Map-Territory Distinction

Thinking ToolsAny stepAlso known as The Map Is Not the Territory

Originated by Alfred Korzybski (Science and Sanity, 1933) and extended by Gregory Bateson (Steps to an Ecology of Mind, 1972).

Treats every model, dashboard, and plan as a partial picture of reality, and asks where the picture and the ground have drifted apart.

Nine facts from a quarterly review, sorted by whether the dashboard shows them and whether they change the call

The dashboard says three things: the Northeast is green, the product is healthy, support is at capacity. The plans are to raise the Northeast's target, cut the product's budget, and hire four agents. The headline number is placed as an example. Select each fact found on the ground and pick its box: the top-left box is where the map and the territory have drifted apart. Illustrative.

Items (select one, then a box)
  • Not placedPlacedDrift: go and lookOn the map, and decisiveTracked, not decisiveDetail the map can drop
  • Not placedPlacedDrift: go and lookOn the map, and decisiveTracked, not decisiveDetail the map can drop
  • Not placedPlacedDrift: go and lookOn the map, and decisiveTracked, not decisiveDetail the map can drop
  • Not placedPlacedDrift: go and lookOn the map, and decisiveTracked, not decisiveDetail the map can drop
  • Not placedPlacedDrift: go and lookOn the map, and decisiveTracked, not decisiveDetail the map can drop
  • Not placedPlacedDrift: go and lookOn the map, and decisiveTracked, not decisiveDetail the map can drop
  • Not placedPlacedDrift: go and lookOn the map, and decisiveTracked, not decisiveDetail the map can drop
  • Not placedPlacedDrift: go and lookOn the map, and decisiveTracked, not decisiveDetail the map can drop
  • Not placedPlacedDrift: go and lookOn the map, and decisiveTracked, not decisiveDetail the map can drop

Would it change the decision?

YesNo
Drift: go and look
On the map, and decisive
Detail the map can drop
Tracked, not decisive
Not on itOn it

On the dashboard?

