The Science of The Job Search, Part II: Racism, Outgroup Bias & KFC

I have a dream that my four little children will one day live in a nation where they will not be judged by the color of their skin but by the content of their character. — Martin Luther King, Jr.

Tomorrow is the end of Black History Month. In a few weeks, it’ll be the 50th anniversary of Martin Luther King Jr.’s death. How much does race still matter in America?


At least in hiring, race (still) matters — a lot. After analyzing thousands of job applications, outcomes and applicants, we discovered 3 key things:

  1. Non-white job applicants got 2.3x fewer interviews than their white counterparts;
  2. For non-white job applicants, if a resume mistake reinforced a racist stereotype, it basically disqualified them, e.g. African-Americans are lazy, Asian-Americans can’t speak English;
  3. On the other hand, very few people were consciously racist. When non-white applicants followed specific tips (see below) that forced hiring managers to consider them objectively, they were given a 54% fairer chance.

Let’s break it down, shall we?

Non-White Applicants Are Less Hireable

It’s hard out there for a black man. (And black women and Hispanic men and…)

Why? Two reasons (see below):

  1. When a white person applied to a job, they had a 13.0% chance of hearing back, compared to 6.7% for Asian-Americans and just 2.3% for African-Americans!
  2. When white applicants heard back, they heard back in 12.4 days, compared to 18.8 days for Hispanics and 41.2 days for African-Americans (nearly 6 weeks)!
Ethnicity% Interview Rate# Days for Replyp-value

A common explanation for this sort of effect is that non-white job applicants have lower educational attainment, etc. We’ve omitted that analysis here for brevity but, in short, the effects here aren’t explainable by education, work history, skills or geography, e.g. African-American job applicants had nearly 37% more education than their white counterparts. (President Obama was right…)

If you’re a minority (and you’re ever planning on looking for a job), this is pretty terrible news. What can you do about it?

Disqualifier: Don’t Reinforce Racist Stereotypes!

When you look at the data, it becomes pretty clear that certain resume mistakes are basically disqualifiers for certain races. Those disqualifiers fit a general pattern: racist stereotypes.

Put another way, if you in any way reinforce a pre-existing racial stereotype, you’ll be punished for it. Check out the following graph, for instance:


If you’re from an immigrant family (or your name looks like it might be) and you make a resume faux pas, your chances drop precipitously.

#5: Asian-Americans: Dorks Who Can’t Speak English [100% PENALTY]

Asian-Americans are awkward math dorks who can’t speak English — that’s the stereotype, right? Witness Engrish, Apu in the Simpsons, and this scene from the Big Short (one of my favorite movies):

If you’re Asian-American, a resume mistake that in any way reinforces the Engrish/bad communication skills stereotype can be a fatal mistake. Don’t believe me? Check out this tweet a few weeks ago from Seattle:

Look at her dad’s original email above. It’s not great English, sure, but I’ve seen hundreds of applications like that from non-Asian applicants too. Do you think Bruce would reply like that to Chad?

Job Search Tip: Especially if you’re Asian- or Hispanic-American, make sure that you’re not making a resume faux pas (100% penalty).

(If you’re worried about making an unintentional faux pas, ApplicationAssistant can automatically optimize dozens of variables for you, including writing personalized cover letters for each job application.)

#4: African-Americans: Lazy Welfare Queens [PENALTY]

This doesn’t apply just to Asian- and Hispanic-Americans.

Let’s talk about two people: Tyrone Robinson, who got zero replies after ~4 months of job-hunting in SF, despite applying to 100+ jobs; and, DeAndre Jackson, who got multiple offers after just 1 month of job hunting in Los Angeles. (Names and locations changed to protect the innocent.)

Consider Tyrone’s work history:


Marketing Assistant, Gap, 2015 — 2017
Sales Associate, Gap, 2014 — 2015
Sales Associate, Best Buy, 2013

Looks good, right? 5 years of work experience. Good career progression. So, why didn’t he get the job? Consider DeAndre, who had less experience but did just fine:


Marketing Assistant, American Apparel, 2017 — Present
Marketing Assistant, Banana Republic, 2015 — 2017
Office Manager, Law Office, 2014 — 2015

What’s the stereotype for African-Americans? They eat fried chicken, they’re lazy and they’re welfare queens. If you’re African-American, a resume mistake — no matter how small or innocent — that reinforces the lazy welfare queen stereotype will doom your job search.

