Better calendar defaults for the historical statistics #329
Replies: 2 comments
-
Thanks for the suggestions @vermorel, this is very useful. I agree with the first 2 points.
What would the difference be between this view and and the The reason main purpose of having
Agreed that 6 months is pretty pointless if we have 12 months. One idea we've had is to allow you to select both the timeframe (e.g. |
Beta Was this translation helpful? Give feedback.
-
The difference between 28 days and 30 days is minor, and clearly this isn't a showstopper. At Lokad (we specialize in statistics), I routinely witness a client or an employee misinterpreting the data in front of their eyes because they misread the day-of-week pattern. Thus, it would just a be a minor thing to help your userbase fall into the "pitfall of success". About the monthly reports, I see you point. My professional opinion remains that monthly statistics are going to be accidentally noisy, but I understand that some businesses may want this, and it's fine by me. I would still suggest to unify those two reports into one, as the comparison would be more than the sum of the two independent reports. Letting the end user select daily/weekl/monthly is the best. I did not suggest this because it's more dev efforts and I wanted to focus on the low hanging fruits, but clearly if you are willing to go down this path it would be awesome. |
Beta Was this translation helpful? Give feedback.
-
Most analytics apps (including Google Analytics) do a terrible job at reporting their historical statistics because they don't take into account the two dominant cyclical patterns that drive web traffic. It would be awesome if plausible where to provide those better default.
Web traffic follows (nearly always) two strong cyclical patterns: day of the week, and week of the year (seasonality). Not "abiding" by those cyclicities make the result hard to read and confusing. Thus I would suggest a short series of changes in the default of the calendar selection:
Last 30 days
, I suggest theLast 28 days
. Sticking to full weeks make numbers easier to interpret, precisely due to the day-of-the-week effect.Last 7 days
. Frankly, I am not to sure to see any practical situation where this report would be more useful than the last 28 days which is going to be fine even on mobile.Last month
andPrevious month
by a single reportCurrent vs Previous Month
that superposes the last 28 days (yes, it's cheating) with the previous 28 days.Last 6 months
andLast 12 months
. Cheating again, but the monthly statistics are just invariably skewed by the day-of-week effects. Display a timeline with 52 points is just fine. Also, theLast 6 months
can probably be removed, I doubt that it serves any real purpose.Finally, it would be really awesome if there was a
Current vs Previous Year
view where the last 52 weeks are compared to the 52 weeks that precede. Indeed, as web traffic tends to be highly seasonal, the only way to make sense of the traffic is compare to the relevant period last year.Beta Was this translation helpful? Give feedback.
All reactions