Northeast at 104%. Northeast bookings are 104 percent of plan, so its tile is green.Not placed yet.Northeast at 104%. Northeast bookings are 104 percent of plan, so its tile is green.Drift: go and look: Off the map, and it changes the call. This is what the dashboard hid; the decision waits until someone has seen it.Northeast at 104%. Northeast bookings are 104 percent of plan, so its tile is green.On the map, and decisive: The dashboard carries it and the decision rests on it. Check it against the facts in the top-left box before acting.The map's headline, and the reason for the plan to raise the region's target.Northeast at 104%. Northeast bookings are 104 percent of plan, so its tile is green.Tracked, not decisive: On the dashboard but it would not move the decision. A candidate to drop from the review.Northeast at 104%. Northeast bookings are 104 percent of plan, so its tile is green.Detail the map can drop: Off the map and rightly so. Every map leaves things out; this is the kind it should.Two clients, 60%. A ride-along shows two clients account for 60 percent of the Northeast's bookings, and one is reviewing vendors.Not placed yet.Two clients, 60%. A ride-along shows two clients account for 60 percent of the Northeast's bookings, and one is reviewing vendors.Drift: go and look: Off the map, and it changes the call. This is what the dashboard hid; the decision waits until someone has seen it.Invisible on the tile, and it turns a stretch target into a concentration risk.Two clients, 60%. A ride-along shows two clients account for 60 percent of the Northeast's bookings, and one is reviewing vendors.On the map, and decisive: The dashboard carries it and the decision rests on it. Check it against the facts in the top-left box before acting.Two clients, 60%. A ride-along shows two clients account for 60 percent of the Northeast's bookings, and one is reviewing vendors.Tracked, not decisive: On the dashboard but it would not move the decision. A candidate to drop from the review.Two clients, 60%. A ride-along shows two clients account for 60 percent of the Northeast's bookings, and one is reviewing vendors.Detail the map can drop: Off the map and rightly so. Every map leaves things out; this is the kind it should.Discounts at 22%. Deal files show the Northeast's average discount rose from 12 to 22 percent this quarter.Not placed yet.Discounts at 22%. Deal files show the Northeast's average discount rose from 12 to 22 percent this quarter.Drift: go and look: Off the map, and it changes the call. This is what the dashboard hid; the decision waits until someone has seen it.Bookings are on the map; margin is not. The region bought its green.Discounts at 22%. Deal files show the Northeast's average discount rose from 12 to 22 percent this quarter.On the map, and decisive: The dashboard carries it and the decision rests on it. Check it against the facts in the top-left box before acting.Discounts at 22%. Deal files show the Northeast's average discount rose from 12 to 22 percent this quarter.Tracked, not decisive: On the dashboard but it would not move the decision. A candidate to drop from the review.Discounts at 22%. Deal files show the Northeast's average discount rose from 12 to 22 percent this quarter.Detail the map can drop: Off the map and rightly so. Every map leaves things out; this is the kind it should.Weekly users up 8%. Weekly active users of the product rose 8 percent, so its tile reads healthy.Not placed yet.Weekly users up 8%. Weekly active users of the product rose 8 percent, so its tile reads healthy.Drift: go and look: Off the map, and it changes the call. This is what the dashboard hid; the decision waits until someone has seen it.Weekly users up 8%. Weekly active users of the product rose 8 percent, so its tile reads healthy.On the map, and decisive: The dashboard carries it and the decision rests on it. Check it against the facts in the top-left box before acting.What the map says about the product. Read it with the quiet annual accounts beside it.Weekly users up 8%. Weekly active users of the product rose 8 percent, so its tile reads healthy.Tracked, not decisive: On the dashboard but it would not move the decision. A candidate to drop from the review.Weekly users up 8%. Weekly active users of the product rose 8 percent, so its tile reads healthy.Detail the map can drop: Off the map and rightly so. Every map leaves things out; this is the kind it should.Annual users quiet. Ten calls to customers show that most annual-contract accounts stopped logging in two months ago; new monthly signups carry the user count.Not placed yet.Annual users quiet. Ten calls to customers show that most annual-contract accounts stopped logging in two months ago; new monthly signups carry the user count.Drift: go and look: Off the map, and it changes the call. This is what the dashboard hid; the decision waits until someone has seen it.Q3 renewals are at risk. Cutting the budget now would be the wrong call.Annual users quiet. Ten calls to customers show that most annual-contract accounts stopped logging in two months ago; new monthly signups carry the user count.On the map, and decisive: The dashboard carries it and the decision rests on it. Check it against the facts in the top-left box before acting.Annual users quiet. Ten calls to customers show that most annual-contract accounts stopped logging in two months ago; new monthly signups carry the user count.Tracked, not decisive: On the dashboard but it would not move the decision. A candidate to drop from the review.Annual users quiet. Ten calls to customers show that most annual-contract accounts stopped logging in two months ago; new monthly signups carry the user count.Detail the map can drop: Off the map and rightly so. Every map leaves things out; this is the kind it should.Support 98% busy. Support agents are busy 98 percent of their logged hours.Not placed yet.Support 98% busy. Support agents are busy 98 percent of their logged hours.Drift: go and look: Off the map, and it changes the call. This is what the dashboard hid; the decision waits until someone has seen it.Support 98% busy. Support agents are busy 98 percent of their logged hours.On the map, and decisive: The dashboard carries it and the decision rests on it. Check it against the facts in the top-left box before acting.The reason for the plan to hire four agents.Support 98% busy. Support agents are busy 98 percent of their logged hours.Tracked, not decisive: On the dashboard but it would not move the decision. A candidate to drop from the review.Support 98% busy. Support agents are busy 98 percent of their logged hours.Detail the map can drop: Off the map and rightly so. Every map leaves things out; this is the kind it should.A third are repeats. Sitting with the team shows a third of tickets are repeat calls about one billing bug.Not placed yet.A third are repeats. Sitting with the team shows a third of tickets are repeat calls about one billing bug.Drift: go and look: Off the map, and it changes the call. This is what the dashboard hid; the decision waits until someone has seen it.Fix the bug and the team has room. The hire is not needed yet.A third are repeats. Sitting with the team shows a third of tickets are repeat calls about one billing bug.On the map, and decisive: The dashboard carries it and the decision rests on it. Check it against the facts in the top-left box before acting.A third are repeats. Sitting with the team shows a third of tickets are repeat calls about one billing bug.Tracked, not decisive: On the dashboard but it would not move the decision. A candidate to drop from the review.A third are repeats. Sitting with the team shows a third of tickets are repeat calls about one billing bug.Detail the map can drop: Off the map and rightly so. Every map leaves things out; this is the kind it should.Calls per rep. The dashboard tracks calls logged per sales rep each week.Not placed yet.Calls per rep. The dashboard tracks calls logged per sales rep each week.Drift: go and look: Off the map, and it changes the call. This is what the dashboard hid; the decision waits until someone has seen it.Calls per rep. The dashboard tracks calls logged per sales rep each week.On the map, and decisive: The dashboard carries it and the decision rests on it. Check it against the facts in the top-left box before acting.Calls per rep. The dashboard tracks calls logged per sales rep each week.Tracked, not decisive: On the dashboard but it would not move the decision. A candidate to drop from the review.Watched every week, and it would not change any of the three decisions.Calls per rep. The dashboard tracks calls logged per sales rep each week.Detail the map can drop: Off the map and rightly so. Every map leaves things out; this is the kind it should.Office moved floors. The Northeast office moved to a different floor in its building.Not placed yet.Office moved floors. The Northeast office moved to a different floor in its building.Drift: go and look: Off the map, and it changes the call. This is what the dashboard hid; the decision waits until someone has seen it.Office moved floors. The Northeast office moved to a different floor in its building.On the map, and decisive: The dashboard carries it and the decision rests on it. Check it against the facts in the top-left box before acting.Office moved floors. The Northeast office moved to a different floor in its building.Tracked, not decisive: On the dashboard but it would not move the decision. A candidate to drop from the review.Office moved floors. The Northeast office moved to a different floor in its building.Detail the map can drop: Off the map and rightly so. Every map leaves things out; this is the kind it should.True, and rightly left off. A map that showed everything would be as large as the ground.

