Network Research · Ars Network
See the network you have.
A research-grade network survey platform built for the way knowledge actually moves through your organization. Three clean CSVs on study close: nodes, edges, responses. Ready for R or Python, every time.
Built for researchers, consultants, and people leaders across education, healthcare, the social sciences, and the workplace.
The Problem
The questions your org chart cannot answer.
- Who in your organization is actually using AI, and who are they learning it from?
- Who do people really go to for advice, and who is carrying more of that load than anyone realizes?
- Where do the practices on the ground diverge from the policy on paper?
- Which connections did your program actually create, and can you show a funder the evidence?
Every leader and every researcher studying organizations is sitting with some version of these questions. They are network questions: not "what does each person think?" but "who is connected to whom, for what?"
The answers exist. Your people can tell you exactly who they go to and who they learn from. The problem is that almost no tool collects those answers in a form you can analyze. So the questions go unanswered, and the policy stays on paper.
The Honest Comparison
You could run this in the survey tool you already have.
Generic survey platforms are good tools, built for a different problem. Here is what running a network study through one actually costs you.
The data comes out flat
Generic tools store one row per respondent. Network analysis needs one row per relationship: who named whom, for what, how often. That difference is structural. No amount of clever question design in a generic tool produces edge lists and node tables.
30 to 40 hours of wrangling
That is the typical cost of restructuring generic survey exports into analyzable network data, per study. Three weeks of CSV (comma-separated values) cleanup before you can see a single network. Most teams do not have a network specialist sitting around for that.
The instrument is on you
Name generator wording, roster design, and follow-up logic determine whether your network data means anything. Generic tools give you a blank page. Getting the instrument wrong is the most expensive mistake in network research, and you find out after the data is in.
How It Works
Three steps from question to clean data.
Launch a study. Share one link. Get clean network data.
Build from a validated instrument
Start from a field-tested template or compose your own study from eleven network-native question types, including roster-based name generators and reactive alter loops that ask follow-up questions about each person a respondent names. Paste in your own institution-approved consent text, verbatim. The platform never supplies consent language for you.
Launch to your roster
Invite respondents by tokenized link. One question per screen, accessible to WCAG AA (Web Content Accessibility Guidelines) standards, with save-and-finish-later built in. Launching locks your roster automatically: the platform enforces your IRB scope mid-study, so memory doesn't have to. Response caps pause collection gracefully. Your data is never locked away or discarded.
Export and analyze immediately
On study close, download three clean CSV files: nodes, edges, and responses. Excel-friendly, and ready to load straight into R or Python (igraph, networkx, statnet), UCINET, or Gephi. Zero wrangling. That is the product.
Who It's For
Built for the people who run network studies.
Different jobs, same underlying need: relational data you can trust, in a form you can use.
Defensible from consent to export
Your institutional review board (IRB) approved specific consent language; respondents see it verbatim, because the platform never substitutes its own. We cannot view your workspace contents at all: support access exists only through a time-limited grant you issue and revoke.
Exports load straight into R or Python, so analysis starts the day collection closes. Academic licensing is available, including letters of support for grant proposals.
Repeatable instruments, client after client
Organizational network analysis (ONA) engagements live or die on instrument quality and turnaround. The validated template library gives you field-tested name generators and rosters you can rebrand and redeploy per client, instead of rebuilding from scratch each time.
The export bundle drops into your existing analysis stack, so your margin stops going to data cleanup.
See adoption, then act on it
Map who is actually using AI and who they learn from, or surface the advice network your org chart hides. Once you can see who carries the collaboration load, you can rebalance it. Once you can see where adoption stalls, you can target enablement instead of blanket training.
Run it as a wave: baseline now, repeat after the intervention, and show the change.
And the method travels much further than three job titles. Education researchers, public health teams, healthcare clinicians, social scientists, foundations, nonprofits, and coalition builders all run the same kind of study, because the core network questions recur across every field that studies relationships. Only the noun changes:
"Whom do you turn to for advice about teaching?"
Teacher collaboration, instructional networks, program evaluation.
"Who do you turn to for information to get your work done?"
Organizational network analysis, AI adoption, advice and energy networks.
"Who do you go to for health information?"
Care coordination, health information flow, community coalitions, social support.
"Which organizations do you partner with, and how closely?"
Grantee and coalition networks, collective impact evaluation, evidence for funders.
Instrument Library
Start from instruments that already work.
Every template is a complete, validated study instrument: question wording, roster logic, alter loops, and export schema designed together. Validated starter instruments ship with every plan; the full library is available by subscription or included in Team and Enterprise.
Advice Network
The canonical who-do-you-go-to study. Surface the informal advice structure your org chart cannot show you, from hidden brokers to overloaded experts.
Collaboration & Workflow
Trace how work actually moves between teams and departments. Built for reorganization planning, merger integration, and cross-functional friction diagnosis.
Learning & Professional Networks
For education and research settings: map learning ties, mentorship, and program-attributable connections across cohorts, schools, or professional communities.
See It With Your Own Study
Bring a real question to the demo.
The fastest way to evaluate Ars Network is with a study you actually want to run. We will walk your instrument through the editor live.
Request a DemoWhy Networks
The org chart is not the organization.
Decades of organizational research point the same direction: work, advice, and innovation flow through informal networks that formal structure does not capture. Once you can see those networks, you can act on them.
Research on collaborative overload finds that in most organizations, 20 to 35 percent of value-added collaboration comes from only 3 to 5 percent of employees. Seeing who carries that load shows you exactly where burnout risk and single points of failure sit.
Running a network study through a generic survey tool typically costs a research team 30 to 40 hours of data restructuring before analysis can begin. Ars Network's export bundle removes that step entirely, so analysis starts the day the study closes.
