Last updated: 5 Aug 2016
What lessons from Google can we learn from the insane number of people experiments run there?
What do we know about working at Google?
For those who’ve never been into Google, it might sound like paradise on earth. Even after going in, it looks like….
What makes the Google culture tick isn’t all these material trappings though. Over the last 10 years, their head of “People Operations” (AKA Human Resources for non-Googler mortals) has run countless experiments to nudge – and sometimes shove – greater efficiency and effectiveness in extracting productivity from the prodigious talent they hire.
In a recent announcement, Laszlo Bock has chosen to step down after a decade of running Google’s people strategy. It made me reflect on the lessons from his experiments that I’ve learnt personally, and had the benefit of reading about in his neatly packaged book.
Are these insights relevant to me if I’m not in tech?
I’d say so.
Yahoo’s recent sale to Verizon as the “saddest $5bn deal in tech history” is yet another tangible indication of Google’s dominance in search. While it is tempting to point fingers at Yahoo’s current CEO Marissa Mayer and her sometimes draconian employee policies, what they did wrong wasn’t the perks offered to employees – which sometimes looked a lot like Google’s.
It is what was under the hood about the Google’s culture and approach to HR that made the difference – something very replicable in other companies from other industries and cultures.
These lessons even apply to SMEs (Small and Medium Enterprises)
Even though Google is now large and wealthy, their approach to HR has been consistent since they were small and resource-strapped. They grew from 8 employees in 1999, to 800 in 2004, to a gargantuan 57,000 in 2015.
While you might argue that much of that growth could be driven by their virtual monopoly in search (about 90% of Google’s income still comes from advertising), Yahoo’s failure to capitalise on that space despite being the first-mover suggest a deeper story.
Maintaining bench strength and a talent attraction engine nonpareil through their growth stages takes considerable strategic HR muscle.
As we run through the 3 biggest lessons below, pay more attention to Google’s problem solving approach, and less to the actual steps that they took.
Lesson 1: Build a solid process for your company’s most important people decision
One of the critical sets of experiments run at Google are those in hiring.
While it is true that the salaries offered at Google are high, Laszlo suggests that this just means they get more applicants, and not necessarily better applicants, nor that they can automatically apply a better filter to sift the great from the mediocre.
With the high salaries and blue chip brand, Google receives on average thousands of applicants for each job opening. Their challenge isn’t talent attraction, it is talent identification.
This is an important point in the process that many companies miss: narrowing the focus of your recruitment problem, and clearly defining it, allows for meaningful experiments and tests downstream that would help solve the root cause of the problem.
So what actually works in separating the wheat from the chaff?
Most traditional hiring tools have a remarkably bad record of predicting performance on the job:
- Unstructured interviews: 14%
- Reference checks: 7%
- Years of work experience: 3%
- Graphology (handwriting analysis): 0.04%
What does work instead?
For years, Google had been relying on gimmicky techniques like math puzzles on billboards and brainteaser interviews (eg “if you were shrunk to the size of an ant and put in a blender, how would you get out?”), but they found that these had very little correlation to on the job performance too.
What they did find, was the top three predictors of job performance were processes that very few companies adopted.
- Work sample test: 29%
- General cognitive ability (IQ): 26%
- Structured interviews (a consistent set of questions with clear criteria to assess the quality of responses): 26%
They also found that combinations of assessment techniques are better than any single technique.
From these insights, Google’s recruitment team continues to iterate and track the results of their experiments. They believe in the process so much that “people join and on their first day are trusted and full members of their teams” – a crucial part of getting the best out of their talent, which we’ll get into next.
Lesson 2: Systematically work out how to get the best from your people
In many Asian companies, the concept of employee motivation and engagement is still seen as a bit fuzzy and foreign. Traditional Chinese businessmen still believe that because they pay their people a salary, they should be thankful they have a job and work hard in return.
People management is seen as a chore at best, and a molly-coddling distraction at worst.
