Response Rates for Network-Based Listening

Overview

One of the most common questions about survey-based organizational network analysis (ONA) is: what response rate do we need? The answer matters more for ONA than it does for traditional engagement surveys.

This article explains why response rates are especially important in ONA, what the recommended thresholds are, and how different factors can raise or lower the bar for your specific use case.

Why Response Rates Matter More in ONA

In a traditional engagement survey, a low response rate creates gaps in your attribute-based coverage. For example, if a team has a low response rate, you may not have enough data to report results for that team. That is a sampling problem, but the responses that were collected are still valid on their own.

In network analysis, a missing respondent does not just create a coverage gap; it distorts the structure of the network itself. Here is why:

•  Every person in the network is a node.

•  Every relationship between people is an edge.

•  A non-respondent is a missing node, and missing nodes mean missing edges.

When someone does not respond, you lose not just their answers, but all of the outgoing connections they would have nominated. This can make highly connected people appear less central than they actually are, make teams look more siloed than they really are, and obscure key brokers or connectors who bridge different parts of the organization.

Recommended Response Rate Thresholds

The following thresholds reflect general guidance from the organizational network analysis literature and OrgAcuity's experience with survey-based ONA deployments.

These thresholds apply to the population being analyzed. If you are conducting a department- or team-level network analysis, the relevant denominator is the size of that group, not the organization overall.

Factors That Affect the Threshold

The thresholds above are starting points, not hard rules. Several factors can raise or lower the bar for your specific situation.

Roster-Based Nomination (Lowers the Bar Slightly)

OrgAcuity uses roster-based ONA, meaning respondents select from a predefined list of colleagues rather than typing in names from memory. This partially mitigates the impact of non-response, because other respondents can still nominate a non-respondent, preserving at least their incoming connections even if their outgoing connections are missing.

This is one reason why survey-based ONA with a roster is more resilient to non-response than open-ended network surveys.

Scope of the Analysis (Raises the Bar for Smaller Groups)

Team-level and department-level network analyses are more sensitive to non-response than org-wide analyses, simply because the denominator is smaller. In a team of 10, two non-respondents represent 20% missing data. In an organization of 500, two non-respondents are negligible.

If you are planning to analyze the network at a granular level (e.g., within a specific business unit or function), aim for response rates at the higher end of the recommended range for that group.

Purpose of the Analysis (Affects Required Precision)

Not all ONA use cases require the same level of structural precision. Consider:

•  Identifying broad influence clusters or cultural patterns — more forgiving of lower response rates. Directional patterns tend to be robust even with moderate non-response.

•  Identifying specific change champions, connectors, or brokers — requires higher precision. Missing data is more likely to misclassify individuals.

•  Modeling contagion risk or predicting information flow — most sensitive to structural distortion. Aim for 80%+ wherever possible.

Which Network Metrics Are Most Affected

Some network metrics degrade faster than others when response rates drop:

•  Betweenness centrality (brokerage) — highly sensitive. Missing connectors disproportionately underestimate brokerage scores.

•  Degree centrality (volume of connections) — moderately sensitive. Incoming ties from others partially compensate for missing outgoing ties.

•  Clustering and community detection — moderately sensitive at lower response rates, but directional patterns tend to be stable at 65%+.

Tips for Driving Higher Response Rates

Because response rates directly affect the quality of your network data, investing in survey promotion pays dividends in ONA in a way that goes beyond traditional surveys. A few approaches that tend to work well:

•  Communicate the purpose clearly. Employees are more likely to respond when they understand how the data will be used — and that their individual responses will remain confidential.

•  Get visible leader sponsorship. A message from a senior leader explaining why the survey matters significantly increases participation, especially for ONA questions that may be less familiar than engagement items.

•  Keep the survey focused. Combining ONA questions with a broader survey increases response rates compared to fielding a standalone network survey. OrgAcuity is designed to integrate ONA natively into your standard survey program.

•  Use targeted reminders. Rather than sending blanket reminders, focus follow-up communications on groups or teams with lower participation rates.

•  Set realistic field windows. A 2–3 week field window typically balances urgency with accessibility. Shorter windows may disadvantage employees in different time zones or with variable schedules.

What to Do If You Fall Short

If your response rate falls below the recommended thresholds, you have a few options:

•  Re-field the survey. If you are below 65% across the full population, consider extending the field window or running a targeted push to non-respondents before closing.

•  Narrow the scope of the analysis. If response rates are strong in some parts of the organization but weak in others, you may be able to run a reliable network analysis for the high-response segments while flagging limitations for the rest.

•  Adjust your reporting thresholds. For results in the 65–79% range, flag the limitations in your deliverables and focus reporting on directional patterns rather than specific individuals.