How to identify the right people analytics project
When it comes to People Analytics (PA) there are always more questions than answers. The newness of the field, the complexity of human behavior and the ever changing nature of organizations generates a huge list of potential questions. It can be really easy for the aspirations and imagination of business leaders to overwhelm their people analytics teams with demands and unrealistic expectations.
The drive to deliver and show progress can quickly lead people analytics teams to over extend. However the fastest way to lose the interest of key executive stakeholders is to miss their expectations. Every emerging or growing, people analytics team needs a process for identifying which questions get answered and which do not.
Over our 10+years in the People Analytics space, Visier has supported hundreds of teams to get established and to scale their impact. From this we have identified and built a rich knowledge base of what works and what does not. The following decision model is the result of that work and something which can help you quickly assess, justify, and explain the work that your people analytics team will and won’t do.
Determining Project Viability
It is easy to generate interesting questions about how people are impacting the business. One of the exciting aspects of the growing field of people analytics is that there are more questions than answers! The opportunities to find new knowledge, new approaches and to differentiate the business through people programs is huge and in the early stages of being codified. However all organizations have finite resources, and there is no way to allocate resources to every question that comes up.
This combination of factors leads to the need for a decision model that is focused on balance. The model needs to balance multiple components related to value, data and resources. The goal of the model is to establish which projects or “products” are a go, and which are not. In doing so it needs to cover the spectrum of possibilities, from the most basic of reports to the most compelling and complex options for new research.
The diagram below shows the key components that should be assessed and balanced when making a decision about a specific people analytics project or product.
The 3 components are as follows:
Value: The goal of any people analytics activity should be to generate value for the business. It can be easy to lose sight of this when mired in all of the complex references and permutations of people data. In simple terms organizations spend up to 80cents of every expense dollar on their people. Actions or decisions which improve outcomes related to people will deliver value to the business.
The first thing to gauge for every project or activity is how directly it will deliver value to the business, and how clearly are the outcomes of people analytics work linked to this value. Is the linkage between action and value well established - like retention and cost saving. Or is it vague and undefined - like the impact of network connections on overall revenue.
For example, risk of exit models are now very commonly used. This is because their value is clear and easy to link to a business outcome. Not only does retaining an employee save the costs of replacing them, it can also be associated with an impact on revenue or risk. Visier’s own research shows that retention projects consistently provide a large ROI to the organization.
When it comes to gauging value there are two key questions to answer:
1.Is there a clear, well articulated and commonly understood link between the PA work and the proposed business impact.
This is a yes/no question. If you cannot articulate this link you should not proceed until the link is clear to all.
2. What is the scale of the value that can be delivered?
Data: The people analytics revolution has been fueled by the increased availability of, and ability to process, people data. Organizations record vast amounts of information about their people, past, present and future, every day.
For a people analytics project to be viable it has to be supported by data and that data needs to be properly structured to be useful. A recent conversation with a group who were exploring how to analyse learning behavior revealed that they had not adopted an effective data capture mechanism. Whilst they had lots of data, there was nothing in the data that would uniquely identify a specific learning experience. The same Youtube video or Harvard paper could be represented in any number of different record sets, and there was no way to break out the required lower level of detail. There was a record that indicated a person had experienced some learning. There was no way to determine what learning this person had experienced compared to anyone else.
As you can imagine it was a source of frustration and disappointment to this group that the data they thought was available was not, due to how their transactional system had been set up.
In this example the lack of properly structured data made it impossible to answer any questions about the impact of learning consumption. It demonstrates the critical importance of data as the fuel that makes people analytics work possible.
More commonly the judgement in relation to data is about whether or not there is enough data, of sufficient quality, to provide a trustworthy answer. People analytics teams should be able to perform a data review, processing a data set to see if it is sufficiently complete and well structured to support answers to relevant questions. There is rarely a need for perfect data, the focus needs to be on data that is fit for purpose. The people analytics team needs to be capable of evaluating if the available data is fit for purpose, and articulating this to their stakeholders.
It is also crucial that the organization has an ethics standard for the people data they will, and will not use. The data required to solve for the business question being presented must fit within this ethical standard.
When it comes to gauging the viability of the data related to the business question the information needed is as follows:
- Does the team have access to a properly structured data set, that aligns to our ethics policy, that will support effectively answering the business question.
Hopefully the answer to the above question is yes. However if the answer is no, that does not mean the question should be put aside. What becomes important is the cost or resources needed to generate the required data.
Resources: Resources covers a wide range of inputs, from the people on the team, to the technology, and external support that is available. You may have a high value question, for which data is available. However if you do not have the people with the time and skills, the technology to produce and maintain the product, the relevant stakeholder support, or access to the right people from other departments, the project will not progress.
An example of resource misalignment comes from a presentation I saw a couple of years back. A team had built a risk of exit model. It had taken 2 PhDs, 6months to assemble the data, run the model and generate the outcome. The CEO was excited by the result, and could see the value. The CEO wanted the risk of exit information provided monthly.
The PA team could not meet this expectation. Their methodology and infrastructure would not allow them to produce the result more frequently than once every 3months. They had the data and the value was clear. What they lacked were the resources to deliver the consistency of insight that the business needed.
This pattern is all too common, where an analytics team take a project-focused approach and provide a one-off answer. The answer may be interesting, but the value to the business only comes once the analysis can consistently support the decisions of relevant stakeholders. This is why it is important to consider resources up front. It is also important to consider resource demands for the full life-cycle of the project from conception, design, test, results and then scaling the output to deliver the full value to all the business stakeholders.
The best way to gauge the resource demands related to the business question is to use the structure below:
NOTE: One outcome of the above model is to look at technology investments, not as one-off tools, but on the basis that they can support a wide range of projects, and help to expand the volume delivered by the PA team.
Allocating Projects
Having outlined the components of the viability model the next phase is understanding how to put it into action. Project viability is a balance between the business value, data quality and the resources you have to approach the project. It would be possible to turn project viability into a complex study in its own right. We recommend, and practice, something that is more pragmatic, allowing for quicker decisions and only going into detail where the stakes are high.
The matrix below shows the projects which automatically get onto the list, those that never get on the list and those that require a judgement call. In each of these cases the team must have access to a properly structured data set, that will support effectively answering the business question and aligns to the organization’s data ethics policy.
High value projects with a low resource demand are clear winners. Business questions with low value and high resource demand are non-starters. The projects that fall into the muted green category are those that require careful judgement or further investigation before they get onto the teams worklog.
If the data set is not available then it becomes relevant to determine whether the potential business value from the project makes it worthwhile investing the resources to source the required data.
Summary
The practice of people analytics has matured over the last five years to the point where proven approaches and processes are being established and followed. One of the core practices which determines the success or failure of an emerging people analytics group is the ability to select and deliver only the most relevant pieces of work.
Through balancing the components of business value, data availability and resource capacity it becomes clear which projects should be prioritized and which should be dropped. This model also helps to communicate why these decisions are being taken to a broad range of stakeholders.