Analysis by time slot
What the time slots show, and don't show
Before looking at any number: in this cut, day of week and time slot are perfectly confounded. Friday was only measured in the afternoon, Saturday only in the morning, and Sunday only at midday. We can't separate whether an observed effect is from the hour or from the day.
Summary
The Friday/afternoon group has 142 calls, but only 13 returned capturable text, producing just 10 mentions in total. Even limiting the comparison to calls that did return text, Friday/afternoon's average mentions per call (0.77) is far below Saturday/morning (5.73) or Sunday/midday (5.59). With a sample of only 13 calls, this could be statistical noise, not a real pattern of that time slot.
Why this page doesn't compare "morning" to "afternoon" in general
This cut's design assigned a single time slot to each day of the measurement window, not a different slot repeated across multiple days. That means any difference between groups could be due to the hour, the day of week, or both at once, and this cut has no way to tell them apart. A future cut that combines several time slots across more than one day would let us isolate the hour's effect from the day's.
This cut's three day · slot groups
Each bar shows the group's total calls; the detail states how many of those calls returned capturable text and how many mentions they produced.
Frequently asked questions
Questions about this page
So does the AI recommend worse on Friday afternoons?
We don't know from this sample. 91% of that group's calls (129 of 142) have no capturable text because of a capture-pipeline gap, not because the assistant recommended nothing. With only 13 usable calls, that's not enough of a base to claim anything about that time slot.
Why didn't you re-run the 129 calls with no response?
This cut is frozen exactly as it was captured, gap included, precisely so it can be documented transparently. A future cut can re-run that slot with the capture issue fixed.
Keep exploring the dataset
This is one of six dimensions we analyze. The rest of the variables, and the full dataset to explore on your own, are on the data page.