Interesting View: Google and Hiring


Most books about business aren’t worth reading because they’re either a selective collection of anecdotes with unknown biases and completeness (e.g. Malcolm Gladwell), or the writings of some obscure PhD focusing on some relatively secondary point (e.g. employee morale, while ignoring strategy and the competitive environment).

Fortunately, this book is an exception – summarising HR practices employed at Google, along with (in some cases) their evolution.

A key point made immediately:

Google managers cannot unilaterally decide whom to hire or fire, how someone’s performance is rated or rewarded, when software code is of sufficient quality to be incorporated into working systems, final product design and/or launch date, or who is promoted. Instead, each of those decisions is made by a group of peers, a committee, or a dedicated and independent team.

Outcomes are thus calibrated across groups, and managers instead focus on clearing roadblocks and inspiring their teams. Google contends that giving non-managerial individuals and teams, decision-making authority, providing learning opportunity outside what is needed to do one’s job, and increasing reliance on teams, works best. (See Kamal Birdi for related academic research.)

Block tells readers that Google relies on three cornerstones in its HR practices.

1.  Mission Statement – Cornerstone of its culture, intended to convey a mission that matters, a moral rather than business goal. Why? Nothing is a more powerful motivator than knowing you are making a difference in the world.

2.  Transparency – Weekly, Larry and Sergey host the entire company (live and internet video) for updates from the prior week, product demonstrations, welcoming new hires, and fielding 30-minutes of questions from anyone on any topic. (Staffers submit questions, discuss, and vote on their priority. ‘Hangout On Air Q&A’ is the software utilised for this and is also utilised by Obama to handle and prioritise questions from audience members when he speaks.)

Transparency is valued because it promotes information-sharing, encourages improvement (e.g. New York’s posting CABG mortality rates for each hospital, brought an overall death rate reduction of 41%). Bridgewater Associates (world’s largest hedge fund) goes further – recording every meeting and making it available to employees – shows how decisions are made, encourages more precise thinking and communications, reduces politicking/backstabbing, stops the ‘I never said that.’ Complaints about someone within an email are allowed to be forwarded to that individual.

3.  Employee voice – Google has an annual Bureaucracy Busters Program where staffers identify their

greatest frustrations, vote on them, and initiate actions.

Bock believes that most assessments of potential hires occurs in the first 3-5 minutes, if not sooner. Further, most interviews are subconsciously biased towards people like the interviewer. Finally, most interview techniques are worthless. (These assertions are backed by research cited in the end-notes.)

In addition, other research has concluded that about 90% of training doesn’t bring sustained performance improvement or behaviour change. Thus, Google spends considerably more on recruiting than the average firm, with less emphasis on training.

Even the best-intentioned managers compromise their standards as searches drag on.

Side Notes:

1.  As late as 2010, most new Googlers took significant salary cuts when joining – some as much as 50% or more. Hints at the value of adding Google to one’s resume.

2.  Google now prefers to take bright, hardworking students who graduated at the top of their class at a state school over an average or above-average Ivy League graduate.

3.  It also looks for humility and conscientiousness. Google only hires about 0.25% of those considered.

4.  As of 2008, Google continuously trawls the Web and reprocesses its findings several times a day.

5.  Continuing, Bock states that only 10% of an organisation’s applicants (at best) will be top performers – this creates a need for more interviews. Moreover, most top performers aren’t currently looking for work.

Two teams of senior leaders (one for product management and engineering roles, another for sales, finance, and all else), plus a final reviewer (Larry Page) enforce Google’s high hiring standards. A major rationale – this avoids declining standards over time, as well as hiring preference for friends.

6.  Google no longer uses case interviews and brainteasers – researchers have found them useless.

Google analyses have found that academic performance didn’t predict job performance beyond the first 2-3 years after college, so they’ve stopped requiring grades and transcripts – except from recent graduates.

7.  At one point, over half of new hires came from employee referrals, and then dropped off. The reason – they’d exhausted their networks.

Google now helps boost recollection by having events at which staffers attend through their Facebook, LinkedIn, and other networks, with HR individuals standing by. Why? Google was taking too long to contact those referred and not providing feedback to the referrer. That too has changed – now, an initial call is made within 48 hours and the referring Googler is given weekly updates.

8.  Since so many of the best potential applicants don’t apply, Google has developed in-house recruiting staff (using ‘gHire’) that now brings in over half of its new hires. It stopped using job boards (e.g. in 2012.

9.  Researchers Schmidt/Hunter (1998) researched how well assessments predict performance. Results showed: unstructured job interviews ‘explained’ 14% of performance variation, reference checks 7%, experience 3%. Work-sample tests 29% were best (Google uses). General cognitive ability 26%, when combined with assessment of conscientiousness, brought the level to 36%.

10.  Google now uses ‘qDroid’ – one chooses the attributes they want to test for and supply the job needing to be filled – the system provides questions appropriate for the interview. Bock also commends the sample questions available on the U.S. Dept. of V.A. website.

11.  In 2007, Google started looking for themes across the 10,000 or so individuals it had hired, and the millions it had not hired. Four predictive attributes were identified:

  • General Cognitive Ability
  • (Emergent) Leadership
  • Googleyness’ – enjoying fun, having humility, conscientiousness, comfortable with ambiguity, and
  • Role-Related Knowledge. The latter was least important, and Bock sort of waffles about it. Google now requires all interview feedback to comment on each attribute. Subsequent reviewers may disagree with the conclusions reached by those previously involved, and the system rates interviewers on their longer-term accuracy

12.  Google became infamous for its long, drawn-out hiring and interview processes. Analysis found that four interviews were enough (provided 86% confidence), and that each additional interview beyond those four only added another 1% confidence in the decisions made. Acting upon this information has cut the median hire time to 47 days, down from 90-180.

