Every company has a sentence that quietly redraws its org chart: "Ask the person everyone asks."
That person may be a manager. They may also be an analyst 4 layers down, an assistant who remembers how every decision was made, or an engineer who can explain the part of the system everyone else approaches with professional caution.
The official chart says where authority is meant to sit. The unofficial one says where people go for help, context, candor, and permission they trust.
Both are real.
Managers get into trouble when they treat this as a contest between formal structure and the supposedly authentic network underneath it. The formal organization is not fake. It controls reporting relationships, budgets, hiring, evaluation, and escalation. The informal organization is not automatically wise. It can spread knowledge and solve problems, or hoard information, exclude outsiders, and make one helpful person the human API for an entire company.
The job is to see where the 2 structures reinforce each other, where they compensate for each other, and where the gap has become dangerous.
What the evidence says
- Informal networks carry advice, trust, and work-related communication across formal reporting lines.
- David Krackhardt and Jeffrey Hanson described ways to map those 3 networks in a 1993 Harvard Business Review article.
- Social network analysis can reveal connectors, bottlenecks, isolated groups, and missing relationships that a formal chart cannot show.
- A network map is evidence about relationships, not a score of human value. The question asked and the data collected determine what it can reveal.
- Peer-reviewed ethics research has warned that network analysis creates special risks in managerial settings. Purpose, consent, access, aggregation, and how results affect employees all matter.
The company behind the chart
David Krackhardt and Jeffrey Hanson gave the idea a durable name in their 1993 article "Informal Networks: The Company Behind the Chart". They argued that much of an organization's real work travels through relationships that cross functions and divisions.
They separated 3 networks that are often bundled together:
The advice network shows who people approach for technical help, judgment, or information needed to get work done.
The trust network shows where people take sensitive concerns, political questions, or doubts they are not ready to announce widely.
The communication network shows who regularly discusses work with whom.
Those networks overlap, but not perfectly. The person who knows the answer may not be the person people trust with uncertainty. The person who appears in the most meetings may not be the person whose advice changes the decision. A manager can occupy the center of one network and the edge of another.
That distinction is more useful than a single list of "influencers." Influence is not a substance people carry around at a fixed level. It depends on the question. Someone can be central to technical advice, peripheral to social trust, and irrelevant to a decision about hiring.
The formal chart remains necessary because these relationships do not answer every organizational question. An advice network cannot tell you who conducts a performance review. A trust network should not determine who is legally accountable for a regulated decision. The chart defines stable formal relationships; the networks reveal how people navigate the work inside and across them.
Why the informal organization appears
No formal design can anticipate every conversation a company will need.
Work changes faster than reporting lines. Projects cross functions. An experienced employee remembers why a strange exception exists. People discover who responds quickly, who gives useful context, and who can disagree without turning a question into a career event.
Informal networks grow from repeated experience. They are the organization's memory of which relationships work.
This can be excellent. A strong connector carries information across a boundary that would otherwise slow a customer decision. A trusted expert helps a new team avoid a mistake the old team already paid for. Colleagues form relationships that let them coordinate without escalating ordinary issues through 3 managers.
It can also conceal structural debt.
If every complex request depends on one veteran employee, the organization has not necessarily discovered a star. It may have failed to distribute knowledge. If 2 departments cooperate only because their longest-serving managers are friends, the relationship is an asset with an expiry date. If employees routinely bypass a formal leader to get a safe answer, the network may be routing around a problem the chart politely ignores.
An informal workaround becomes dangerous when the company mistakes its success for proof that the system is healthy. The work still gets done, so nobody notices that it requires private favors, historical memory, or one person's permanent availability.
That is how helpful people become bottlenecks by being helpful.
What organizational network analysis can show
Organizational network analysis, often shortened to ONA, applies social network methods to relationships at work. Instead of starting with the hierarchy, it starts with a defined question about connection.
Researchers and practitioners may ask employees whom they rely on for expertise, who helps them solve problems, or where they discuss sensitive issues. They may also study records of communication or collaboration. The result is a network of people or groups connected by a specific kind of relationship.
Rob Cross, Stephen Borgatti, and Andrew Parker described how organizations could use this approach to support collaboration in their 2002 article "Making Invisible Work Visible". The important word is not visible. It is work. A useful analysis begins with an organizational problem, not the opportunity to make an impressive web of dots.
Depending on the question, a map may reveal:
- people who connect otherwise separate groups;
- experts who receive a disproportionate share of requests;
- teams with strong internal ties and weak outside contact;
- formal leaders who are missing from a relationship their role requires;
- new employees or remote groups that remain isolated;
- duplicated channels carrying the same information;
- a dependency on one person that becomes obvious only when they take leave.
These are patterns to investigate, not verdicts.
A highly central person may be a generous connector. They may also be overwhelmed, controlling access, or simply copied on too many messages. A quiet employee may be isolated, or their work may require long stretches of independent concentration. The network cannot decide which interpretation is true.
The map is only as good as its boundary. Study one department and you may label its most outward-looking employee peripheral because the people they connect with were excluded. Ask who gives technical advice and you will not discover who provides emotional support. Count messages and you will mostly learn who sends messages, which is not the same as whose judgment matters.
