Posted above for illustrative purposes, is a typical Usage history graph for an individual AP (over 30 days). Looking at around Aug 14/15 there is a peak that indicates ~1.8Mbps.
My assumption has always been that peaks are 'flattened' in the historical graphs as the data point counts increase, but are also spread out over a longer overall time period.
(ie. If I'm looking at a 2-hour interval, the data points are minute by minute. If I'm looking at a 30-day interval, the data points are hour by hour.) <<< Example only, I have no idea if this is accurate...
I have two questions that I'm hoping to have answered:
 Is my assumption that usage peaks are flattened correct, or is the actual traffic peak presented on the example graph (~1.8Mbps)?
 If my assumption is correct and the peaks are indeed flattened, can anyone advise what the actual data point spread is? 1-min, 5-mins, 30-mins, etc?
1. Yes, they absolutely are. You can see this simply by toggling between "last 2 hours" and longer, like "last week".
2. No idea... This is a good question.
What may be helpful to you often the API calls for similar data allow you to specify custom time ranges as opposed to the dashboard default last two hours, last day, last week, and last month. I'm not sure how summarized these are though.
Thanks guys - you've corroborated what I'm seeing as well. I've generally noted higher usage 'spikes' when viewing historical data over shorter time periods (not on every AP, depends on the usage of course).
The good sauce will be to actually confirm the details of how that data is presented. Time to open a case!
I wonder if a vertical "bar" for each sample might be better for the graph, where the bar goes from the average to the peak thoughput.