Many speak about it, not so many have done actual projects in this space. A larger group is just starting to look into Organizational Network Analysis or ONA in short. When you are interested in ONA, you probably already know what it is. In short, ONA is about studying communication and other networks within formal organizations. In this post, I will give you some guidance and hints on what to focus on when you are starting off your first ONA project.
There are four main areas to consider when you are doing ONA projects. These areas are relevant in people analytics and machine learning projects too, but due to the nature of ONA some become much more relevant for the success of your projects.
In my experience, there are four major roadblocks . The first one is no clear goal or objective, the second is the definition and availability of the data used in ONA. It shouldn’t surprise you that Legal & ethical topics are on this list too. The last one and the biggest difference with other projects is communication with your key stakeholders and the subjects of the study.
Without a goal or objective, there is no point in ‘doing ONA’
This is not different from any project in analytics, yet with ONA there is much more at stake. You can only do ONA wrong once, so you better do it right. The best way is to choose a clear objective of what you want to achieve.
Focus on one question or problem and go into depth, get somebody on your project team with experience in ONA, but make sure you have business acumen and somebody who understands your organization too.
Don’t just ask a question to the data but also consider what you could potentially do with the answer. “Do we have silo teams in the organization?” isn’t sufficient, make clear what you would do once you find out which teams are operating in Silos. An example could be that you then consider the responsibilities of this team and whether you would expect them to work together with certain other teams. Once you know this you can follow up and foster collaboration between different teams.
If you don’t know what you want to do with ONA, a good idea is to start with your corporate values, strategy or HR initiatives that you are running.
Objective first, data and metrics second
Many companies will start with collecting data, for example through data collected in Microsoft Outlook. I encourage you to set the objective first. Data and metrics can follow the second.
Get your data sorted and relationship defined
Once your objective it’s clear, it will be time to determine the relationship you want to observe for your objective. “Work together on a project”, or “have exchanged at least 7 Emails over the last 3 months”. Determining the definition of your relationship, especially when using passively collected data, is extremely important and can make or break the rest of your ONA. Be prepared to spend some time on this.
From your objective, you should also be able to derive which network metrics you can get from the data. If you want to find bridge builders, you can use betweenness centrality. If you want to find silo teams, you can use community detection algorithms.
Let your objective drive the metrics and the data collection, not the other way around
Now that you have defined your relationship and your network metrics, you can derive several options on how to collect the relevant data. You can collect the data through Email-traffic, phone, instant messaging, calendar entries, RFID tags or survey data.
ONA projects require opt-in or opt-out from your subjects, the rate of subjects joining the study be large, ideally over 80%. Why? Because gaps in your network will distort the results significantly.
Have your Legal and Ethical points sorted
This is not something that can wait until the end of your project. Ideally your organization already has an ethics charter in place to build on, if not now is a good time to create one.
ONA can be sensitive for your employees, so make sure they are taken care of. With regulations such as GDPR, it is best to get your legal team involved and get advice about what is possible and whatnot. I would encourage you to understand the legal and ethical implications yourself such that you can explain these to anybody who asks.
Don’t just comply, link to your company values when deciding what to do with ONA
Some questions you could ask are
- Who will be affected by my analysis?
- How could the analysis positively or negatively affect the subjects?
- Ask yourself: would this analysis help me personally?
- Should the subjects be anonymized?
In most cases, you will be required to either follow an opt-in or an opt-out strategy with the subjects of your study. This and whether you want to anonymize your subjects will depend on the objective and data used in your study.
Communicate, communicate, communicate
Start in time with communication to the subjects, already before they become the subjects of your study. You need the subjects to opt-in or opt-out, and we have seen we ideally need at least an 80% participation rate. In most studies there has been a much higher participation rate when communication was done early and often.
Communication is key for the success of your ONA project
Engage actively with your subjects, allow them to ask you questions about what will be done. Be open about what will be done with the analysis, and make sure the objective will be positive for the subjects.
Consider these four points when you want to get started
Get a clear objective. Then define the relationship and metrics, derive the data you need. Consider legal and ethical implications. Communicate with your subjects early and often.
Get these points right and you will be better than most organizations in‘doing ONA’.
About me: I am an Analytics Consultant with a passion for pretty network pictures, Organizational Network Analysis, and effective collaboration.