This actually became a serious question at Google: does management actually matter? [Note the clear statement of a hypothesis again] With a traditionally flat structure, and many brilliant engineers giving qualitative feedback that they thought their managers only got in the way of their productivity and added bureaucracy, the role of people managers at Google needed a strong reason to exist.
Meanwhile, Google’s regular employee survey, Googlegeist, contained rich data on what Googlers thought about their workplace, providing a baseline for useful analytics.
Takeaways from Project Oxygen
Not only did this study prove conclusively that good management was essential for high performance at Google, it also yielded 8 characteristics of great Google managers for replication. These 8 behaviours were then introduced into Googlegeist as a means of tracking manager performance, as well as into training programmes to bring new managers up to speed.
Important points to note:
- Taking action on survey results is vital in creating an upward spiral of performance. “There’s a virtuous cycle here: We take action on what we learn, which encourages future participation, which then gives us an ever more precise idea of where to improve.” The faster you can close the loop in this cycle, the faster you can achieve results.
- Good management matters. A lot.
Lesson 3: Stop grading performance on a bell curve
“Performance management as practiced by most organizations has become a rule-based, bureaucratic process, existing as an end in itself rather than actually shaping performance. Employees hate it. Managers hate it. Even HR departments hate it.”
– Laszlo Bock, Work Rules
Many of the discussions around performance management – whether to rank-and-yank, to use a 3- or 5- or 7-band system, how often to give performance appraisals – miss a critical point. They are all premised on the assumption that performance can and should be graded on a bell-curve.
It cannot. And shouldn’t.
For most jobs, human performance in organisations follow a power law distribution instead of a normal distribution. Bill Gates remarked once that “a great writer of software code is worth 10,000 times the price of an average software writer.” While the magnitude of the difference may be more pronounced in software than say, retail salespersons, the nature of the difference is the same.
A great writer of software code is worth 10,000 times the price of an average software writer – Bill Gates @EngageRocketco
If this is true, it also suggests that the rewards to extreme performance should have the flexibility to be equally extreme – assuming that measuring performance is tractable and accurate.
Transparency helps to improve the process: Laszlo recounts an experiment where a team of 100+ disgruntled engineers were given blinded performance data and asked to come up with a system to distribute rewards fairly.
Also, to alleviate accusations of promotions granted based on proximity to headquarters or job role, Google published promotion data broken down by location and job function to allow engineers to process the data for themselves. Interestingly, Laszlo strongly emphasises that extreme reward systems have both distributive and procedural justice.
Separate development from evaluation
Finally, in giving performance appraisals, Google differentiates between performance evaluation and people development. Divorcing developmental and evaluative feedback is essential: without doing so, employees focus on the extrinsic reward (a raise, a higher performance rating), and their openness to learning shuts down.
This is certainly true for the performance appraisal conversation, but I would add that while evaluation (necessary for the distribution of finite promotions or bonus dollars) needs to happen once or twice a year, development conversations need to happen as regularly as 3-4 times a month.
- Recognise that performance occurs in your organisation in a power law distribution
- Separate your evaluation conversations (1-2 times per year) from your development conversations (3-4 times per month)
Final thoughts on Google’s lessons
With close to 60,000 employees spread across so many countries around the world, it’s “not all rainbows and butterflies” working at Google. Things still slip through the cracks and poor management does rear its ugly head.
The fact is, “People” operations is exactly that: finding a way to help groups of people come together and achieve something great is always fraught with complex challenges. Yet with the right approach and evidence-based experimentation, Google has done a good job of consistently attracting and retaining the best coding talent in the world, even if it occasionally gets it wrong.
We need to start finding ways to adopt a similar approach to the people decisions we make in our own companies. If you get it right, maybe – just maybe – you’ll be sitting atop the next 50,000+ employee company.
Author: Chee Tung
CheeTung is the CEO of EngageRocket, an HR tech startup that analyses employee feedback in real-time to advise you on how to build a better culture, one team at a time.