Tracker history helps you judge a goal using what you recorded. Start with a consistent definition: one entry might mean pages read in one session, or minutes spent practicing. A count of entries and a total of their values answer different questions.
Use Trackers to find an existing tracker. Read its description and unit before logging more. If no tracker fits, agree on a definition with your AI before creating one. Changing the meaning halfway through makes comparisons harder to interpret.
Work through an example
For this fictitious example, your Reading sessions tracker is a Metric with unit pages. Each entry records pages read in one session, not the page number you reached. Your goal for October 5–11, 2026 was 100 pages.
The four recorded sessions are 20, 30, 15, and 25 pages: 90 pages in total, averaging 22.5 pages per recorded session. The data shows four sessions. It does not show what happened on days without entries or whether a longer book was more difficult.
Ask your connected AI to check the evidence:
Read my private Penny tracker "Reading sessions" and its definition. Review October 5–11, 2026 in America/Chicago. Show the entry dates and values, the number of sessions, total pages, and average pages per recorded session. Verify the calculation from the entries.
Compare the total with my goal of 100 pages. Treat missing entries as unknown, not zero. Separate observations from possible explanations. If the data or a result list is incomplete, say so before drawing a conclusion. Do not add or change entries.
For those four values, the recorded total is 10 pages below the goal. “I read 90 pages” is stronger than the evidence if you know some sessions went unrecorded; “I logged 90 pages” is accurate.
Decide what to change
You might choose a smaller goal, a different routine, or better logging. A suggested explanation is not a measured cause. Save the decision separately from the observations. Replace the example period and numbers with your verified results before copying:
For real work, replace every example name, fact, date, and value with information you have verified. If you are rehearsing the fictional example, prefix new records with Practice: and explicitly label their saved text as fictional practice data. Keep example updates out of real Projects.
Use the actual records and figures verified in this review. If this is the fictional example, use new records named Practice: and label their text as fictional practice data. Do not update a real Project with practice data.
Read my private Project "Reading practice." If none exists, propose one and wait for my approval. Save a dated review note with the period, session count, recorded total and goal verified in the review; missing observations remain unknown.
Attach the review note and tracker as Resources to that Project. My proposed next experiment is to read for 20 minutes on Tuesday and Thursday next week, then log the actual pages. Add those as suggested next steps only. Update Progress with the evidence and this plan. Read back what saved and report any failure.
Verify the saved result
Open the tracker and check the four dates and values against the review. The dashboard shows recent entries; a long history may extend beyond the visible list. Ask the AI for the specified period rather than assuming the screen contains every observation.
If an entry is wrong, use Correct entry on that observation instead of adding an offsetting entry. Recheck the totals after correction. Open Projects to inspect the review note and linked tracker. Confirm the proposed reading sessions remain suggestions until accepted as To-dos.
See logging and correcting tracker entries and working with Projects.