What is Organizational Network Analysis? A Comprehensive Guide

1. Introduction to Organizational Network Analysis (ONA)

Organizational Network Analysis is the practice of analyzing an organization as a network - it’s about collecting data on who is working with whom inside of an organization and then utilizing that data to provide insight and ultimately improve the way that organizations function. 

Networks are incredibly important in organizations. We are often used to thinking about organizations in terms of the formal reporting structures that exist - the organization chart - but the reality is that the way that organizations function is far more complicated than that. There are multiple different types of networks inside of an organization - there are advice networks, friendship networks and mentorship networks to name just a few.

Organizational Network Analysis differs from traditional analytics through its use of networks - most of the data that you are likely used to analyzing is in rows and columns. Network data on the other hand is all about entities and connections between those entities. It’s more complicated (and also more flexible!) than that traditional row and column data format. ONA is all about applying network science to this network data, for example finding who has outsized influence in the organization by calculating network metrics such as PageRank.

2. Why Use Organizational Network Analysis?

Organizational Network Analysis is not a new field - though it has seen a resurgence of interest in recent years. ONA's origins can be traced back to the 1930s when sociologists manually created diagrams representing social connections using pencil and paper. They then calculated metrics to understand network behavior and identify influential individuals. Today we have modern tools and software to make this process far easier. Still the key benefits and applications of Organizational Network Analysis are much the same as they were a few decades ago. At Polinode we often call these core use cases the “bread and butter” of ONA and they broadly fall into two categories: 

1. Improving collaboration: What happens when sales and marketing aren’t talking to each other? What about the product team and sales? Often the successful execution of a company’s strategy depends on multiple teams co-ordinating and collaborating together. All too often there are gaps in collaboration and/or the quality of relationships between teams inside of organizations (and this is true of both large and relatively small organizations). ONA allows you to pinpoint those collaboration issues (including silos and over collaboration) and then to intervene to get things back on course.  
2. Identifying influencers: Often some of the most influential individuals in an organization don’t sit in senior management roles - they can often be in more mid level or junior positions. And too often the importance of the role that those individuals have been playing is only uncovered when they leave the organization. ONA allows you to uncover these often hidden influencers. It’s by no means the only application but utilizing these influencers as change agents is one of the most popular and common ways of leveraging the result of an influencer identification process.  

There are a range of other applications of ONA, including DE&I, organization design, onboarding and sales performance to name a few. Our intention in this guide is not to provide you with an exhaustive list but rather just an overview of some of the core or fundamental applications of ONA.

3. Foundational Concepts in ONA

In Organizational Network Analysis there are two foundational terms that it’s important to understand - nodes and edges. A node in a network is just an entity in that network. In ONA nodes are usually people (in particular, employees of the organization that is the subject of the ONA). However, nodes are not restricted to people or employees. They can, for example, be external connections or other organizations as is often the case when ONA is applied in the Non-Profit space. An edge is simply a connection or relationship between two nodes.

What does that edge represent? Well, it really depends on the source of the data. Often an edge results from asking people questions about their relationships with others. For example, we can ask them who they go to for advice, in which case the edge would represent an advice relationship. We also often access the digital communication of an organization in order to uncover the working relationships that exist and in that case an edge would represent a communication relationship such as via email or calendar appointments.

This is a great time to distinguish between two types of ONA - Active ONA and Passive ONA. Active ONA is where we run surveys to ask people about their key working relationships (such as the advice relationships example above). Passive ONA is where we tap into the digital exhaust of an organization, for example using email or calendar data.   

It’s also important to distinguish between formal and informal networks in ONA. The examples we have given above are all informal networks, i.e. the connections that exist between people as they work together to achieve the objectives of the organization. Formal networks on the other hand are reporting relationships that exist between two individuals (often represented by the Org Chart). ONA can also be used to analyze the formal network as it’s just a particular type of network that exists inside of an organization and it can also be used to compare the formal and informal networks (and to help inform the formal structure through organization design or redesign).

