← Mark McGranaghan

Thoughts on Recruiting


Recruiting is so important. It’s perhaps the biggest factor in company success. It determines where individuals will spend years of their professional lives. But the sad reality is that we’re terrible at is an industry.

Below I’m growing a set of tweet-size ideas on recruiting. I’ll keep adding / editing over time:

  1. “The tech industry is fundamentally unserious about how it recruits, hires, and retains candidates.” — @patio11
  2. When companies say “recruiting” they usually mean having good people working at the company. The best way to improve that may be training and retention, not new hires.
  3. The most likely reason you’re having trouble hiring is that you’re offering an undesirable job (company / role / compensation).
  4. The second most likely reason is that you’re offering a desirable job but people don’t know about it.
  5. Read Dan Luu on hiring market pathologies.
  6. A fast-growing engineering org will spend much of its bandwidth on growth: recruiting, interviewing, onboarding, training, org changes, team (re-)forming, process updates, etc. Can easily be e.g. 1/2 of total team bandwidth.
  7. Fast engineering org growths tends to tax your tenured staff the heaviest, exacerbating this productivity hit.
  8. This makes growth efficiency sensitive to attrition rates of tenured staff.
  9. Companies atomize the recruiting process as they scale: sourcer, recruiter, on-site coordinator, hiring manager, interviewing manager, executive closer, etc. This is more convenient for the hiring company, but is it a better experience for the candidate vs. a single point of contact?
  10. Much "standard" sourcing, screening, and interviewing is cargo-culted process, institutionalized superstitions, internal politics, and metastasized principal-agent problems.
  11. Read Thomas & Erin Ptacek on hiring processes.
  12. The best predictor of future performance is recent performance in similar situations.
  13. Work sample tests give you the most recency (now) and pretty good similarity (can be made close to real work scenarios).
  14. Whiteboard coding is not similar to a real work scenario.
  15. Hands-on-keyboard coding is only a part of the job of software engineer and should be only the part of a work sample test.
  16. Thorough work history reviews are not totally recent by definition, but give a good view into how people behave in real scenarios.
  17. Ask 2+ layers of follow-up questions on work history reviews. People construct nice personal stories but have to fall back to reality on follow-up questions.
  18. Interview processes should be designed as much to help the candidate choose to join your company as to identity the strongest candidates.
  19. Candidates impute the quality of a company from the quality of their interview process.
  20. Capture-the-flag competitions are wildly underused both for identifying talent and signaling that the hosting company is a clueful employer.
  21. You can measure the information- theoretic effectiveness of interviews at predicting performance in the role, but very few companies do. (Deep down, they know the results would contradict their hiring superstitions.)
  22. Before selling a candidate on joining, understand their criteria for choosing where to join.
  23. The problem of false negatives is rarely systematically considered in designing hiring approaches, and AFAIK almost never empirically measured.
  24. Candidates with pedigrees (e.g. elite undergrad school) may be more likely to do well in the role than other candidates (maybe!). But most candidates who would do well in the role don’t have pedigrees. And everyone is already competing for the pedigreed candidates.
  25. Many very strong engineers come from totally unassuming / random backgrounds. It’s a huge mistake to judge candidates by the brand name of their university or past employers.
  26. Read Patrick McKenzie on salary negotiation.
  27. The idea that “startup equity is worthless” is a classic case of “it’s true until it’s isn’t”.
  28. Properties of startup equity vary with the stage and quality of the company, in ways that are legible to candidates. As a candidate you can’t pick winners, but you can give yourself a much better chance of a good outcome.
  29. The terms on equity for employees are usually really bad (low %s, opaqueness, preferences, illiquidity, exercise windows, etc.). It’s been getting worse with increasing times to IPO.
  30. Companies will find a way to improve these terms as a way of attracting candidates. (So far we’ve seen only modest efforts around the edges.)
  31. When considering whether to ask for more cash vs. more stock when negotiating, note that it’s more likely cash comp will be normalized within the firm over time, vs. stock comp.
  32. The most valuable benefit from working at an outstanding firm is increased professional capital: experience, skills, network, etc. (This is not an excuse for low $ compensation.)
  33. There are a lot of mediocre staff making $300,000 - $500,000 / year (or more) in total comp at the tech giants.
  34. Companies pay ~20% of first-year salary (so about $40,000-$50,000) to external recruiters for hired candidates. Remember this when you’re negotiating as a candidate.