Your sort matched 0 of 9.Your sort matched 1 of 9.Your sort matched 2 of 9.Your sort matched 3 of 9.Your sort matched 4 of 9.Your sort matched 5 of 9.Your sort matched 6 of 9.Your sort matched 7 of 9.Your sort matched 8 of 9.Your sort matched 9 of 9.

Drift: go and look
Off the map, and it changes the call. This is what the dashboard hid; the decision waits until someone has seen it.
On the map, and decisive
The dashboard carries it and the decision rests on it. Check it against the facts in the top-left box before acting.
Tracked, not decisive
On the dashboard but it would not move the decision. A candidate to drop from the review.
Detail the map can drop
Off the map and rightly so. Every map leaves things out; this is the kind it should.

The sort

  • Northeast at 104%: On the map, and decisive. The map's headline, and the reason for the plan to raise the region's target.
  • Two clients, 60%: Drift: go and look. Invisible on the tile, and it turns a stretch target into a concentration risk.
  • Discounts at 22%: Drift: go and look. Bookings are on the map; margin is not. The region bought its green.
  • Weekly users up 8%: On the map, and decisive. What the map says about the product. Read it with the quiet annual accounts beside it.
  • Annual users quiet: Drift: go and look. Q3 renewals are at risk. Cutting the budget now would be the wrong call.
  • Support 98% busy: On the map, and decisive. The reason for the plan to hire four agents.
  • A third are repeats: Drift: go and look. Fix the bug and the team has room. The hire is not needed yet.
  • Calls per rep: Tracked, not decisive. Watched every week, and it would not change any of the three decisions.
  • Office moved floors: Detail the map can drop. True, and rightly left off. A map that showed everything would be as large as the ground.

