Your Index is a reading from 0 to 100 of what your organization has attempted, carried and finished. Every point rests on something that happened, with a date on it. The dial reads 68, and that made-up reading is worked through below.
Each of the six areas is scored from 0 to 100 on what you can show. The six are then weighted into one number. Here is what each share is worth.
| Area | Weight | What its share buys |
|---|---|---|
| Ambition | 20% | What the organization sets out to do |
| Individual reach | 15% | How far one person carries it |
| Working knowledge | 15% | What can be handed over |
| Machine capacity | 15% | What actually does the work |
| Direction | 20% | Who chooses, and who answers |
| Compounding | 15% | Whether it gets cheaper each time |
Ambition and Direction carry a fifth each. They are the two ends of one decision: what to set out to do, and who will stand behind it. The four areas between them are how the work gets done, and those four are the easier ones to buy. Every competitor can rent the same machine workers on the same morning, at the same posted prices. Nobody can rent the nerve to pick a harder problem, or a person willing to put a name against it.
The weights are ours. No standards body has certified them, and no study proves this exact split. Here is what they were set against, so you can argue with the reasoning and not only the numbers. The middle four are the buyable ones, so they carry less. The two on the ends carry more for one reason. The published work on why AI programs stall keeps returning organizational answers rather than technical ones. In one study of sixty-five practitioners, four of the five root causes were about the organization and its people. That is the whole derivation, and it is a judgment read off the record rather than a measurement.
Then test how much the split matters before you spend a meeting on it. Move any single weight by five points on the worked reading below, taking the five from the other five areas in proportion. The total never moves by a full point. The largest move is 0.9, on Machine capacity. The order of your own six numbers is what changes the answer, not our arithmetic.
What to do. Take the six area names into your next management meeting. Ask which of them has ever held the room for a full hour.
Who: you, in a meeting you already hold. How long: ten minutes of the agenda. You end up with: the two areas nobody can remember discussing. Those are where your reading will be lowest, known before you answer a single question.
One rule sits above the arithmetic, and it is an accelerator when you read it the right way.
Direction covers who chooses the work and who answers for what it does. No other area may read more than twenty points above it.
The reason is practical rather than moral. Capability that nobody will answer for tends not to stay in use. That is our assumption and not a measured finding. It has a shape most leaders will recognize. A pilot runs for six weeks. Its sponsor moves to another team, and nobody renews it, because no one person was ever named as its owner.
The accelerator works like this. Picture an organization with a rehearsed change of supplier, a complete register, and machine workers running with real room to act. Put a name on each decision those machines make. The same board will then let those machine workers run wider: more of the business, higher value, less asking first.
The reading prints both figures. What an area earned sits beside what the rule allowed. The ceiling is never hidden, and you can see how many points are waiting for a name.
What to do. Find one process where a machine’s output moves money. Write down the name of the person who allowed it.
Who: you, with whoever owns that process. How long: one afternoon. You end up with: a name on one process, or the knowledge that Direction is your ceiling and how many points it holds back.
Each band names a state of capability. None of them is a warning, and none of them is a grade.
The work is done much the way it was two years ago. Tools are present, often bought one at a time and never recorded. Nothing anybody sets out to do has changed.
The existing work goes faster. People are quicker at their own tasks, and the handoffs between them sit where they always sat. You can point to time saved and to nothing new attempted.
Work that used to need a team has shipped, carried by one named person. Something has come off the declined list. Every process that moves money has a named person who can halt it, and one has.
The limit is what you dare to choose, not how many people you have. Each quarter’s work makes the next cheaper. Forward calls are scored, and a change of supplier has been performed and timed.
The bands describe what somebody walking around would find, not what an organization says about itself. Moving up a band is usually a year of work, which is the order the route is written in.
What to do. Read the band above yours and find the sentence in it furthest from true today. That sentence is your next twelve months.
Who: you, then the person who owns amplification. How long: fifteen minutes. You end up with: one sentence, with one name against it, on next week’s agenda.
Every point rests on a completed act with a date on it. Six kinds of act raise a reading.
Five things score nothing at all. A license bought, or seats issued to everybody. A policy nobody has ever acted on. A second supplier named in a document and never tested. A stated ability to change supplier with no timed record behind it. A trial that produced a presentation and left the work exactly as it was.
What to do. Take last quarter’s list of achievements and strike every line carrying no date and no name.
Who: whoever wrote that list. How long: half an hour. You end up with: the dated, named half of it. That is roughly your reading, before you answer a question.
Two figures print beside every reading: the date, and how much of it was checked against documents.
Ask people how much faster they have become and the answer comes back generous by thirty-nine points. That is the measured gap in the randomized developer trial, set out in full with its design and its dates. It is why nothing here scores on how the year felt.
The same disagreement runs through whole economies. Three official surveys of the United States report AI adoption at 18 percent, 41 percent and 78 percent. They are not in conflict. They count different things. The 18 counts firms, from the U.S. Census Bureau’s business survey for November 2025 to January 2026. The 41 counts workers. The 78 counts the labor force at firms that have already adopted. A Federal Reserve staff note of 3 April 2026 sets the three side by side. It gives five reasons they diverge, one of which is the pressure senior leaders feel to report usage. Take a number like that into a board pack with its unit of measurement attached, or leave it out.
The checked figure is the one you can take into a meeting. A reading built on records will hold up when people disagree with it. A reading built on how the year felt will not.
What a checked reading buys is a price on your next move. On the worked reading below, twelve points of Direction are worth 3.6 points of Index. That is 2.4 at Direction’s own weight, and another 1.2 released from under the ceiling. It is what turns a number into an argument for spending: not what you are, but what one area is worth if you move it.
What to do. Take the reading twice. Once quickly, from memory, and once with the documents open beside you.
Who: you both times. How long: twenty minutes, then an afternoon. You end up with: two numbers and the distance between them. That distance is the most useful thing you will learn this quarter.
Three limits, stated once. The reading does not say whether your organization is safe. It does not forecast anything. And it cannot see whether the hours the machines handed back went to harder questions or to nothing much. That is why what you set out to do carries as much weight as anything else in the table.
One made-up organization is used across this whole site. The dial on the front page and the sample the check prints are this same company. Its arithmetic is not made up, and every line of it can be checked.
Read it from the bottom. Compounding at 55 says this quarter’s work is not yet making next quarter’s cheaper. Machine capacity earned 92 and prints 84, because Direction reads 64 and the ceiling allows twenty points above it. The outlined bar is what was earned and the solid bar is what was printed.
The total is 68, which lands in Amplified: work that used to need a team has shipped, and something has come off the declined list.
| Area | Earned, then printed | Weight | Points toward the total |
|---|---|---|---|
| Ambition | 74 | 0.20 | 14.80 |
| Individual reach | 62 | 0.15 | 9.30 |
| Working knowledge | 71 | 0.15 | 10.65 |
| Machine capacity | 92 → 84 | 0.15 | 12.60 |
| Direction | 64 | 0.20 | 12.80 |
| Compounding | 55 | 0.15 | 8.25 |
| Your Index | — | 1.00 | 68.40, rounded to 68 · Amplified · the ceiling holds one area |
Here is where the 3.6 comes from. Lifting Direction from 64 to 76 moves the whole reading from 68 to 72. Twelve points of Direction are worth 2.4 at its own weight. The same twelve release the eight points the ceiling was holding back from Machine capacity, worth another 1.2. Direction is the only area that counts twice, which is why twelve points there move the total more than twelve points anywhere else.
What to do. Draw your own six bars on one page and mark the one you would defend in front of the board. Then take the check and see which guess was generous.
Who: you, before anyone else sees it. How long: twenty minutes for the page, twenty for the check. You end up with: your guess and your reading side by side, and the area where they disagree most.
The reading is yours. You run it, you keep it, and it is never published against you. No table of similar organizations is printed, because there is no honest pool to build one from yet.
So it is used twice. The first reading says where to spend the coming quarter: the lowest area, and whichever one holds the ceiling down. The second, ninety days later, says whether the spending worked.
A reading goes stale in plain view, which is the point of the date. A number from last year records what used to be true. On any day you take it, the only fair comparison is with your own earlier reading.
What to do. Put today’s date on your first reading and a date ninety days out in the calendar. Send the six bars to the person who owns amplification and ask which one they intend to move.
Who: you send it, and the person who owns amplification answers. How long: five minutes to send, one week to hear back. You end up with: a dated baseline, a second date in the calendar, and one named area somebody has agreed to lift.