If you look at Tyrone’s resume, you can see that he had two unexplained gaps in work experience. For any other person, it might not be great, but it would’ve been just fine. For Tyrone, it was a fatal (and 100% innocent) mistake. (He’d had to take care of his ailing mother.)

Resume Tip: If you’re African-American, make sure you explain (or remove) any gap in work experience.

Explanation: Outgroup Bias and KFC

Here’s the rub: none of these resumes contained photos. Hiring managers could only infer an applicant’s ethnicity based on their name.

That split-second inference has tremendous (subconscious) power: a first impression (a home team jersey or how a name sounds) fires off millions of neurons in your brain, giving you dozens of working assumptions in a few hundred milliseconds — it’s what we call a “gut feeling” in everyday life.

And it’s important! We’re bombarded by information everywhere we go and if we didn’t have our gut feelings, we’d be sloppy, slippery puddles of anxiety on the floor. But, sometimes it goes awry.

Whenever someone seems foreign or unfamiliar to you, those gut reactions prepare you to be skeptical and wary — in psychology, it’s known as outgroup bias. This made a ton of sense when we lived in warring clans in the Irish highlands or African savannah, but it’s less helpful now that most (corporate) folks’ biggest physical threat is a paper-cut.

In the couple seconds the average manager takes to review an average resume, outgroup bias means Tyrone Robinson gets bucketed as a lazy, KFC-eating flake and Minh Huynh gets bucketed as an Engrish-speaking dork in a few hundred milliseconds. They never even had a (real) chance.

Sucks, right? But, it turns out there’s hope—

Equalizer: Force an Objective Mindset

In general, when hiring managers are reviewing your resume, you’re operating at their subjective, subconscious “gut feeling” level. And if you’ve got a name like Tyrone Robinson, Maria Torres or Minh Huynh, that means you’re in trouble.

But, remember: very few people are consciously racist. What if instead of fighting the subconscious, you forced them to consider you consciously — objectively?


Non-white job applicants saw up to +199% higher interview rates when they forced hiring managers to consider them objectively. This roughly translated to closing the racial discrimination gap in hiring by 54% (a 1.6x race penalty vs. 2.3x originally).

So, how do you force hiring managers to consider you objectively? In addition to not breaking resume etiquette, you need to follow at least two of the resume tips below:

#3: Use Concrete Numbers [+23% BOOST]

As we mentioned last time, every 3 sentences, use at least 1 number to demonstrate your (concrete) impact. Between the two people below, who would you hire?

Helped increase sales by 31% by working with Operations Manager to reduce time to 1st customer reply.

Collaborated with Operations Manager improve customer reply times.

The first one is better than the second for everyone. But, if Chad and Tyrone both say the second, Chad is always going to win.

Resume Tip: Especially for people of color, quantifying the impact that you made with numbers helps remove subjective bias (+23% boost).

#2: Add Industry Buzzwords [+34% BOOST]

Also as last timeadd 15-20 specific skills, industry buzzwords, acronyms, etc. to your resume. Although it’s helpful for everyone, it’s especially helpful for minorities to anchor their expertise in objectively-known and -respected foundations.

Resume Tip: Include 15-20 specific skills, industry buzzwords and expertise in your job achievements (+34% boost).

#1: Don’t Be a “Team Player” [+63% BOOST]

Finally, don’t be a “team player”. Don’t mention these sorts of collaboration-oriented words more than once or twice in your resume:

team player
results-driven collaborator
supporting member

Why? These sorts of words discount your achievements to hiring managers. Saying it once conveys that you work well in teams. Saying it five times screams, “I don’t know how to get anything done individually.” If you’re a person of color, hiring managers are already (subconsciously) discounting your achievements — don’t help them.

(Quite frankly, this is a lot like the problem that many women face in the workplace. More on that in a post next month. In the meantime, check out #banbossy.)