The process begins with resume screening by someone familiar with all Google jobs (Bock omitted the initial computer screening and key words/phrases utilised), a phone or Google+ ‘Hangout’ interview to assess general cognitive ability, then in-person interviews with hiring manager, peers, subordinates, and a cross-functional representative, followed by summarisation of the formal structured-prior interviews, along with ‘backdoor references’ (e.g. information from current Googlers who knew the individual before, perhaps in college), an interview with a senior leader, then the CEO. The average Googler spent 1.5 hours each week on hiring in 2013.

13.  Google tries to reduce ‘distance’ between layers – no executive dining rooms, parking spots, or pensions.

14.  Decision-making is based on data, as much as possible.

15.  HR practices are tested via surveys, test data, and sometimes double-blind experiments.

16.  Rejected candidates are sometimes hired to see how they perform.

17.  Purported biases are taken seriously and checked out via data – do those reporting to upper managers get more promotions and better ratings. What about those on the more important projects? (Analyses were conducted and the results posted. When the analyses uncover or bear out a problem, Google tries changes.)

18.  Engineers are allowed to spend 20% of their time on non-work-related projects (presumably related somehow to bettering Google – some projects have to do with HR changes), and a sizable proportion do so.

19.  Sidebar: Bock states that part of the reason women are paid less than men is that they negotiate less – he backs this up with data. Google disseminates this finding to staffers in an effort to encourage them to be more assertive in requesting raises and promotions.

20.  All supervisors receive structured feedback from their subordinates. Those with 100 or more subordinates have summaries of that feedback posted for all to see. (These results are not factored into their performance ratings or pay decisions). Google also regularly uses five questions that predict whether employees are likely to quit – action is triggered if favourable responses fall below 70%.

21.  Google Board member John Doerr introduced Intel’s goal-setting practices (Objectives and Key Results – OKR) to Google in the early 2000s. Results must be specific, measurable, and verifiable. (Bock suggests having both quality and efficiency measures, and cites Edwin Locke and Gary Latham as guiding authorities).

The firm deliberately sets ambitious goals, believing that if they achieve all of them, the goals are not aggressive enough. (‘If you set a crazy, ambitious goal and miss it, you’ll still achieve something remarkable’ – Larry Page)

Larry sets OKRs for Google at the start of each quarter. Everyone’s OKRs are visible on Google’s internal website. Since research has shown that hours spent cascading goals up and down a firm doesn’t improve performance, and those grossly out of alignment are quickly obvious. Google spends little time on that.

22.  Google stopped doing quarterly ratings in 2013 (now every six months) and went to a 5 point scale (from 41). Managers assign a draft rating to each of their employees and then sit with about 5 other managers and review the outcomes to help standardise/calibrate them. This is seen as reducing the incentive/pressure to inflate ratings, and improving fairness.

Prior to beginning the sessions (last about 3 hours), the group reviews the seven types of likely bias (e.g. frequency, central tendency). Not every individual is discussed, but they do look at the various ratings distributions.

Google separates the ‘how you did’ conversation from the ‘how to do better’ discussion. Managers are assisted in this by Google-provided guides.

Google believes that some individuals in technical jobs are worth far more than others at the same level – even after accounting for differences in project impact. Bock cites Bill Gates: ‘A great writer of software code is worth 10,000 the price of an average software writer.’ (When I wrote software the generally-accepted figure was a much more modest 20k – still quite impressive).

Unfortunately, this runs afoul of typical rewards systems. Google tries to account for these differences with varying stock grants, though these are infrequent (usually generate resentment); public recognition and less dramatic rewards (e.g. trip to Hawaii) are more frequent.

23.  Typical reward systems assume a normal-curve distribution of talent/contributions. Bock contends a ‘power’ (exponential) curve is more appropriate – if for no other reason than the fact that the worst performers don’t get hired or, are fired if they are hired.

He then documents that power-curve distribution of talent in a number of areas (academic publishing, professional sports, musicians, etc.) using data from O’Boyle/Aguinis and their study of 633,263 individuals.

Those falling in the bottom 5% at Google are provided training, sometimes given a new position (usually they rise to average), or fired. The range of rewards within a level at Google can easily vary 300-500%. A consequence is that average performance is rewarded with less than average compensation. As for failures – they too are rewarded if seen as thoughtful and well-attempted.

Bock/Google is relatively dismissive of training – Bock references research concluding that most is a waste, primarily because of a lack of evaluation.

24.  Google tries to use fellow workers as trainers (more credible and knowledgeable) and to assess behaviour/practise/performance changes that result.

25.  Bock ends by reporting that while Google does offer an excellent menu of benefits, many are free or are of very low cost to the company. Examples include allowing onsite provision of services for employees that help simplify their busy lives – cleaning or bicycle repairs for example. Employees pay for the services, although sometimes Google is also able to negotiate lower rates.

Overall – this is an excellent description of how Google recruits, retains, and motivates its staff.

To read Work Rules!: Insights from Inside Google That Will Transform How You Live and Lead, Laszlo Bock click here


One thought on “Interesting View: Google and Hiring

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s