People Analytics 101: How to Start When You’re Small

Published by:
Cameo Choquer
February 4, 2022
10
min read

Debiased Hiring Bundle

“We have charts and graphs to back us up... so f**k off!"

This was a sticker slogan by Google’s people team then led by Laszlo Bock. People analytics have come a long way since then. There are many teams and tools advancing people analytics and embedding it into company culture from the get-go.

When you’re not Google and you don’t have the same investment or resources to form a robust people analytics team, it means that you’ll be collecting and stitching together a lot of the data yourself and making more assumptions than the data-driven number crunching teams at FAANG.

In this article, I’ll explain what people analytics is and why collecting good data on your company’s people processes is incredibly important to inform decisions and strategy. In this post we’ll look at people analytics, ways you are already using people analytics, and why (and when!) making assumptions and finger in the air guesstimates is okay.

What is People Analytics?

People analytics, the application of scientific and statistical methods to behavioral data, traces its origins to Frederick Winslow Taylor’s classic The Principles of Scientific Management in 1911, which sought to apply engineering methods to the management of people.

People analytics is the practice of collecting, analyzing, and communicating people data from your organization to inform business decisions and outcomes. On the surface this can sound a bit…dry. But when you start to realise all the tools you use, your processes, and information to hand that can help shape your culture, it starts to get a bit more interesting.

As they say, culture eats strategy for breakfast and people analytics is the best tool companies have to reflect on their culture.

People analytics is more important than ever as the world has turned to automation and remote work to be more efficient, combat the pandemic and better understand the Great Reshuffle.

Organizations are realizing the processes and tools they have in place may no longer suit their needs in this new way of working.

Why is People Analytics Important?

HR is turning more into a people analytics function with 70% of company executives citing people analytics as a top priority. Many HR tech companies are developing mobile apps and tools to automate processes.

With that comes extra points of data that help companies move faster and focus more energy on creating trust and deeper connections between colleagues.

With the automation of our people processes and relying on the tools we use, making sure they’re used efficiently and ethically is also important.

There are some screening tools that have a bias against women, text fill-ins that autocomplete women’s names to men’s names, and engineering roles being shown to more men than women because the ads were more expensive to show to women.

The number one reason why people analytics matters is because you matter. While we like to rage read through anti-work subreddits and blogs and think that many executives and big corporations only care about profits, the people and their needs as employees and their experience with a company really matter to how well they run.

This is a balancing dance that companies have to learn and people analytics is their best instructor.

HR teams are lagging behind in their ability to collect and analyse data. If you speak with anyone on a people team going through a procurement process — they’ll tell you the HR ecosystem is fragmented and noisy.

There are lots of tools that do many things, and some of them doing too much of the superficial things. And these tools are incredibly important because they are how you collect data. If you choose the wrong tool, it can have a domino effect for people decisions — bad data = bad analytics = bad decisions. Below are the insights and traps to watch out for when starting to think about people analytics.

Getting the numbers

“It’s a shame that spreadsheets and cocktail parties don’t mix better.” This is the trick about working on a people team. It’s the point I have to reconcile with the most day to day. People aren’t numbers but in order to provide a great work environment, we have to work with them — whether it’s employee pay, benefits, pensions, eNPS scores, engagement, or emails — you have to work with numbers.

The more numbers, i.e. data, you work with, the less biased you’ll be in your decision making. This is a difficult one for those of us more comfortable working with people — we always think we know better even though research tells us time again that we are biased.

At the same time, we are under pressure to move forward and get things done which means getting comfortable working with the little information we have to hand — and finding comfort in discomfort.

How and what data we collect is key

When you’re getting a new piece of software that collects employee or candidate data, you should understand where the data sits, how accessible it is, the limits of what you can do, and if there are any algorithms you should understand in-app (sentiment and attrition analysis are the ones that come to mind). In other words, make sure you have scoped out why you’re getting it in the first place and what requirements are needed before buying the newest and shiniest thing on the market.

What data to collet for people analytics
https://www.mckinsey.com/business-functions/people-and-organizational-performance/our-insights/how-to-be-great-at-people-analytics

If you’re too small to invest in specific software and are using free tools — you can still reflect on what would be useful further down the line by researching benchmarks like eNPS, salaries, or staffing ratios.

When collecting the data it’s important to be consistent, even if you’re not using a piece of software. A great example of people analytics and selection is the information we collect at the sifting stage during recruitment and the questions asked. All of these questions are points of data that can help refine the processes.

A few questions to ask yourself:

  • Did we select these questions in a biased way?
  • Were the questions we asked the same for every candidate?
  • Did we use a consistent review guide (scorecard) when assessing the candidates?
  • Did we unconsciously change the wording of the question for female candidates?

These sorts of questions and data collection will help you massively the next time you do another recruitment round.