Another key concept in ONA is that of network metrics. There are a range of network metrics that can be calculated for the nodes, edges and the overall network. Polinode includes over 30 of these metrics, here though we will just highlight a few of the most commonly used metrics:

1. In Degree: In Degree is the simplest measure of centrality, i.e. it highlights which nodes (usually individuals) are more central in the network by calculating the total number of incoming edges or connections that each node has. In the context of Active ONA, this is the total number of nominations that each person has received from the other respondents to the relationship-based survey. 
2. Betweenness Centrality: Betweenness centrality helps you find the “bridges” or “brokers” in your network. The way that it is calculated is by computing the shortest paths between every pair of nodes in the network (i.e. the Nx (N-1) shortest paths) and then summing up the total number of times that each of those shortest paths passes through each node. A node is said to have a high betweenness centrality (i.e. be a “bridge” or “broker”) if a large number of these shortest paths pass through that node. Intuitively, to reach one part of a network from another part of the network you need to pass through that node. 
3. Pagerank: Pagerank is an algorithm that was an important part of the initial Google search engine (the internet is after all just a network!). It is an alternative measure of centrality for each node in the network but, instead of simply summing up the number of connections that each node has, it takes a recursive approach, assigning a node a larger rank if it has incoming edges from other more central nodes and so on.   
4. Community Detection: A community detection algorithm is an algorithm that partitions the nodes in the network into groups that are more closely connected to each other than they are to the rest of the network. One of the most popular community detection algorithms is the Louvain community detection algorithm and, in the context of ONA, this algorithm is commonly used to help understand the informal groups or communities that exist in an organization based solely on the structure of the network. This can then be used to help inform organization design decisions amongst other things.  
5. Density: The four metrics above are all node metrics, i.e. they provide a value for each node in the network. This last metric is a bit different as it’s an overall network metric. Density is a measure of how many connections exist in the network relative to the total number of connections that could potentially exist. Intuitively if you have a large number of nodes in the network but just a few connections then that network is not going to be very dense or, alternatively, if you have a small number of nodes but essentially all of those nodes are connected to each other node then that network is going to have a very high density. Density can give us insight into how strongly (or weakly) connected a group within an organization is. 

4. Getting Started with ONA: A Step-by-Step Guide

Step 1: Set Clear Goals for Your Analysis
‍Given the large number of applications of ONA and the flexibility of the approach, one of the most important steps is to first identify the strategic priorities of the organization that is the subject of the ONA, i.e. what are the particular organizational challenges or opportunities that this organization faces? This then should help inform the design and approach of the ONA. 

Step 2: Identify Stakeholders and Key Participants
‍The next step is to identify the stakeholders and participants in the ONA. It’s generally advisable to work with one or more executive sponsors to ensure the right level of internal support. You will also want to identify the population that data will be collected on. Often this is the entire organization but not always. Sometimes an ONA will be limited to a particular location or division. For larger organisations in particular an ONA may be limited to just a certain level and above. Small ONAs in large organizations can also be surprisingly effective, for example involving just the executive leadership team + 1 level beneath them. 

Step 3: Design Your Data Collection Process
‍When designing the data collection process for an ONA an important first question will be whether to run an Active ONA (i.e. collect data via relationship-based surveys) or a Passive ONA (tap into say email data or calendar data). Often the answer is actually both - the combination of Active and Passive Data can be considerably more powerful than either alone. 

If you elect to run an Active ONA you will want to design a set of relationship questions that will help you to achieve the goals of the ONA. No two organizations are the same and at Polinode we believe strongly in the benefits of tailoring an ONA to the particular strategic priorities and objectives of an organization. That said, we do provide an Enterprise template Active ONA that is often used as a base to be adapted as needed. 

If you are also (or alternatively) collecting passive data you will also want to identify the group that data will be collected for as well as answer important questions around privacy and optional anonymization of data. There is no single right answer to the privacy question in the context of passive data and the approach will often depend on the culture and location of the organization as well as the identified use cases. 

With Polinode it’s also possible to incorporate data from other sources as it’s possible to import literally any network or connected data with any attributes attached to the nodes / individuals in that network. For example, 360 degree review data is natural network data that can easily be imported into Polinode for analysis. Additionally, relationship data is often combined with HRIS demographic data from an HRIS demographic export.  

Step 4: Choose the Right Tools and Software
‍The next step is to choose your tools! You will want to make sure that the ONA tools that you select are able to both collect the data that you are targeting as well as visualize and analyze that data. Some of the questions that you may want to ask, include: 

1. Is the ONA tool flexible enough? Can it collect both Active as well as Passive data? Can I attach any kind of demographic data and import my own networks from external data sources?
2. Is the analysis functionality intuitive and well documented with a high level of support available?
3. How can the insights be shared with others? Here interactive network diagrams are generally better than static exports or outputs.
4. How secure is the ONA tool and is that security posture independently verified?

We would recommend a single integrated ONA tool like Polinode as this leads to a faster time to insight and more powerful / more actionable analyses. 

Step 5: Collect the Data and Visualize the Network
‍If an Active ONA is to be run it’s critically important to develop a high quality communication plan for all of the respondents. This should include an announcement at an event such as a Town Hall or regular meeting and the benefits for both the organization and the individual should be clearly communicated. Reminder emails and follow-ups should then be sent. 

