What should a useful weekly review tell you?

Begin with the outcome, not the streak

The goal is the standard for the review. If you want a more reliable lunch routine, meal times may matter more than the number of foods recorded. If you want to speak German at work, a comparable speaking attempt tells you something that study minutes alone cannot.

Your reason matters too. Saving for a course, feeling more organised or having a calmer afternoon can lead to different next steps. Your agent should use the reason you shared rather than invent one.

Check how much of the week is actually visible

Three recorded days do not describe all seven days. No purchase entry does not mean no money was spent. A review should say what data it used, what is missing and whether the previous period has similar coverage.

HAB can analyse the records available to that agent. A background review cannot use information that remains only on your phone. Estimates, self-reported values and measured values should stay distinct.

Example“You recorded lunch on four workdays. Three were before 1 pm. Last week, only two workdays had lunch records, so there is not enough comparable coverage to call this an improvement.”

Compare like with like

Use the same measure, units and time span where possible. Compare a category of spending with the same category, and keep currencies separate. Do not count money transferred between your accounts as a purchase. For learning, compare similar tasks rather than one easy exercise with one difficult exam question.

A pattern is a clue, not proof of cause. If an earlier bedtime and a better morning appear together, that is worth exploring. It does not establish why the morning was better.

Choose one change you can test

A review becomes useful when it leads to a manageable next step. You might protect one lunch break, plan one home-cooked dinner or practise a specific conversation. The action should follow from the evidence and fit the constraints you have shared.

Avoid changing five things at once. With one clear experiment, the next review has a better chance of explaining whether the change helped and whether it is worth keeping.

Keep the review in the conversation

HAB’s recurring analyses appear with the relevant agent. A normal weekly rhythm is a starting point, and the cadence should fit the goal. A short project may need more frequent reflection; a slower goal may not.

You can ask for a review, change its cadence or pause it. The agent should only say a future review is scheduled when that arrangement is actually saved. An automatic analysis does not imply permission for an extra phone call.