Analyzing the numbers

If you’re at a start-up or scale-up there’s a good chance you don’t have (m)any data scientists or Business intelligence analysts to support your people team. This is where we often rely on the reporting mechanisms we have in our tools (like Applied, Culture Amp, Payfit, etc).

This isn’t a bad thing by all means and good data is better than none at all. Remember, don’t let being perfect get in the way of being good.

However, when we have a limit in capacity or expertise to analyse the data, we need to address 2 traps: context and sample size. If you are collecting answers to a survey you sent out but there’s only 20–40 people in your company, you may be getting better answers by conducting interviews with a few people in the company directly rather than sending out surveys.

This is because when we’re asking more complex questions like “do you feel comfortable asking questions in meetings?” we need a much larger sample size to ensure that one or two individuals aren’t influencing the entire outcome of the survey.

Even for basic questions that we know to be true, for example, men weigh more than women, we still need a sample of n=46.

When we are doing people analysis like this, we’re aiming for statistical significance to tell us something, to inform what decision we should take.

“Statistical significance helps quantify whether a result is likely due to chance or to some factor of interest

When we have a finding that is statistically significant, we can more confidently say that the numbers we have are enough to take action.

The other way that this could impact our decision making is context. If we send out a survey to the entire company but a quarter of the company has just joined, but the question relates to the away day they weren’t a part of, the context is totally off and the results of the survey would be skewed.

When conducting surveys, you should ask yourself :

  • What teams/individuals are answering these questions?
  • Is there enough people to answer these kinds of questions?
  • Are we confident the data isn’t spiky?
  • Are there any demographic gaps in survey responses?

A note on performance

Performance in sport is much easier to analyse than performance at work — it’s far more consistent, has clear rules for engagement and success — it’s a much more ‘controlled’ environment than the workplace.

The infographic below is a clear representation of how an individual’s performance can change depending on their context — how you perform can totally depend on what field you’re playing in. In Baseball there are fields known as ‘hitter fields’ and ‘pitcher fields’ which means there are some environments that are just built to favour one type of player over another.

How environmental factors impact individual performance
How environmental factors impact individual performance

When it comes to performance reviews, make sure that context is factored in and ask yourself: Do employees in the same role have similar roadblocks or accessibility to information? Did they start at the same time? Are they in a different time zone?

One way to combat this is to have consistent questions for all of your line managers and specific criteria for their reviews, and even better, tying those reviews into specific frameworks — either for their role or specific criteria your company has for promotions.

Mind the people analytics traps!

We’re all suckers for a good story. As humans, we try to find patterns when there aren’t any, try to connect all the dots, and create narratives in hindsight. These are the biases that help us to ignore the persistence of data and confuse causation with correlation.

underfitting, optimal, and overfitting models
Underfitting, optimal, and overfitting models

The most recent and spicy example of this is the Great Resignation.

When it comes to hiring, the number of jobs that go live will change from quarter to quarter — we have to accept that hiring markets are seasonal. The Economist recently crunched the numbers on the great resignation and analyzed if the data held up to the observation that people were quitting in droves.

When they looked at the numbers across the globe, the only places where there seemed to be a sliver of truth were in the US and the UK.

When they looked even closer and considered the seasonality of the labour market, fast wage growth, and the unusually high number of vacancies, the great resignation seemed even more transitory.

Callum Williams at the Economist explains that an increase in job vacancies is correlated with number of job quits — if there is an increase in job listings, you’re more likely to find something that you like and therefore likelier to quit.

Meaning, we’ve all been caught up in the exaggerated narrative that people are quitting in droves out of the blue, when in reality, it isn’t persistent data and is correlated to job growth.

If you’re reflecting on the retention at your company, keep these things in mind and be aware of the persistence and causal v correlation of data. Use your survey tools, conduct interviews, conduct your own collection of data rather than relying on the noisiness of LinkedIn to tell you what trends to expect at your company.

Three things to remember

If you’re a people manager at a start-up or scale-up, you’re on an exciting adventure of building a company and at the very beginning of your people analytics journey. You’re at the “fun” part where you have the opportunity to collect, experiment, and work iteratively.

Like all the other work you do, you’ll make mistakes and get things wrong. But if you keep in mind these three simple reminders, you’re well on your way to building out a rigorous and data driven people function.

  1. Start collecting and experiment — take an iterative approach to planning and decision making in the early days.
  2. Remember to scope out what questions you are asking and trying to answer — you won’t be able to focus on everything so speak with your leadership team and decide on what you want to focus on for the year.
  3. Mind your people analytics traps — keep in mind sample size, context, seasonality, and correlation vs. causation

At Applied, we think one of the biggest factors in a company’s success is its people — and the best way to shape that is when they come through the door. So, if you’re looking on where to start — start with hiring!


Applied is the essential platform for debiased hiring. Purpose-built to make hiring empirical and ethical, our platform uses anonymized applications and skill-based assessments to identify talent that would otherwise have been overlooked.

Push back against conventional hiring wisdom with a smarter solution: book in a demo