Assuming you have your timestamps extracted correctly, you can do something like this:
search your data over some time range
| bin span=1d _time
| stats sum(count) as count by _time BRAND
| stats sum(count) as total latest(count) as today by BRAND
| addinfo | eval days = round(info_max_time-info_min_time) | fields - info_*
| eval average = total / days | eval change = 100 * (1 - today/average)
First three rows get a daily sum for each brand.
Fourth row gets a total sum and the last day for each brand - note, if a brand didn't report "today" then it's using the most recent day. Whether that's correct or not depends on the specifics of your scenario.
Fifth row calculates the number of days in your time range.
Sixth row calculates daily averages and the percentage change for the last day.
... View more
