's data set is that we can look at our data not just for demographic traits, but also for the skill displayed on our technical screen. It's reasonable to imagine that particularly good engineers might differ from the population as a whole, so we sliced up our motivator data further by how candidates scored on our screen.

For the purposes of this article, we’re arbitrarily defining “great” candidates as those scoring between 95th and 98th percentile on our technical interview, and “the best” candidates as those scoring at 98th percentile or above. We pass a little over 30% of candidates who take our interview (the exact number fluctuates depending on incoming candidate quality), so these two groups together make up a little under a sixth of our passing candidates that actually end up on our platform visible to companies that hire through us. Only passing candidates select motivators, so all of the data in previous sections of this article has been on this passing 30+% percent of candidates.

As it turns out, highly skilled 'great' engineers do indeed differ from the rest of our candidate pool:

Great engineers answer “mentorship” much less, which makes sense - in most cases, they're already quite a bit better than the person who would be mentoring them. “Opportunities for professional growth” also drops, probably because at this skill level job prospects are much rosier. Much more surprising is the massive drop-off in “open communication”: again, it's hard to tell why a candidate would avoid a selection, but one possibility that these are the sorts of engineers who like to be handed a problem and then left alone.

On the other hand, great engineers answer “salary”, “impressive team members”, “autonomy”, “fast-paced environment” and “equity” much more. “Autonomy” suggests that these engineers know that they're skilled and can be trusted to work with minimal supervision, and “salary”, “fast-paced environment”, and “impressive team members” reflect career ambition appropriate to that skill level.

In short, great engineers want to be paid well and allowed to exercise their skills in challenging environments .

This isn’t very surprising. We expected that these differences would persist, or be magnified, when we looked at the very best engineers; after all, it seems logical that more of whatever traits make someone a good engineer would make someone an even better one.

Our expectations turned out to be quite wrong!

The best engineers have completely different priorities from great engineers. In many cases, the trends from the previous graph actually reverse when looking at the best engineers. For example, great engineers selected “work/life balance” less often than the population as a whole, but the best engineers rank it as their top priority by a gigantic margin, making it the most selected option in any cohort at 65%. “Impressive team members”, on the other hand, was selected more among great engineers but plummets to near the bottom among the best.

The best engineers notably de -emphasize ambition. They’re so good that their employment is almost certainly secure at least in the short term, and so they can work where they want - this group includes an electrical engineer, a professor at a small college in New England, a remote developer for an education platform, and a musician. With concerns about their professional future out of the way, they can seek meaning (“product-driven”) and comfort (“flexible work arrangements” and “work/life balance”).

In other words, the priorities of the very best engineers are very different from those of merely great engineers . The very best engineers want comfortable and meaningful work and the ability to choose their work environment and timing.

This has very interesting implications for hiring. Most companies do not need the very best engineers, and trying to appeal to them might discourage great - but slightly less superlative - engineers who would be able to meet their needs just fine. In other words, companies should consider whether they need good engineers or the very best engineers, and choose how they market to candidates accordingly.

In Short

What can we take away from all of this?

First, engineers, regardless of demographic, want the chance to learn and to advance their technical skills . This is especially great news for new startups, since playing up the opportunity for engineers to learn new skills and technologies is a great way to set themselves apart. Relative to typical Bay Area software engineer salaries in the $150,000/year range, devoting (say) $1,000 or even $5,000 to classes or other forms of professional growth is not particularly expensive, but it can have a disproportionate effect on recruiting.

Companies willing to put this investment into their engineers also have a competitive advantage. They can hire strong senior engineers who have fallen behind the latest technologies and thus source candidates who are much less competitive than other candidates of a similar skill level. Offering even moderate investment in training opens up a candidate pool that is currently underrated by the job market.

Second, women engineers prioritize growth to an even greater extent than men, and they place particular secondary emphasis on inclusivity and comfort with their work and environment . Putting time aside to work on technical debt and taking the time to build proper infrastructure are often discussed in terms of their business benefits, but it's rare to think of them as a recruiting tool to improve gender diversity in an engineering department.

Explicit inclusivity is still important, however, and so is the growth that engineers as a whole want. In some cases, a desire for inclusivity is tied into a desire for growth: for example, a woman engineer who wants to learn through mentorship may not feel comfortable asking for it if she feels her environment is not supportive. It's important to note that an inclusive environment will help with more general recruiting, too: the top male engineers are also slightly more likely to mention inclusivity as an important value.

Finally, it's critical to decide whether your company needs good engineers or the very best engineers . It turns out to be quite difficult to target both, because the desires of a 95th percentile engineer and a 99th percentile engineer are different and often mutually exclusive. Most companies do not need the very very best engineers available. If your company doesn't - and it probably doesn't - then you can gain some competitive advantage by targeting less than the best engineers. They aren't in as much demand, they have a very different set of priorities, and they are likely just as capable for your purposes as the best engineers would be.

If your company wants great engineers, emphasize pay, autonomy, and challenge. If your company wants the very best engineers, let them work from home, keep your expectations around hours worked reasonable, and go out of your way to emphasize meaningful product.