Editor’s note: This is part one of the People Analytics Operating Model in the Age of AI, where we examine the roles and goals of people analytics, assess the existing operating models, and provide practical guidance on how to set up a Lean People Analytics Operating Model that connects technology, impactful consultation, and strategic decision-making, ultimately enabling HR success.
Stay tuned as we cover various aspects of operating models over the coming weeks, from challenges of people analytics, shortcomings of existing models, evolving skills in people analytics, and how to set up your PA team so it scales the value it brings to the business.
In this part, we look at:
- defining the goals of people analytics
- measuring people analytics success
- the current state of the people analytics operating model
- challenges and shortcomings of the current operating models
One of the best illustrations of the value of people analytics is the book-turned-movie, Moneyball.
And the traditional lesson is this: data-driven decision making will win out over gut-feel and even expert knowledge in the long-run.
So, given the current shift in maturation of people analytics, propelled in part by generative AI and modern technology stack, how does this lesson play out for people analytics and HR?
- What will this data-driven decision making look like?
- Where are we headed and what is the challenge to overcome?
- How can we best set up our people analytics teams?
- And which model to use?
We set out to answer these questions and propose an operating model that ensures the success of people analytics and HR for the current and — the shifts to come.
People analytics at a crossroads
Looking back, people analytics did not always have the respect amongst HR and business leaders as it has now. We had to learn, grow, and change in the past to earn the standing we have now.
To remain relevant, and perhaps even flourish like never before, in the future, we will have to continue that growth.
But right now, we are at a phase-change in maturation of the people analytics function.
The bad news is we are in completely uncharted territory.
The good news is we have been in uncharted waters before.
And as we go forward, we need a change to the way we operate.
Defining the goals of people analytics
What are the key goals that we’re to meet and that we’re struggling with as a people analytics community? When you’re a CHRO, how do you set KPIs for your People Analytics team? What is it that you want them to achieve?
Instead of building a structure for the sake of structure, let’s first start by examining the purpose behind people analytics teams. Let’s examine what we’re trying to achieve before creating a system that will take us there.
Starting point to building a scalable operating model for people analytics is answering the following questions:
- What is a successful people analytics function?
- What would be the criteria of success that divide successful people analytics functions and teams from mediocre or bad ones?
The top line is: adding value to the business. That’s all that matters.
But just how do we do that?
It is not as straightforward as the sales function, isn’t it?
Measuring people analytics success
While surveying a number of CHROs on how they evaluate the success of their people analytics teams, we heard a number of different answers that can boil down to a handful of success measurements.
Four key quantitative measurements surfaced:
- data integrity: data integrity refers to the quality of data and its alignment with the finance dataset and its general accuracy;
- time-to-value: how quickly can you produce an analysis that is needed to make it relevant and on-time;
- user-adoption rates: key stakeholder’s weekly active usage [WAU] of analyses output or engagement with the information and insights served, and
- time-to-action: how quickly the organization is able to use the analysis and turn it into tangible motion.
The qualitative measurement that kept reappearing was related to stakeholders’ assessment of the:
quality of insight: the quality of insights could be captured through structured interviews, open feedback analysis, or surveying and should assess the accuracy, methodological ethics, completeness, fit for purpose, confidence, and impact of the insight.
The question that follows is this:
How do you build a structure, an organizational model, that will increase the likelihood of your people analytics units achieving those goals?
Our task here is to propose the mix of talent, technology, and organizational/structural integration elements that will increase our data accuracy.
And they will also produce timely and impactful insights that will be frequently used by stakeholders at different levels to drive decision-making and action at the HR and business level.
What is the current state of the people analytics operating model?
The current state of the people analytics operating model has gone through many iterations and has had many influences.
One of the most popular models is the Insight222 operating model which was published in 2020. It is only one example of many competing models that exist.
Insights222 Operating Model. Source: Insight222: A New Operating Model for People Analytics
However, none of these models account for post 2021 market realities, and do not incorporate generative AI and modern technology stack into operations.
Overview of people analytics operating models evolution
Phase One: Reporting & data operating model
This model was a necessity in the beginning, but definitely not the place to plateau. The business needed data and people analytics reporting teams managed the data and provided the reports. Hello Microsoft Excel, goodbye free-time and being a “strategic partner”.
Phase Two: Project-based operating model
Taking a project-based approach added value, albeit narrow, to organizations and built pockets of deep expertise within people analytics. However, this model lacked scalability and required highly skilled resources to deploy insights.
Phase Three: Product-based (build or buy) operating model
This model was a positive step forward for our field in terms of technology adoption and scalability. At the same time, it brought about excessive overhead, cost, and waste due to the high level of investment it required for firms. Which, in turn, has led to the layoffs we’re seeing now in people analytics. People analytics must run leaner.
*NEW* Phase Four: Lean Operating Model (with Generative AI & Strategic Consultation built in)
We’ll cover this phase in the coming articles.
Challenges of the current operating models
Before we propose a new model, let’s first examine the current models, along with their biases and failures to tackle a few challenges along the way.
In short, those challenges could be described as:
- the potential failure points in our current people analytics setups to achieve the aforementioned goals
- fundamental inefficiencies that are lowering the ROI of the function, and
- the hidden costs of people analytics that are not accounted for by existing models.