When organizations ask "who do you go to first?" about a new practice such as AI tooling, the answers reliably reveal brokers and bottlenecks leadership did not know existed. That is where targeted enablement beats blanket training.
Built for Research
Designed around the standards research demands.
Privacy by architecture
We cannot view the contents of your study workspace. Access for support exists only through a time-limited grant that you issue and you revoke. For studies under IRB oversight, that is a property of the system, not a promise in a policy document.
Your consent language, verbatim
The platform never supplies default consent text. You paste in the exact language your review board approved, and that is exactly what respondents see.
Accessible by default
Respondent surveys meet WCAG AA contrast as a floor, with body text at AAA. Full keyboard navigation, screen reader support, and 200 percent zoom without breakage, on every study, with no configuration required.
Caps that never cost you data
We never bill per response or lock you out at a cap. When a study reaches its soft cap, collection pauses and every response already submitted stays safe and exportable. No surprise overage invoice, no held-hostage data, ever.
Locked at launch
The moment a study goes live, its roster locks. Enforced by the platform at every layer, not promised in a policy: you cannot accidentally expand your IRB scope mid-study. Need new participants? Close the wave, duplicate it as a fresh scope-clean wave, and launch. The right thing is the easy thing.
From draft to data in two clicks
Launch generates your shareable respondent link, locks the roster, and starts the audit trail in one atomic step. On close, one click downloads the full CSV bundle.
A Short Primer
What is social network analysis?
Social network analysis (SNA) is a research method for studying the structure of relationships among people, teams, or organizations. Instead of asking only what each person thinks or does, it asks who is connected to whom, for what, and how strongly. Organizational network analysis (ONA) applies the same method inside workplaces.
The raw material is simple: a list of people (nodes) and the relationships between them (ties). From those two ingredients, network analysis answers questions that attribute surveys cannot touch. Who is central, and who is isolated? Who bridges groups that would otherwise never talk? Where has a community formed, and where is a silo?
What makes the method demanding is not the math. It is the data. Network data has to be collected relationally, person by person and tie by tie, with carefully worded questions. That is the part Ars Network is built to get right, so that the analysis, in whichever tool you prefer, starts from data you can trust.
Five Terms, Plainly
- Node
- A person, team, or organization in the network. Your roster defines the nodes; respondent attributes (role, department, tenure) become node data.
- Tie (or edge)
- A relationship between two nodes: advice, collaboration, learning, trust. Each tie is one row in your edge list, with its own attributes like frequency or strength.
- Centrality
- How connected someone is. People with high centrality are hubs: information reaches them fast, and losing them hurts. Several variants exist; degree centrality simply counts connections.
- Broker
- Someone who bridges otherwise disconnected groups. Brokers control how information moves between clusters, which makes them both valuable and overloaded.
- Cluster
- A group of nodes more densely connected to each other than to the rest of the network. Clusters show you where collaboration is strong and where silos have formed.
Questions, Answered
Frequently asked questions.
What exactly is in the export bundle?
Three structured CSV files: edges (one row per tie, tagged by tie type), nodes (one row per person, with roster attributes), and responses (columns derived from your question keys). Excel-friendly encoding, loading directly into R or Python (igraph, networkx, statnet), UCINET, or Gephi with no restructuring.
Does Ars Network draw the network maps for me?
Not today, and we would rather tell you that plainly than imply otherwise. Ars Network's job is the hardest part of the pipeline: collecting relational data correctly and delivering it analysis-ready. Visualization and metrics happen in your analysis tool of choice, where you have full control. Hosted analysis and visualization are on our roadmap.
Why can't I just use the survey platform my organization already pays for?
You can collect network-ish answers in any survey tool. What you cannot get out of one is network data. Generic platforms store one row per respondent; network analysis needs one row per relationship, plus rosters, name generators, and per-person follow-up loops. Teams that go that route typically spend 30 to 40 hours per study restructuring exports before analysis can start, and the instrument design is entirely on them.
How does this work with IRB oversight?
Three ways, all structural. Your institutional review board's approved consent text appears verbatim, because the platform never supplies its own. We cannot view your study workspace contents; support access exists only through a time-limited grant you issue and revoke. And respondent-facing surveys meet WCAG AA accessibility standards by default, which an increasing number of protocols require.
What happens if my study hits its respondent cap?
Collection pauses, gracefully, and we never bill per response. Every response already submitted stays safe and exportable, and you can raise the cap and resume. Your data is never locked behind an upgrade prompt and never discarded. We consider that a non-negotiable property of a research tool.
Do I need to be a network scientist to use this?
You need someone who will analyze the data: a researcher, a data-capable analyst, or a consultant. What you do not need is a specialist to design the instrument or wrangle the data, because validated templates handle the first and the export bundle eliminates the second. If you want help interpreting results, ask us about analysis partners.
Can I use it for regular, non-network survey questions too?
Yes. Standard question types (multiple choice, rating scales, open text, demographics) are built in, because real network instruments need them. Every attribute answer becomes a column in your node table, already joined to the network data, so one study can carry your climate or adoption questions alongside the network questions: one instrument, one consent flow, one export.
If a study has no network questions at all, a generic survey tool will serve you fine, and we will tell you so. The reason to run it here is when you want both kinds of data, joined.
Is there academic pricing?
Yes. Grant-funded and university research is why this platform exists, and academic licenses are available for individual researchers, labs, funded studies, and university sites. We are finalizing pricing with our first cohort, so ask. We also provide letters of support for grant proposals that include Ars Network in their data collection plan.
Get Started
Have a network study this quarter?
We are onboarding a small group of first customers. If you are planning AI adoption mapping, an advice network study, or any organizational network project, we would like to talk.
Request a DemoPrefer email? jmcclure@arsinnovate.com