Resume Tip: For people of color, being explicit about your specific contributions is crucial to remove subjective bias (+63% boost).

(Given the different variables to balance, this one’s a bit tricky. Our ResumeOptimizer can automatically suggest corrections for teamwork cliches and other common mistakes.)


To recap:

  • Race (still) matters, a lot — non-white job applicants were 2.3x less hireable than white job applicants.
  • If you (in any way) reinforce a pre-existing racist stereotype, outgroup bias will cause hiring managers to instantly disqualify you, e.g. African-Americans are lazy welfare queens, Asian-Americans are dorks who can’t speak English.
  • On the other hand, if you forced hiring managers to consider you objectively, they gave you a 70% fairer chance. For example: demonstrating personal achievements, concrete skills, quantified impact.

As with everything involving race, you’re playing on a knife’s edge here: do too little and you’re nothing, do too much and you’re labeled try-hard (or worse, blackballed). You have to do the right thing in exactly the right amount. Balancing the hundreds of variables that go into this is hard, for humans.

But, it’s easy for robots: our AI-driven systems can simultaneously optimize 1,000+ variables. We’re just getting started with these sorts of discrimination-related hiring issues, but our existing tools can already help a lot:

  • ApplicationAssistant automatically identifies jobs you’re objectively qualified for, so you can emphasize your objective qualifications and mitigate subconscious racism/outgroup bias; and,
  • ResumeOptimizer instantly scans your resume for dozens of potential issues putting you at a disadvantage, including the three objective-mindset tips above.

In addition, our TalentAdvocates have helped people of every ethnicity and background navigate the especially complex, confusing, frustrating job search landscape that minorities face. Let us help.


We used the core dataset, analysis and visualization from The Science of the Job Search, Part I.

To calculate which resume tips were disqualifiers, we looked only at non-white subgroups and filtered for tips where their absence resulted in a <5% interview rate. For equalizers, we filtered for tips where their presence resulted in an interview rate within 2 standard errors of the overall mean. The above graphs are plotted with bokeh on Python.

Why Are We Doing This?

With ApplicationAssistant right now, we can boost the average job-seeker’s hireability by 5.8x. But, what makes ApplicationAssistant work has been an internal company secret until now. We’re fundamentally a mission-driven company and we believe we can help more people by sharing our learnings. So, that’s exactly what we’re going to do.

Creative Commons

We’re not only sharing this but also sharing all of it under the Creative Commons Attribution-NonCommercial-ShareAlike license. In other words, as long as you follow a few license terms, this means you can:

  • Share: Copy, redistribute the material in any medium or format.
  • Adapt: Remix, transform, and build upon the material.
Did you like this post? Share it with your friends!

The Science of The Job Search, Part I: 13 Data-Backed Ways To Win

It’s the New Year! And what does that mean? Fireworks, champagne and New Year’s Resolutions. It turns out fully 63% of people’s New Year’s resolutions are about jobs: negotiating that promotion, quitting that job you’ve always hated, or getting that new job you’ve always wanted.

But, as we all know, getting a job is hard. Clearly, there are jobs out there. And clearly, some people are getting those jobs. In fact, although people have a ~2% interview rate for online job applications on average, some TalentWorks subscribers have a 40%+ interview rate!


We call these folks our “A-List Talent.” What’s so special about them? What’s their secret? And is there anything you can learn from them?

We analyzed 4,000+ job applications and job applicants from the past few months and, using some fancy math and a bit of elbow grease, identified 13 key factors out of 100+ possible factors that drove up our A-List Talent’s interview rates. So, without further ado, let’s get to it: What can you learn from our A-List Talent’s super-high hireability to (finally) get that job you deserve in 2018?

Factors you can’t control

#3: Go back to school. [+22% BOOST]


Having a 2nd degree boosts your chances of getting an interview by +21.9%.

Why is this something you can’t (easily) control? I don’t know about you, but not everyone can put their life on hold, spend four years and tens of thousands of dollars of school in the hopes of getting a better job… four years from now.