If data is to be collected via a Passive ONA we would suggest using Polinode’s passive data integrations to make the data collection easy. This will typically require navigating a privacy and information security process as oAuth connection(s) needed for the collection of that passive data will need to be made to one or more data sources such as Office 365, Google Workspace and/or Slack data. 

Once the data is collected, networks will need to be generated. It’s straightforward to do this in Polinode by either clicking the “generate” button for an Active ONA survey which will generate a series of interactive networks or by running a “query” for a given time period over one or more connected passive data sources which will also produce a network.

Once the network data is available, a number of metrics such as those detailed above like In Degree and Betweenness Centrality should be calculated and overlaid on the network. These metrics can be combined with HRIS demographic data such as Level, Gender, Team, Function, Location and Tenure to create interactive visualizations and to understand how the organization is truly functioning. 

Step 6: Analyze and Interpret Results
‍At Polinode we can take these interactive network diagrams and produce an organizational report for you. These reports typically cover areas such as: 

1. Understanding and Improving Collaboration
2. Identifying Influencers
3. Looking at Greater Access
4. Diversity, Equity & Inclusion
5. Informal Communities
6. Engagement Analysis

We generally recommend one or more “sense making” discussions with key stakeholders to walk through the results and ensure that the context that only exists inside the organization is brought to the analysis. In this way we can turn the raw network data into a set of actionable insights. 

Step 7: Take Action and Communicate Findings

The final step in any ONA is to take action based on the findings. This really depends on what the objectives of the ONA were to begin with but can include: 

1. Changes to reporting lines or organization design to improve decision making and better align the formal structure with the informal structure
2. Running a program with identified influencers to connect senior leadership with shopfloor influencers
3. Identifying high impact change agents for specific transformation or change projects
4. Bringing two or more groups together that display a low degree of collaboration or low trust and running a workshop to improve the situation

ONA data is also the kind of data that can often be used for specific interventions or programs over the next 12 - 24 months. 

We always recommend communicating findings back to the participants, typically in an anonymized way. Frequently this is also complemented by sharing personal network reports with the individuals as well - these are short reports that communicate key metrics and insights to the individuals and can be used as a development tool. Polinode can also produce these personal network reports on request.  

5. Best Practices for Successful Organizational Network Analysis

To ensure a successful ONA we suggest: 

1. Define clear objectives upfront: It’s important to know upfront what are the high value objectives that you are looking to achieve from the ONA. It can be tempting to run an ONA without clear objectives in mind but this is generally not advisable. ONA can be used for a large number of use cases but the best outcomes are normally achieved when all of the key stakeholders understand the “why” from the outset.  
2. Engage stakeholders early: Relationships, even relationships within an organization, are generally seen as somewhat sensitive. It’s important to be clear as to why an organization is running an ONA and what the benefits of that ONA are for both the organization and the individual. Engaging key stakeholders early, including all participants, and proactively communicating these benefits is important. 
3. Run a small pilot: A lot can be learned from a small pilot. If the ONA is a large one or has some complications around it, then a “mini-ONA” for a subgroup is often run first. For example, where a whole-of-company ONA is being run, the same ONA can be run for HR only as a pilot first. Where the ONA is not particularly large or complicated, we would still recommend running the ONA on a small group (often of around 5 - 10 people) first. This way you can receive feedback on the specific questions and ensure that the pick list isn’t missing any data before launching the full ONA. 
4. Address confidentiality: The nature of Organizational Network Analysis means that it’s generally not possible to collect data anonymously. For example, respondents to an Active ONA survey need to select others by name. The one exception to this is passive data collection - it’s possible to collect passive data anonymously (and Polinode enables this) but this comes with some trade-offs around the insights that can be gathered. Most commonly, an ONA is not run anonymously but individuals are not identified outside of the small group that is running the ONA and conducting the analysis unless permission is received from that individual (for example, when identified as an influencer).

6. Common Pitfalls and How to Avoid Them

With over a decade of experience and hundreds of successful ONAs at Polinode, we have seen a few common pitfalls and have some suggestions for how to avoid them: 

