Every AI answer comes with a confidence between 0 and 1 (0% to 100%). The threshold decides what happens with it:
- At or above the threshold, the row gets the status Categorized.
- Below the threshold, it gets the status Review and a light fill, so it stands out.
- A row where none of your categories fits gets the status Other, whatever its confidence.
The default threshold is 80%. Change it in Settings; the new value applies to the next run. The add-in remembers the value you set, so a threshold you changed earlier stays as it is.
Choosing a threshold #
Raise it, for example to 90%, when a wrong category is expensive: bookings that end up in a tax return, or a chart of accounts full of close neighbours. More rows go to Review, so there is more to check, and more of the doubtful answers land there instead of passing as Categorized. Lower it for a first rough pass: fewer rows go to Review, and more uncertain answers are marked Categorized.
A quick review routine #
- Click Show review rows and go through them. These are the answers the AI was least sure about, often rows that could belong to two of your categories.
- Look at the Other rows. If several of them have something in common, a category is probably missing. See Why did a row get Other?
- Spot-check a handful of Categorized rows, especially where your list has categories that are close to each other.
To correct a row, type the right category into its Category cell. The next run skips rows that already have a result, so your correction stays.
Confidence is a guide, not a guarantee #
A high confidence means the AI is sure, not that it is right. Most doubtful rows land in Review, but not all of them: a row that reads like one category and belongs to another (a restaurant charge that was really a client gift, say) can be Categorized with a high confidence. Treat the results as a very good first pass and check them before you book, report or pay anything based on them. How accurate is it? explains where it gets it wrong.