The Org Chart Is 170 Years Old (and Showing It)
In 1855, Daniel McCallum drew the first formal organizational chart. He was the general superintendent of the Erie Railroad, working with civil engineer George Holt Henshaw. The Erie was one of the longest railroads in the world at the time, and McCallum needed a way to make sense of who reported to whom across hundreds of miles of track and thousands of workers. The diagram he produced was lost for over a century. Researchers Charles D. Wrege and Guidon Sorbo Jr. located the original at the Library of Congress in 2005.
What McCallum drew baked in three industrial-era assumptions. First, that hierarchy is stable, with the same boxes pointing to the same boxes for years at a stretch. Second, that reporting lines describe the actual flow of work. Third, that employees stay in one role long enough for the diagram to remain accurate.
None of those assumptions hold in 2026.
The Modern Org Doesn't Match Its Chart
Start with the multi-team data. A 2018 Harvard Business Review piece by Mark Mortensen, Bradley Kirkman, Gilad Chen, and John Mathieu reported that between 81 and 95 percent of employees globally serve on multiple teams simultaneously. Mortensen and Heidi Gardner had already shown that overlapping team membership is now the default, with knock-on costs in coordination and burnout. Gallup data suggests 84 percent of US employees are matrixed to some extent.
Now layer in the squad and tribe models. ING's 2015 agile transformation reorganized roughly 3,500 headquarters staff into around 350 nine-person squads inside 13 tribes. The Spotify model, described in Henrik Kniberg and Anders Ivarsson's 2012 whitepaper, defined squads as 6 to 12 people with end-to-end ownership and tribes as 40 to 150 people. Kniberg has since cautioned that the model was a snapshot, and the structure kept evolving. McKinsey research found that 80 percent of respondents in agile units reported moderate or significant performance gains after their transformations, and that agile transformations are three times more likely to land in the top quartile of performers.
An org chart describes none of this. It cannot show that a senior engineer is on three squads this quarter. It cannot show that a marketing lead is informally running incident response for a different tribe. It cannot show roster fluidity at all.
The Chorus of 2025 and 2026
The pile-on has been building. In Fortune on March 31, 2026, Aneesh Raman, LinkedIn's Chief Economic Opportunity Officer and co-author of Open to Work: How to Get Ahead in the Age of AI, put it bluntly: "The org chart was built in the industrial age to bring order, predictability, and stability to rapidly growing organizations. Companies need to let that go, as it's going to hold back innovation."
Keith Rosen made an even sharper version of the case in February 2026: "When structure prioritizes reporting lines over the flow of work, leadership effectiveness degrades." Org charts, in his framing, "structurally create the silos leaders want to eliminate."
The argument is moving from the edges of strategy writing into the mainstream of operational discourse.
AI Is Flattening the Hierarchy in Real Time
The structural change is happening on calendar dates you can point at.
On October 28, 2025, Amazon announced roughly 14,000 corporate cuts. CEO Andy Jassy had set a goal in September 2024 to increase the IC-to-manager ratio by 15 percent by Q1 2025, with explicit references to AI as the reason fewer layers were needed. On April 23, 2026, Microsoft offered voluntary buyouts to roughly 7 percent of its US workforce, around 8,750 employees. It was the first voluntary retirement program in the company's 51-year history. The same day, Meta announced 8,000 cuts, 10 percent of its workforce, with cuts beginning May 20, 2026, framed as a push deeper into AI.
Meanwhile, the leverage equation shifted. Cursor, the AI coding tool built by Anysphere, reportedly hit over $1B ARR with fewer than 300 employees, valued at $29.3B as of November 2025. Gartner predicted in October 2024 that through 2026, 20 percent of organizations will use AI to flatten their structure, eliminating more than half of current middle management positions.
And the people who would have filled those middle slots are walking away from them. A 2025 Robert Walters survey found that 52 percent of Gen Z workers said they would prefer not to be middle managers. The trend has a name now: conscious unbossing.
The historical lesson on what comes next is worth remembering. Zappos adopted holacracy in 2014, offered buyouts in March 2015, and 18 percent of employees took them. By March 2017, the company had walked back to a "circular hierarchy". Removing the chart without replacing it with something better produced more confusion. The lesson: pulling out the chart leaves a vacuum. Something has to fill it.
What a Dynamic Team Intelligence Map Shows
The map that fills that vacuum captures what an org chart structurally cannot.
- Behavior under pressure. How does a team respond when a deadline collapses? Who steps up, who freezes, who escalates, who delegates?
- Decision velocity. How long does it take from problem identification to commitment? Hours, days, or weeks?
- Recovery from setbacks. After a missed launch or a failed quarter, how fast does the team get back to baseline performance? Does it ever?
- Cross-functional throughput. Where does work actually move across functions? Which informal pairings carry disproportionate load?
- Roster fluidity. Which people are on which teams this week, this sprint, this campaign? Where is the latent multi-team load that the chart hides?
The org chart cannot answer any of those questions. The Team Intelligence map is built around them.
The ONA Precursor
Organizational network analysis was the first serious attempt to make this kind of map. Deloitte's 2017 Global Human Capital Trends report found that "only 8 percent of companies in this year's survey are using ONA today, with an additional 48 percent of companies experimenting." That was nearly a decade ago. The thesis was correct. The tooling was early.