An example

A software company's quarterly review ran on one dashboard, and the dashboard said three things. The Northeast region was green, at 104 percent of plan. The flagship product was healthy, with weekly users up 8 percent. Support was at capacity, busy 98 percent of logged hours. The draft decisions followed: raise the Northeast's target, move budget away from the healthy product, hire four support agents.

The chief operating officer delayed the calls by a week and sent people to the ground. A ride-along in the Northeast found that two clients made up 60 percent of its bookings and that discounts had risen from 12 to 22 percent. Ten customer calls found that annual-contract accounts had gone quiet two months earlier, while new monthly signups carried the user count. A day with support found a third of tickets were repeat calls about one billing bug. The target held, discount approval moved to the regional vice president, renewal outreach began, and the bug was fixed before anyone was hired.

Why it works

A dashboard is a map: a small picture of a large territory, drawn by deciding what to leave out. That is what makes it useful, and it is also how it misleads. The things left out keep happening, and over time the picture and the ground drift apart. A team that manages the picture keeps acting on the drawing long after the ground has changed.

The discipline is to treat every model, plan, and report as a map and to ask two questions of it before an important decision. What does it leave out? And would any of the omissions change this call? Most omissions do not matter, which is why the map exists. The few that do are found only by going to the territory: the deal files, the customer, the shop floor.

The thinking behind it

Alfred Korzybski Alfred Korzybski, a Polish American scholar, founded general semantics. He presented the idea in a 1931 paper to the American Association for the Advancement of Science and developed it in Science and Sanity (1933). His dictum, the map is not the territory, holds that words and models are abstractions from reality and must not be mistaken for it. Any map leaves things out; trouble starts when people act as if it did not.

Gregory Bateson Gregory Bateson, an anthropologist and cyberneticist, carried Korzybski's distinction into his account of mind and communication. In Steps to an Ecology of Mind (1972) he argued that what we know of a territory reaches us only through the channels by which we perceive it. The mental world is a series of maps, and the territory never gets in at all. For a manager, every report is one of those channels.

How to run it

  1. Name the map. Identify the dashboard, model, or plan the decision rests on, and who drew it.
  2. Write what it says. State its claims in one line each, such as the region is green.
  3. List what it leaves out. For each claim, note what the measure cannot see, such as margin behind bookings.
  4. Ask which omissions could change the call. Keep only the gaps that would flip or resize the decision on the table.
  5. Go to the territory. Send someone to look at each of those gaps directly: deal files, customers, the floor.
  6. Decide on what you found. Make the call on the combined picture, and say which facts came from outside the map.
  7. Redraw the map. Add a measure for any gap that keeps mattering, and drop measures that never change a decision.

Where it goes wrong

The first failure is distrusting every map. A team that insists on seeing everything for itself never decides. Go to the ground only where an omission could change the call.

The second is adding every gap to the dashboard. Each new tile makes the map larger and harder to read, until no one reads it. Add a measure only when the same gap has mattered more than once.

The third is sending someone to the ground who will report the map back. A manager whose bonus rides on the green tile will find it green. Send someone with no stake in the picture, and ask for what they saw before what they concluded.

When to use something else

  • Goodhart-Aware Metric Selection. When the gap is people optimizing the measure itself, and the job is choosing measures that resist it.
  • Construct Validity. When the question is whether a measure captures the thing it claims to measure at all.
  • Thick Description. When going to the ground means recording what people do and what it means to them.

Where it sits in SOLVE eX

  • Any step: Tools that serve every step: habits of thought and checks on the thinking itself.

Go to the source

  • Science and Sanity, 1933
  • Steps to an Ecology of Mind, 1972

Back to the catalog