Tool 3 of 32 · a few days
Baseline Vitals
The health numbers written down on day 0, so day 100 is measured against them instead of felt.
Captured day by day across the diagnosis window, each vital dated the day it first exists, before any treatment of the thing it measures.
What it is for
To write down the health numbers that fit this business, with a value and a source, so later claims that it is working can be checked against a starting point.
The failure it catches
It is working stays a feeling because nobody wrote down the baseline the numbers started from. Progress narratives fill the vacuum, and narratives always trend positive.
Why it works
A dated number with a named source is the only thing a later story cannot quietly rewrite. Day 100 then compares against a starting point instead of against a mood.
How to run it
- Pick the health numbers that fit this business: delivery pace, time for problems to settle, the numbers customers vote with.
- Add the vitals other tools produce as they run: knowable-to-known lag from the Deadline Replay, time-to-vindication from the Dismissed-Warning Search, divergence count from the One-Paragraph Answer, runway from the Runway Trace.
- For each vital, write the value and the source, dated the day it first exists. Runway at intake. Desk vitals as the records are read. Session vitals as those sessions run.
- Baseline means captured before any treatment of the thing it measures, not before a calendar day. A patch count captured before any fix runs is a clean baseline even if it arrives later in the window.
- A vital without a dated value and a named source is not captured. The set is complete when every later re-measure has a number to compare against.
How to read the result
The sheet is the reading. Capture the few that the diagnosis says matter. Vanity metrics wearing a suit do not count. Thirty vitals dilute review attention.
How to run it well
- Write the source next to the number. A number with no source cannot be re-run the same way.
- Empty slots are allowed. Label them empty with the day they should arrive. Claim nothing for an empty slot.
Where it breaks
- Capturing thirty vitals so none of them get looked at.
- Substituting a friendlier metric later, which is why the re-measure uses the same source and the same method.
What would prove it wrong
A re-measure that routinely needs vitals nobody captured would mean the selection procedure is misfiring.
What we watch
Whether the vitals chosen at the start are the ones the checkpoints actually need, or whether a number nobody captured keeps becoming the argument.
Ask yourself
- Which health numbers would you actually use to decide if this is working?
- Are those numbers written down with a date and a source, from before you started changing the thing they measure?
- If you re-measured tomorrow, would you use the same source and the same method?
- Which numbers are you tracking because they look good, rather than because they describe how this business actually lives?
Fed by: One-Paragraph Answer, Deadline Replay, Dismissed-Warning Search, Runway Trace
Feeds: Decision Decay Curve, Leading-Indicator Shield, Baseline Re-Measure, Checkpoint Review