Rob Cross has been building the empirical case ever since. Across more than 20 years of research at over 300 organizations, he has documented a recurring pattern: 3 to 5 percent of people in a typical network account for 20 to 35 percent of value-adding collaborations. In one professional services case, only 5 percent of people accounted for 25 percent of all revenue-producing collaborations. After ONA-driven interventions, the firm saw a 30 percent increase in proposals up to $1M and roughly 10 percent annualized revenue growth. Cross has also shown that collaborative work consumes 85 percent of employees' time, and that top performers are 18 to 24 percent more efficient than peers at managing that load.
The structural argument runs even deeper than collaboration data. Conway's Law, articulated in Datamation in April 1968, observed that "organizations which design systems are constrained to produce designs which are copies of the communication structures of these organizations." Carliss Baldwin and Lyra Colfer's 2016 review of the mirroring hypothesis surveyed 142 studies and found empirical support in over 80 percent of cases. Your org structure shapes your products. If your structure is mismeasured, your products are too.
Today's commercial ONA tools have made parts of this legible. Polinode visualizes networks of up to 50,000 nodes and 250,000 edges. Worklytics pulls more than 400 metrics from over 25 sources. Microsoft Viva Insights ONA Change Management, generally available since 2024, surfaces cross-group collaboration, insular collaboration, and brokers of information flow.
What's Missing in Today's Map Tools
Every ONA tool on the market is, in effect, a collaboration heatmap built from email, calendar, and chat metadata. They show frequency, density, and broker positions. They are good at it.
They do not measure behavior under pressure. They cannot tell you that the team responded to a P0 incident with three ICs talking past each other for forty minutes before someone finally called a decision. They do not measure decision velocity in any meaningful way, because calendar metadata does not capture the moment of commitment. They do not measure recovery, because there is no signal in Outlook for "the team got demoralized after the launch slipped."
That layer, the behavioral layer, is where the Team Intelligence map has to add value or it is just another dashboard.
Counter-Arguments Worth Taking Seriously
Four objections come up when this idea hits a real organization.
"We need an org chart for compliance, payroll, and security." Correct. The org chart is an administrative artifact. It can keep doing that job. The Team Intelligence map handles operational reality. Both can coexist. Companies already run an HRIS, a data warehouse, and a BI layer side by side without crisis.
"You can't have two sources of truth." Companies already run multiple coexisting layers. The org chart is the system of record for legal entities, reporting lines, and titles. The Team Intelligence map is the analytical and operational layer. They serve different audiences and answer different questions.
"Dynamic mapping just adds chaos." Chaos is already there. With 84 percent of US employees matrixed and 81 to 95 percent on multiple teams at once, the chaos is the status quo. The map makes existing fluidity legible so leaders can manage it instead of pretending it does not exist.
"This is surveillance." This is the load-bearing concern. Done badly, dynamic team mapping becomes individual surveillance and breaks trust on contact. Done well, leaders see aggregate team trends and strengths-based individual highlights. Voluntary participation, transparent data collection, and strengths-only individual reporting are the difference between a tool people trust and a tool people work around. The QuestWorks team has been explicit about this from day one: the map exists to make teams better.
How Teams Actually Get Mapped
Edmondson's research on "teaming" anticipated where this is headed. In her April 2012 HBR piece "Teamwork on the Fly", she wrote: "Teaming captures the fact that more people are finding themselves having to collaborate across boundaries without the luxury of a stable team structure." Her examples were the Beijing Water Cube and the 2010 Chilean miner rescue. Both were temporary configurations of strangers who had to cohere in pressure conditions.
That description now fits everyday work across most knowledge organizations. Which means the map has to capture how that teaming actually goes inside the moment of work, beyond the metadata trail of who emailed whom.
This is where the QuestWorks Team Intelligence Engine comes in. The platform runs cinematic, voice-controlled multiplayer sessions where teams take on quests together for 25 minutes at a time, 2 to 5 players per group. Players surface as one of nine public HeroTypes, each with a strengths profile teammates and managers can see. Inside each session, behavior under pressure is observable directly: who delegates, who absorbs ambiguity, who escalates well, who recovers from setbacks. Leaders see aggregate team trends in the Weekly Team Health Report and individual strengths-based highlights in the QuestDash. HeroGPT delivers private coaching via Slack and never shares those conversations upstream. Participation is voluntary. Pricing is $20 per user per month with a 14-day free trial.
Slack is the integration layer. The game itself runs on its own platform.
The Map Is the Territory Now
The org chart will not disappear in 2026. It will keep showing up in HR systems, board decks, and onboarding slides. What will change is its status. It will go from being the primary representation of the company to being one administrative artifact among several, mostly useful for legal and payroll purposes.
The thing leaders actually look at to make decisions, the thing investors will eventually ask about in due diligence, the thing that earns the title of "how this company really works," will be the Team Intelligence map. It will be dynamic. It will update as rosters update. It will surface behavior under pressure, decision velocity, recovery curves, and cross-functional throughput. It will not pretend that hierarchy is stable when it is not.
The flat-versus-hierarchical debate has been beside the point for years. Both shapes answer the same wrong question: where do the boxes go? The right question for the next decade is operational. How does this team actually work, and how do we know?
That question has an answer. The answer is a map.
Try it free