  35. The tech giants pay less in markets outside the Bay Area. But the difference can be much smaller than the cost of living difference. Some non-BA staff arbitrage this: get paid almost Bay Area rates, pay nowhere near Bay Area costs.
  36. A good option may be to move to the Bay Area for 4-8 years and then move back out.
  37. Ask the receptionist how the candidate treated them.
  38. Are your current team members genuinely excited to refer their friends to your company? The honest answer may surprise you (if you can get it).
  39. Serious recruiting efforts should have dedicated full-time teams that include product, engineering, design, marketing, and statistics expertise.
  40. If you find statistical thinking weak in the academy, wait till you see how companies talk about their hiring pipelines.
  41. We leave too much of recruiting to Recruiting teams. Functional teams should take more responsibility for their own recruiting.
  42. People vary wildly in their ability to evaluate candidates. Some provide zero or negative signal. We rarely measure or act on this.
  43. Perhaps "technical interviewer" should be a specialized, full-time, highly-paid position. Or at least done full-time on a rotational basis.
  44. Interviewing skill improves with practice. Many people get zero practice before subjecting candidates to their interviews.
  45. Prepare candidates for their interviews. Tell them what to expect and how to prepare. There shouldn’t be any surprises.
  46. Any single type of evaluation — individual coding, pair programming, take-home exercise — will be bad for some candidates. Consider giving them flexibility to show their best work.
  47. Don’t require candidates to front money for reimbursable expenses like travel. (Some will not able able to.) Don’t require a reason for them not to front the money.
  48. A fundamental challenge of recruiting is incomplete and asymmetric information, both about candidates and companies.
  49. Since most companies are bad at recruiting and in fact bad places to work, candidates by default have low priors for your company. A little positive signaling goes a long way.
  50. A similar dynamic goes the other way for candidates.
  51. The critical recruiting problem most companies face is attracting candidates, not filtering them. Yet most focus on filtering.
  52. Job posts should focus on making candidates want to join the company for that role, not filtering candidates.
  53. Every "required" skill / experience in a job posts costs you candidates, some of whom could have done very well in the role.
  54. Resumes show if the candidate can write a coherent document (which is important, and many candidates can’t do). But not much more.
  55. Recruiters screening out candidates at the resume stage based on anything other than document-coherent-ness is therefore suspect.
  56. Interview processes should help candidates assess companies, as well as the more typical other way around.
  57. Be careful when interpreting reference checks; most people are poor judges of talent.
  58. References can help you work more better with someone if you do hire them. ("How did you collaborate most effectively with X?")
  59. Deciding to join a company for probably 1-4 years on the basis of a half day of interviews isn’t good for the company or the candidate. (Doesn’t stop almost everyone from doing it though.)
  60. Most companies underweight recruiting / staff development in performance evaluations of managers and executives.
  61. Hiring managers at fast-growing companies often under-invest in recruiting. Optimal allocation may be 25-50% of their time.
  62. External management hires at director+ should be able to bring multiple candidates from their network into the company in their first year.
  63. Moving fast is the most unreasonably effective way for companies to improve recruiting outcomes.
  64. Examples that every company could do but most don’t: respond to candidate emails same day, schedule interview for first day candidate is free regardless of how much interviews need to move meetings, make an offer the day after the onsite interview.
  65. Externalize your company culture to attract aligned candidates. But remember: culture is how you trade off among plausibly good things, not platitudes.
  66. Recruiting is a long game. Some people take 5+ years to hire. The most effective recruiters work on this timeline.
  67. Someone will land the Starfighter idea and it’ll be awesome. (If you’re working on this, email me!!)
  68. You can only call it "unregretted" attrition before they say they’re quiting.
  69. If "people leave managers, not companies", do they also "join managers, not companies"?
  70. Hiring is a two-sided marketplace problem, with a need for deep vetting in both directions.
  71. We’ve seen some attempts to aggregate and vet supply (candidates). But few to aggregate and especially vet demand (companies).
  72. Successful recruiting companies will verticalize more, beyond job listings and especially into candidate prep, support, and evaluation as well as company evaluation.
  73. The hiring status quo is so bad, especially in terms of principal-agent problems, that it’s hard to fix as a third-party service. The best thing to do with hiring skill may be to capitalize it into a non-recruiting-related startup.
  74. More companies should have alumni programs.

I’m fascinated by recruiting and would love to help you achieve better results. Please do email me: