Why does a checked box tell you nothing?
A checked box records that a person clicked it. It does not record that the schedule template changed, that the insurance plans were rebuilt, or that the front desk is answering the phone the new way. Owners of dental groups tell me this on call after call, in different words: the plan exists, the people know what to do, and somehow the office still runs the old way.
We have been on the wrong side of this ourselves. On a six-location software conversion, the plan showed the insurance module complete at the first office to go live. The front desk was estimating coverage by hand and putting off treatment conversations, because the plans would not open for editing. Everyone who checked that box was telling the truth about what they had done. Nobody had written down what done would look like.
The fix is small and it changes everything: write what done looks like before the work starts, and name the thing that proves it. The rest of this page is that fix, as we run it today.
What has to be written before the work starts?
Four columns, on every task, filled in before anyone begins. The owner fills the first two with the person who will check the work.
| Column | What goes in it | Example from a conversion board |
|---|---|---|
| Task | The verb and the object, nothing more | Rebuild insurance plans for the first office to go live |
| Done when | One sentence a third person could test | Every active plan opens for edit and the top ten carriers verify against a live patient |
| Evidence | The artifact, where it lives, dated | Screenshot of the plan list plus the ten verified patients, in the board |
| Checked by | A named second person | The ASC operator, then the office manager at the Monday meeting |
The done-when sentence is the one most people skip, and it is the one that does the work. If you cannot write it, you do not have a task yet. You have a wish.
That is the conversion above. The box said done. The number that mattered, patients verified before they sat down, said otherwise, and the number is what put it on the issues list.
How does the weekly meeting catch it?
The weekly meeting reads numbers first and tasks second. At the four-office group where we run the office managers meeting, the hour opens with each office reading its scorecard. Then we review the to-dos from last week against their evidence column. Then we spend most of the hour on the three issues the numbers raised.
Three things make that order work:
- The scorecard and the task list are read separately. A task can be closed and the number can be flat. When that happens the meeting opens a new issue with the number attached and leaves the argument about the task alone.
- Evidence is opened, in the meeting, by someone who did not do the work. Ten seconds per task. A report with a date on it, a screenshot, a row in the system. If it is a message that says done, the task stays open.
- Between meetings, the operator checks the offices in short bursts. Five to fifteen minutes per office, weekly or biweekly, looking at the system itself. Asking the team how it is going comes after.
Here is what that catches. At the same group an AI phone agent answers calls in several offices. The scorecard showed thirty-nine calls in one month that the agent failed to hand to a person. That line on the scorecard is what put it on the issues list.
What does a client see at 30, 60 and 90 days?
You should never have to ask whether the work is happening. The structure answers it on its own.
- By day 30 the board exists with every task carrying its done-when line and its evidence column, and the first scorecard is reporting. The technical setup of any system we are rolling out is finished inside this window, so the next sixty days measure adoption; installation is already behind you.
- By day 60 the weekly meeting has run on that board every week for two months. You have a history of which tasks closed with evidence and which came back.
- By day 90 each month has its named deliverable, and you have three months of numbers to read against the boxes.
Two people are on your account the whole time. The operator is inside your boards every day and attaches the evidence. The senior reviews the board and the scorecard with you weekly and is the one who says, out loud, when a closed task did not move its number.
What happens when the box is checked and the number did not move?
This is the moment that separates a board from a to-do list. Our rule is the same one we use for software pilots: the success line is written before the work starts, and the result is judged against it at a set date.
When a task is done and its number is flat, one of two things is true. The task was the wrong task, or it was done wrong. We say which, in the meeting, with the owner in the room, and the item goes back on the issues list with the number next to it. What we do not do is leave the box checked and move on. A board full of green boxes over a flat scorecard is the pattern that gets an operations partner fired, and it should.
What should you ask whoever runs your boards today?
Five questions. If the answer to any of them is a pause, you have found where the checked boxes are hiding.
- Show me the done-when line on the last ten closed tasks.
- Open the evidence for three of them without asking the owner.
- Which number on the scorecard was each of those tasks supposed to move, and did it?
- Who closed each task, and was it the person who did the work?
- Which closed tasks came back as issues this quarter, and why?
If you would rather we ask those questions with you, that is what the first half of the 30-minute call is for. You describe how your boards run today; we tell you where the evidence column is missing and what we would put in it first.
Frequently asked questions
Can you verify work from Tulsa when my offices are in another state?
Yes, because the verification lives in the system. A report, a screenshot or a scorecard number proves the task the same way from any desk. We go on site when the issue needs eyes in the room, and that is a decision we make with you, by issue.
What counts as evidence for a task?
Something a third person could open without asking anyone: the report with the date on it, the screenshot of the setting, the row on the scorecard, the signed checklist. A message that says done is a claim, and claims go in the status column.
Who checks the checker?
Two people are on every client: the operator inside your boards every day, and the senior who reviews the board and the scorecard with you weekly. The owner of a task never closes it alone.
What if the number did not move after the task was done?
Then the task was the wrong task, or it was done wrong. Either way it goes back on the issues list with the number next to it, and we say which of the two it was.
How long before I can see this working?
The board and the first scorecard exist by day 30. By day 60 the weekly meeting has run on it every week for two months. By day 90 you have a named deliverable for each month and three months of numbers to read against the boxes.
Where the proof on this page comes from
Every number and named process above traces to a row in our proof library. Each row names its source and the date we last checked it.
- P-005 Two people on every client: one senior on strategy and the relationship, one operator inside your boards every day.
- P-006 A named deliverable at day 30, 60 and 90.
- P-008 Office-level check-ins of 5 to 15 minutes per office, with weekly or biweekly touch points.
- P-010 ASC runs the weekly office managers meeting (an L10) at a four-office dental group: scorecard, to-dos, issues, with Craig and the ASC operator in the room.
- P-011 Onboarding and offboarding run from role-specific checklists on a board, duplicated and assigned for each hire.
- P-012 Client work runs on a shared board with owners and due dates (the Command Board), reviewed weekly.
- P-024 Conversion lesson: after one office went live, insurance plans could not be edited or deleted, so the front desk estimated coverage by hand until the vendor fixed it. New CBCT scans did not transfer and needed a manual upload.
- P-028 39 calls failed to transfer from the AI phone agent to a person in one month at that group.
- P-043 Pilot rule Craig gives on calls: set success and failure criteria before the pilot starts, judge it 60 to 90 days after setup, and a lift under 15 percent may not justify the cost.
- P-044 Rollout rule Craig gives on calls: finish the technical setup of a system of action inside about 30 days, then track training, adoption and the agreed success measures.
- P-041 The 30-minute call: no pitch deck; by the end of it you know whether ASC is the right fit and what we would do first.