Job Search Tip: Except in rare scenarios, you should not go back to school for a 2nd degree just to improve your job prospects. When you factor in opportunity cost, you usually come out behind — it’s just not worth it.

#2: Be older. (Or younger.) [+25% BOOST]


Age matters. A lot, sadly. Your chances of getting a job at age 20 are pretty bad. At 30, they’re OK. At 40, they’re getting bad again. It might be illegal, but age discrimination is very real.

The best age to get a job is between 28 and 35. During this time, you get a +25.1% hireability boost over everyone else. Up to age 28, your hireability is increasing by +9% every year. After age 35, your hireability drops by -8% every year.

But, here’s the rub: this is inferred age. Hiring managers (subconsciously) guess your age based on your graduation date, how much experience you have, etc. If you don’t show your graduation date, they can’t tell how old you are. If you only have your most recent 2-3 jobs listed, they can’t tell that you started working in the 1980s.

Resume Tip: Don’t list your graduation date if you’re older than 35. If hiring managers can’t guess your age, they can’t discriminate against you based on it.

#1: Be a woman. [+48% BOOST]


Resumes with obviously female names had a +48.3% higher chance of getting an interview. For example:


This effect was initially very surprising to us, but when you think it about it, it’s really not. Dozens of studies show that women often don’t get what they deserve (basically) because they don’t ask for it. This shows that, when women do ask for what they deserve, they’re often recognized for it.

In the past several months, women across the country have become more vocal about their rights, from standing up to sexual harassment to supporting each other in the workplace. Between the clear (data-proven) benefits of hiring women, that women are outperforming men in school, and the fact that most recruiters are women (who want to support other women), it makes 100% sense why women might be getting a boost when they apply for jobs.

Job Search Tip: To all the women out there who might question themselves, undervalue their contributions, or wonder if they truly deserve it, push through the discomfort and demand what you deserve (the job, the raise, the promotion). You miss 100% of the shots you don’t take but, at least with job applications, when you do take that shot, you’ll get a +48.3% boost over the competition.

Factors you can control

#10: Play buzzword bingo. [+29% BOOST]


Buzzwords, keywords, acronyms, industry jargon — call them what you want, but they serve a purpose. Beyond the usual reasons, they help you get past automated screening tools used by many big companies. But if you go overboard, the actual hiring manager might think you’re a tool (even if the robots don’t notice).

Resume Tip: Name-drop a buzzword every 3-6 sentences. Folks who dropped an occasional buzzword saw a +29.3% boost over others.

#9: Demonstrate results with numbers. [+40% BOOST]


I’ve personally hired 100+ people over my career and, during that time, I’ve probably personally reviewed 10,000+ resumes. Even if a resume passes your sniff test, the hardest thing is separating what’s real vs. what’s pink, fluffy, sugar-y cotton-candy-coated horsecrap. Quantifying how you made an impact with numbers goes a long way towards helping hiring managers tell them apart fast.

Resume Tip: Every 3 sentences, use at least 1 number to demonstrate your (concrete) impact. Folks who did that saw gain a +40.2% boost over their competition.

#8: Apply on Mondays. Don’t apply on Fridays. [+46% BOOST]


Enough said.

(Why are there so few applications on Fridays or Saturdays? ApplicationAssistant automatically optimizes dozens of variables for you, including when it applies for a job on your behalf.)

Job Search Tip: Apply on Mondays (+46.0% hireability boost). Don’t apply on Fridays or Saturdays.

#7: Don’t be a “Team Player.” [+51% BOOST]

This one’s a little counter-intuitive, so hold on.


Everyone talks about how important teamwork is. And how our whole economy is becoming about sharing. And collaboration. Lots of it. It’s very cute.

When it comes to actually hiring someone though, the most collaborative candidates get penalized by -50.8% by hiring managers. If that doesn’t make sense, consider these

  1. Owned, analyzed and delivered on-time financial reports for business sub-unit A to management team on monthly basis.
  2. Collaborated with full analyst team to create monthly financial reports for management team.
  3. Assisted management team by creating monthly financial reports as a supporting member of the analysis team.

