[Book review] Work Rules!: Insights from Inside Google That Will Transform How You Live and Lead

This was the second time for me to read Work Rules!. The purpose I read this book again is to know the recruitment process of the early stage of Google and the concrete statistics data related to it. So I skipped the pages which don't relate to this topic at this time.



In this blog, I raise the recruitment process and statistical data that this book introduces.

How to get more candidates

About referral recruitment

Over a long time since its founding, the best channel WAS referral from the existing employees. But in about 2009, the pace of the introduction began to decrease.

In order to increase the referral, they tried "aided recall", which is one of the marketing methods and causes over 30% increase. For example, they changed the way of request "please introduce your excellent friends" into the way to ask "who is the best Ruby programmer in your friends?".

After that, they noticed really excellent persons were not looking for the job because they'd already been satisfied the working environment and got enough salary.

How to get the exact reference

When they get a CV, they try to find the common point with the existing employees, for example, the same university or the same company in the past, and send the email to them automatically to ask if they know its candidate and how they think of them. In this way, they can get more honest references.

How to make the list of candidates

1. List up the excellent companies whose employees can fit Google.
2. Make the list of employees of listed companies above.
3. Check each candidate on this list if someone is very familiar with the special field required for the position that they are looking for.
4. If they find the best candidates, contact them from the existing employees who have the common point (such as school or the past career) or recruiters then make the connections.

Staffing agency

They basically don't use the staffing agencies because they have strong personnel pool already. But in only the case that they can't find the local personnel pool, they use it. In this case, the quality of candidates depended on NOT the agency itself but the consultant in charge. So the selection of the consultant should be more careful than the one of the agency.

Job media

In the past, they used this service but the result was they just got lots of low-quality CVs like spam. So they completely stopped using it in 2012.

The other ways to recruit

These are very special so only big companies can try them, I think.

• If they catch the information of a very excellent person, they hire including their team members.
• If they catch the information of a very excellent person, they build the new office for them.

Summary of the recruitment strategy

Depending on the stage of the company, Google has changed the way of recruitment like this.

1. Define the condition of HR thoroughly and concretely that they require, and let anybody introduce the best talents.
2. Make the recruitment activities a part of the tasks of all members.
3. Request to spend more time to look for the excellent talents to those who have the excellent network.
4. Aggressively measure the result of the recruitment and keep improving it.

How to interview

Refer the next article "Statistical data" to know the reason why we should interview like this.

• Investigate what a candidate can do actually asking "please write the algorithm to realize this" instead of asking "please talk about the case you did this".
• Investigate the learning ability with the general cognitive ability test.
• Do the structural interview. It's composed of the behavior interview and the situation interview. The former is questions about the past result like "please talk about the story when you did ...". The latter is questions with an assumption like "what would you do if you were ...?".
• We can find the sample of the structural interviews on the website of The United States Department of Veterans Affairs.
• Not only managers who can be the boss of a candidate but also colleagues need to interview with them.
• At the end of the interview, make sure to explain why our project is important and how excellent colleagues you will work with.

Keep improving

When interviewing, their main concept is to give candidates the best experience and let them go back home in the mood like they spent the best day. So in order to increase the satisfaction of candidates,

• interviewers always ask them how they felt the recruitment process at the end of interviews,
• Google carries out the office tour after interviews if they have enough time.

As a result of these activities, 80% of candidates who didn't pass answered they want to recommend their friends to apply for Google.

Reference: Statistical data

In the early stage of Google, they interviewed with about 10 candidates from hundreds or thousands CVs. In order to write the feedback to them, they spent 10 - 20 hours. Until they hire one new member, they interviewed 15 - 25 times so they spent 150 - 500 hours totally for the recruitment process.
*This book introduces this fact as a failure experience.

In before and after 2013, Google was increasing about 5,000 employees a year. In order to hire them, they collected 1 - 3 million candidates. So only 0.25% candidates can pass their selection.

The best indicator for predicting job abilities is the work sample test and its coefficient of determination is 29%. This is the test to evaluate how a candidate can do sample tasks which they will do if they join.

The 2nd effective test is the general cognitive ability test (26%). This has the exam which has clear answers.

The structural interview has the same effectiveness as the general cognitive ability test (26%).

There is a "4 times rule". This is the rule that we can predict with 86% reliability if we should hire a candidate or not with 4 times interviews. With more interviews, we can increase only 1% for every 1-time interview.

If only one interviewer interviews with a candidate, the reliability of the judgment on hiring is almost zero.

If you liked this article

Let's subscribe the updates of Scuti!

Share on Google Plus

About Tomohide Kakeya

This is a short description in the author block about the author. You edit it by entering text in the "Biographical Info" field in the user admin panel.