Biases of today’s operating models
There seems to be a bias in today’s operating models (e.g. product-based operating models) towards an abundance market, where every organization is expected to have 20,000+ employees and 20+ person people analytics teams.
This is patently unrealistic.
The people analytics market is as dispersed, fragmented, and diverse as it has ever been.
Therefore, operating models need to keep in step. Solely focusing on a platform doesn’t allow you to scale because every problem is depersonalized and dashboard-ized.
Technology is a critical part of an operating model, but it must be oriented towards technology that will be utilized and adopted, not just implemented to sit on the shelf.
Here are some of the biases of the operating models of today’s median people analytics team:
- Focused on building (rather than buying, or even blending together an in-house tool and a third party solution) PA tools, which is highly expensive, inefficient, and takes a long time to add value
- Heavy business-intelligence, dashboarding focus and not enough use of automation
- Highly variable in the use of evidence-based practices and scientific rigor
- Not integrated and cross-functional enough, due to lack of interoperability with diverse datasets across the business
- Although sometimes aligned with “HR strategy” functions, oftentimes the data is used downstream of strategy, rather than influencing strategy in the first place
- Lacks strategic consultation and decision-making authority in HR
- Not using data orchestration engines and generative AI technology.
Why is this the case?
The current state of the people analytics operating model is a function of the evolution that people analytics has made since its onset.
Our field was sent astray by the relentless pursuit of the ‘predictive’ phase of maturity. This blog (and the operating model it proposes) is intended to diagnose, rectify, and prescribe the pathway forward.
In the past, if a senior people analytics leader wanted to get promoted, they would either have to:
- leave the field and take a leadership role in an adjacent field,
- attempt to become CHRO or other HR leader,
- start a people analytics function at some other organization, or
- get comfortable in their current role.
What if we added some constraints to an effective operating model?
What if, as a thought exercise, you need to maximize the ROI of people analytics with a maximum 5-person team? How would you approach building your operating model?
How would you scale from zero to one, making sure that you deliver value and address those CHRO goals along those steps?
Hint: by adopting a people analytics operating model based on modern technology stack and lean operating principles.
This way, people analytics teams will ascend in importance, seniority, credibility, and decision-making authority in organizations. A new path has formed.
Challenges and shortcomings of the current operating models
Let’s first explore some of the key inefficiencies that are not accounted for by most of the current operating models.
Underestimating data engineering work
Firstly, we are grossly underestimating the data engineering work in people analytics teams.
Anaconda’s 2022 state of data science report breaks down how data scientists spend time and almost half of the time is spent on data preparation, cleansing and administration.
People data comes with additional levels of quality checks which makes this work even more pronounced. Data science and data engineering talent is extremely hard to find — and the one with domain understanding and the passion for people data even more so.
To get fast results and limit the time to value this is the part that needs to be outsourced and heavily automated. These constraints are preventing people analytics from achieving data quality and time-to-value.
Dashboarding self-service paradigm
Secondly, the dashboarding self-service paradigm. The current limitations with the UX of people analytics is tied to our vision of dashboards and power-point reporting as a primary driver of user adoption.
Diagnostics are missing and as a follow up so is the lack of confidence and influence on decision making.
Research tends to lead to the need for more research to suggest tactical steps forward, which makes the field still overly academic. This prevents user adoption rates and increases time to action.
No room for data literacy
Data literacy and upskilling people analytics knowledge in the customers and partners of the people analytics team is a pervasive limitation.
Sometimes this responsibility is outsourced internally to L&D who doesn’t have aligned interests with PA teams, nor the skills to adequately meet the need.
People analytics teams themselves aren’t well equipped to do this either. So we need to be looking for a solution that would provide an opportunity to upskill people analytics knowledge across the stakeholder ecosystem.
Hiring for data expertise in business partnering roles
Speaking of partners of people analytics teams, we need to also revisit business partnering roles from the point of data expertise. Right now, the hiring practices for these roles are not adamant about data expertise. This limits the scalability and ultimately the success of any data-driven project.
For these efforts to scale, we need to challenge the HRBPs hiring practices and diversify HR talent so that it includes more hires with business acumen and data expertise. This acumen can also be shaped through data exercises and challenges that come with the platform and help drive the genuine use of internal data. Let’s make it real.
Cost of stakeholder’s curiosity
Another key inefficiency that is preventing the scale of people analytics is the cost of stakeholder’s curiosity.
In a service-driven model, HR and managers tend to ask questions that might not have a high impact value, but people analytics teams still respond to them and spend additional time fetching this information or insights. In turn, this leads to bottlenecks in people analytics teams which leads to additional hiring of analysts.
A generative AI application can help replace such need and increase the speed of response and user adoption by being tied to key communication channels, chat systems such as Slack and MS Teams.
People analytics only for people analytics’ sake
A quite prevalent problem in people analytics is the equivalent of people analytics navel-gazing. We only look in the mirror and not at the business and HR strategy support.
People Analytics should be geared towards enabling and executing the vision put forth by the CHRO.
However, sometimes people analytics is solely focused on “interesting” pet projects, bridges to nowhere, and research for which there is no audience. Going around hat in hand to try to drum up support for people analytics is not a good look on anyone.
Want to get your copy of the People Analytics Operating Model in the Age of AI and empower your organization with people analytics? Get your guide 👇