Who would you hire? (Or call for an interview?) In the 2nd and 3rd case, I have no idea what work you did (vs. free-loading off your team). Finally, many collaborative words also have passive, subordinate, weasel-word undertones.

Resume Tip: Don’t mention more than once or twice that you’re a “team player,” “results-driven collaborator,” “supporting member”, etc. This is associated with a +50.8% hireability boost over the competition.

#6: Take charge with leadership words. [+51% BOOST]


The converse of avoiding weasel words is also true. Adding strong, active, leadership-oriented words also helps you. Some of the words we detected as strong, active words:


What I want you to get from that: You don’t have to be the CEO of your company to be a leader, and leadership doesn’t always mean managing people or huge budgets. Even if you’re just an intern somewhere, you can still demonstrate leadership traits by proactively communicating with co-workers. And your future bosses want to know that!

Resume Tip: Incorporate 1-2 leadership-oriented words every 5 sentences. Job applicants who used strong, active, leadership-words saw a +50.9% boost over the competition.

#5: Don’t use personal pronouns. [+55% BOOST]


People who used even one personal pronoun in their employment section (not the objective or professional summary section) had a -54.7% lower chance of getting an interview callback.

Resume Tip: Don’t use personal pronouns in your employment section. Ever.

#4: Include a Key Skills section. [+59% BOOST]


You can’t name-drop enough skills, buzzwords and acronyms to get to the optimal number of skills without one.

Resume Tip: For most [*] people, you should add 15-20 skills, buzzwords, acronyms, etc. to your resume. This is associated with a +58.8% boost in hireability on average.

[*] There’s actually a really interesting effect going on here. There’s a clear, second sub-population of special folks for whom 30-40 skills, buzzwords, acronyms, etc. is the right number. More on that later.

#3: Apply in the first 4 days. [+65% BOOST]

We’ve already talked about being first-in-line for a job.

Resume Tip: Applying early gets you a +64.7% boost over your competition on average. (Although it can make up to an 8x difference for a single job application, most people aren’t applying at the worst possible time.)

#2: Apply between 6am and 10am. [+89% BOOST]

We’ve already talked about this too.

Resume Tip: Applying between 6am and 10am gives you an +89.1% boost over your competition. (As above, your competition isn’t applying at the worst possible time so you don’t get the full 5x boost every time.)

And, finally, the #1 most important factor you can control?

#1: Start your sentences with (distinct) action verbs. [+140% BOOST]


If you did anything worthy at a company, you’ll have done something. If you start the sentence describing what you did with an action verb, you’re off to a strong start. And if you describe the different things that you did at that company with different action verbs, you’ll have finished strong.

Say what? In short, say this:

Developed a world-positive, high-impact student loan product that didn’t screw over people after 100+ customer interviews.

Not this:

After 100+ customer interviews, the world-positive, high-impact student loan product was developed by me.

Resume Tip: Describe your job achievements with different action verbs. This one resume tip is is associated with +139.6% boost in getting more interviews.

(Why so few people in the baseline? Our ResumeOptimizer will automatically scan your resume and suggest places where you should use action verbs.)

P.S. A (small) corollary —

Getting a job you deserve is hard, yes. But, it’s not as hard as you think. And I can prove that.

Many folks think to get a better job they have to fundamentally change as a person, gain new skills, learn new habits, network for weeks, etc. And, sure, all of that helps.

But, look again at the #1 most effective tip: it’s about changing the words on your resume for a +139.6% boost. (And not even all of the words — it’s literally about changing the first word of each job achievement.)

On the other hand, look at what a second degree buys you: a +21.9% boost. It’ll cost you tens of thousands of dollars and years of effort, but you’ll get 6.4x more impact for something that’ll take you a few minutes.

Why am I spending precious sentences trying to prove this to you for a blog post that’s already pretty long? Because it’s January 6th. And you probably haven’t made good on your New Year’s Resolutions yet. We know two things:

  1. Every week you procrastinate your New Year’s resolution means you have a 24% lower chance of succeeding at it.
  2. The #1 most-important step to completing a task is to start it. (No, seriously. These folks have studied it.)