Network analysis turns a vague suspicion into a better question. It does not remove the need to ask people what the pattern means.
The privacy problem hiding in "digital exhaust"
Modern work tools record relationships as a side effect. Email headers, chat activity, calendar invitations, document collaboration, and ticket histories can all be converted into network data. That makes informal structure more observable than it was in 1993.
It also creates an obvious temptation: if the data exists, analyze it first and explain later.
That is a poor standard for studying human relationships.
Communication metadata can be sensitive even when nobody reads message content. It can reveal who contacts HR, which groups are isolated, who communicates outside normal hours, and how relationships change during a restructuring. Employees may reasonably experience an undisclosed analysis as surveillance rather than organizational learning.
This concern is not an afterthought attached to newer collaboration tools. Stephen Borgatti and José Luis Molina's 2003 paper on ethics in organizational network analysis argued that network data introduces special problems in managerial settings because the analysis concerns relationships, not only consenting individuals, and because the organization may use the results to make decisions about employees. They also warned that managerial use could change what people are willing to report, weakening the data itself.
The practical lesson is sharper than "be transparent." A company must consider who commissioned the analysis, which question the data can fairly answer, who may be exposed through another person's response, and what decision the map will be allowed to influence. A network diagram gains no moral authority from looking mathematical.
A defensible ONA project should answer basic questions before touching the data:
- What work problem are we trying to understand?
- Why is network data necessary to answer it?
- What is the minimum data required?
- Who knows the analysis is happening?
- Who can see individual-level results?
- How will the organization prevent the map from becoming a hidden performance rating?
- When will the underlying data be deleted?
Aggregation helps. So does asking employees directly, sharing the purpose, limiting access, and reporting patterns at the team level when individual names are unnecessary.
The phrase "digital exhaust" makes relationship data sound like waste heat from a machine. It is still information about people. Treat it accordingly.
When the 2 org charts disagree
Some disagreement is normal and healthy.
A formal structure is deliberately stable. Informal relationships adapt quickly. If every question had to follow the reporting line, the hierarchy would become a telephone tree with payroll.
The useful signal is not that a cross-functional connection exists. It is what the organization depends on that connection to do.
Consider 4 kinds of gap:
Productive bypass. Teams have enough trust and shared context to solve routine problems directly. Managers remain informed where needed, but they do not insist on relaying every sentence.
Compensating relationship. One person bridges a structural gap the organization has not addressed. The arrangement works, but it is fragile and may overload the connector.
Shadow authority. People seek permission from someone who has no formal accountability for the decision. This may reflect valuable expertise or a leader whose informal power is greater than the chart admits.
Excluded group. A team, location, or set of employees sits outside the advice and trust networks required for its work. The formal chart includes them. The operating system does not.
Each needs a different response. Do not formalize a healthy direct relationship just because it bypasses a manager. Do not celebrate a heroic connector when the job is consuming them. Do not redraw the chart merely to acknowledge every person whose advice matters.
Ask what should remain a relationship and what needs a system.
A practical way to map the informal organization
You do not need network-analysis software to begin. Take one piece of work that repeatedly stalls, surprises people, or requires rescue. Keep the scope narrow enough to discuss honestly.
Ask the people involved 5 questions:
- Who do you go to when the documented process does not answer the question?
- Whose advice do you usually act on?
- Who needs to be involved earlier than the formal workflow suggests?
- Where does information routinely stop or arrive too late?
- Whose absence would make this work unexpectedly difficult?
Compare the answers with the formal structure and the ownership agreements beside it.
If one name appears everywhere, do not immediately promote, reward, or remove that person from the flow. First understand what people are seeking from them. It may be expertise that needs documenting, trust that took years to earn, decision authority that belongs elsewhere, or simple responsiveness other channels have failed to provide.
Then reduce the unhealthy dependency without destroying the relationship. Pair the expert with others. Create office hours. Clarify decision rights. Build direct ties between groups that currently communicate through one broker. Make the formal owner easier to reach and safer to approach.
Repeat the exercise after a meaningful change. Networks move. A map becomes nostalgia faster than an org chart does.
Reorganizations move boxes faster than relationships
This is where the second chart becomes especially important.
A reorganization can change a reporting line in an afternoon. It cannot instantly recreate the advice, trust, and communication relationships that made the old arrangement function. Some relationships will cross the new boundaries and help the change. Others will weaken because people no longer share work, time, or context.
That does not make reorganization futile. It makes relationship continuity part of the implementation. Leaders need to know which connections should be preserved, which dependencies should be reduced, and which new ties must form before the new chart can work as designed.
Ignore that layer and the organization may spend months rediscovering who knows what. This is one reason reorgs often fix less than leaders expect.
The official org chart answers a set of necessary questions. Who manages whom? Where does a role sit? Which formal path carries accountability?
The unofficial chart answers different ones. Who helps? Who knows? Who is trusted? Where does work actually cross the lines?
Good managers do not choose between them. They keep the formal structure clear enough to be useful and the informal network healthy enough to make it work.