1. Avoid focussing too much on metrics: Metrics like Betweenness Centrality and PageRank have an important role to play in ONA but there is more to ONA than just metrics. The context that a team or individual is operating in is very important and it’s essential to avoid making overly simplistic judgements based on metrics alone. 
2. Combine qualitative and quantitative data: When running an ONA we will often identify a set of influencers and create outputs such as team to team collaboration matrices. That kind of data is quantitative data and can be incredibly valuable and insightful but it’s important to go a step further and to use that kind of quantitative data to collect qualitative data. For example, to conduct interviews with a group of the identified influencers and to put the collaboration matrix in front of team leaders to understand the broader context behind it. 
3. Don’t overload the network survey: It’s often tempting, particularly when first getting started with ONA, to ask a large number of questions in an Active ONA survey. Long surveys can be counterproductive though and lead to a reduction in completion rates as well as reluctance to participate next time. Between six and eight relationship questions is fairly typical and a great deal of valuable insight can be gained from just those questions.  
4. Understand that networks are dynamic: Some organizations, such as rapidly growing scaleups, are changing very quickly and the networks within those organizations will also be changing quickly. Organizational networks are dynamic, not static as relationships change, people leave and new employees join. Depending on how quickly an organization is changing, an ONA dataset can become stale. For rapidly changing organisations this can be within 6 months. For organisations with a low level of attrition and moderate growth, ONA data can still be valuable in up to 24 months time.    
5. Be sure to act on the findings: Just like engagement data, analysing changes over time from ONA data can lead to valuable insights. For example, understanding how the degree of collaboration between two teams has changed over the last 12 months. That is to say that ONA is most valuable when run regularly (say on an annual basis or once every two years). To ensure consistently high response rates it’s essential to communicate the findings to the participants and to act on those findings (and to make sure that those actions are visible to and valued by the participants). 

7. How Polinode Can Help You Master ONA

Polinode can help you get to actionable insights faster through: 

1. Powerful interactive network analysis and visualization software
2. Full integrated relationship-based survey tool that makes it easy to run powerful and highly flexible Active ONAs
3. Integrations with key Passive ONA data sources
4. Organization reports, individual reports and other professional services for ONA
5. The ability to save multiple views for each network and to share those views with others internally - storytelling via interactive networks
6. A strong community of consultants and practitioners 

8. Frequently Asked Questions About ONA

What is Organizational Network Analysis (ONA)?
Organizational Network Analysis (ONA) is a methodological approach to analyzing an organization as a network by collecting and visualizing data on relationships and interactions between individuals, teams, or departments. It provides insights into who is working with whom, identifies key influencers, uncovers information flows, and highlights opportunities to optimize collaboration and help drive efficiency. By leveraging these insights, ONA ultimately aims to improve how organizations function and achieve their goals.

What types of data are used in ONA?
‍There are two broad types of data used for ONA - Active data and Passive data. Active data results from asking respondents questions in the form of a survey about their key working relationships. Passive data comes from digital sources such as email data, calendar data and enterprise social networks such as Teams and Slack. 

How do you conduct an Organizational Network Analysis?
‍There are seven key steps when conducting an ONA: 

1. Set Clear Goals for Your Analysis
2. Identify Stakeholders and Key Participants
3. Design Your Data Collection Process
4. Choose the Tools and Software (such as Polinode)
5. Collect the Data and Visualize the Network
6. Analyze and Interpret Results
7. Take Action and Communicate the Findings

What are the most common use cases for ONA?
‍The two most common use cases of ONA are improving collaboration and identifying influencers. There are many other applications of ONA too such as organization design, identifying hidden talent, DE&I, improving onboarding, preventing overload and mapping ecosystems. 

What tools are available for Organizational Network Analysis?
‍Polinode provides a single integrated tool for Organizational Network Analysis (both Passive and Active ONA). There are some other tools available that focus on particular areas of ONA, particularly tools designed for network analysis such as: UCINET, Gephi, NetworkX and igraph. 

What is the difference between ONA and Social Network Analysis (SNA)?
‍Organizational network analysis is the application of social network analysis (SNA) to organizations. 

Is it necessary to conduct ONA regularly?
‍Best practice is to conduct an ONA regularly for the whole organization say every 12 or 24 months. This allows you to examine changes over time. However, ONAs are also frequently done for specific projects or to support particular teams or parts of an organization. 

How can ONA help identify key influencers in an organization?
‍The most effective way to identify influencers in an organization is to run an Active ONA as this gives you the ability to ask questions targeted at cultural influence. Typically a number of different metrics are taken into account when identifying influencers (depending on the intended purpose of the influencer identification). This can include measures such as total nominations, betweenness centrality and energy.  

What metrics are commonly used in Organizational Network Analysis?
‍The most commonly used metrics are In Degree, Betweenness Centrality, Community Detection and PageRank. For more details about the metrics available in Polinode please see the metrics documentation here. 

9. Additional Resources for ONA Enthusiasts

If you are interested in ONA and would like to stay abreast of developments and cutting edge applications we recommend you sign up for Polinode Enterprise events. You may also enjoy the tutorial videos we make available. We also regularly run training on ONA - please get in touch to learn more about our training and accreditation and programs.