So, don’t procrastinate. Don’t put it off. It’s not some crazy, big life improvement project to (finally) get that job you deserve. Instead, it’s about making sure you apply at the right time, changing a few words, or adding a few numbers. Go get that job you deserve today! You can do it!

(We can help.)


So, to summarize: Go back to school. Be a woman. Be older. (Or younger.) Sorry, bad joke. Play buzzword bingo. Demonstrate results with numbers. Apply on Mondays. (Don’t apply on Fridays.) Don’t be a team player. Take charge with leadership words. Don’t use personal pronouns. Include a Key Skills section. Apply in the first 4 days. Apply between 6am and 10am. Start your sentences with (distinct) action verbs. (Phew! Trying saying that five times fast.)

Easy peasy. You got this, right? Great. Now do that for every job you have to apply to and we’ll pretty much guarantee that you get the job you deserve.

Or, you can sign up for TalentWorkswe’ll just take care of it all for you! (No, seriously. [*]) Let us help you keep your New Year’s Resolution.

[*] For most things, we can just automatically take care of it for you. For instance:

  • Our AI-driven ApplicationAssistant automatically optimizes the day of week, time of day & delay of your application so you don’t have to worry about keeping track of 100+ applications.
  • Our ResumeOptimizer will instantly scan your resume for all of the potential issues above in addition to dozens of others.

And our A-List Talent? Sure, some of them might be in especially high-demand fields but, more often than not, they’re people who’ve put in a few minutes to optimize their resume for their job search. You should too!


Underlying Dataset

We took a random sample of 4,068 jobs, applicants and outcomes from recent activity on TalentWorks. For each case, we parsed their resumes with our ResumeParser, and annotated various applicant traits including gender, ethnicity, age, etc., and whether they had followed each of 70+ optimizations from our ResumeOptimizer.


Using partial least squares decomposition against interview rate, we then identified 16 principal components from the above dataset. Finally, we hand-selected a subset of the top factors in the first two principal components as the final 13 key factors.


We regressed the impact and estimated standard error of each factor across its domain using a composite Matern kernel. The results above are plotted with Bokeh on python.

Why Are We Doing This?

With ApplicationAssistant right now, we can boost the average job-seeker’s hireability by ~5x. But, what makes ApplicationAssistant work has been an internal company secret until now. We’re fundamentally a mission-driven company and we believe we can help more people by sharing our learnings. So, that’s exactly what we’re going to do.

Creative Commons

We’re not only sharing this but also sharing all of it under the Creative Commons Attribution-NonCommercial-ShareAlike license. In other words, as long as you follow a few license terms, this means you can:

  • Share: Copy, redistribute the material in any medium or format.
  • Adapt: Remix, transform, and build upon the material.

EDIT 2018-01-07 11am: Holy batman, viral post! In just 3 hours, we’ve been deluged by 100,000+ people wanting to learn more about the science behind the job search. Trying to get to everyone’s emails and comments ASAP.

EDIT 2018-01-07 2pm: We’re 6 hours and 250,000+ people in — wow! We’ll definitely be publishing a follow-up to this.

EDIT 2018-01-10: Since I simply couldn’t keep up with everyone’s comments, I wanted to address a few good technical questions that came up below and on Reddit:

  • We’re not trying to claim causation here. This was a 100% retrospective, correlation-based study. Although RCT-based studies are the gold standard and can establish correlation, they also require much more time and resources even after you know what to study. We’ll definitely be spending the time and money on teasing out causation in follow-up studies, but there’s still a lot of value in retrospective studies and we felt holding that back in a world where 51% of all 20-somethings are under-employed was just irresponsible.
  • We’re also not claiming that any one factor above fully explains the observed variance. In statistical terms, any single factor is independently a relatively weak predictor of your hireability [*]. However, when you account for multiple factors jointly, they are fairly strong predictor of hireability. Put another way, the coefficient of determination R^2 of the first two principal components described above is 0.719. For visualization simplicity above, we only graph one factor at a time.

[*] This makes sense, right? If you’re a Harvard MBA, we’re not claiming that applying on Fridays means you’re going to be forever unemployed.

Did you like this post? Share it with your friends!