You can't assume that one population is like the other just because you have no data to suggest otherwise. The proper course of action is to limit assertions to the population where data is provided.
True, but it's not a wild assumption as much as a logical correlation. The US is a free market, so consumers are free to shop anywhere. You'd expect distribution of sales and interests to be uniform across stores appealing to the same demographics. Unless there's reason to speculate Amazon isn't representative of the US as a whole, it should be a litmus test. A few significant demographic populations would have to be excluded from Amazon for it to be substantially skewed.
In 2012 the top selling console for Amazon US was the PS3. Multiple skus, difference in buying habits or a number of other factors could have led to that reality on Amazon. Regardless it doesn't mirror 2012 US NPD sales in terms of order.
The question for me isn't whether Amazon's rankings are 1:1 with retail, but a general barometer. Amazon doesn't give numbers, only ranks, so the difference in that case could have been tiny and due to a few sale options at Amazon. The relevance to sales tracking is that Amazon provides the most frequent sampling of data. Is it data that means something, or should it be completely ignored? I believe the former that it's statistically relevant with a reasonable %age deviation (and crazy noise and range in samples!) from NPD retail results.
Of course nobody references Amazon more than I, they're still fun and informative charts, and the easiest charts to look at.
A major problem with Amazon is how the data is presented. There aren't actual sales figures, only relative measures, and they change at high frequency. As a single store with a single price for items, you'll also get a bigger impact from deals. eg. If a game is notably discounted at Amazon, it'll likely see a bigger uptake than the same game at higher price in other stores. Thus the sampling at Amazon is skewed by Amazon's local pricing and deals.
However, the populace buying at Amazon should be representative of NA unless there's reason to exclude some demographics, which there may well be. We have one suggestion here from Temesgen about kids being unable to buy from Amazon, and they likely represent a notable part of the console market (although maybe less so this early on?). It'd be interesting to see if different income brackets shop differently, and whether Amazon appeals more to lower or upper classes or are truly universal.
Now as to WHY that is, or that you'd have to prove it, I disagree, all we need to do is compare to NPD to see at times there is a discrepancy.
I'd expect a discrepancy. What I'd hope to see, to make Amazon at all worthwhile and not something that should be as ignored as VGChartz, is correlation within a sensible degree of accuracy. eg. Maybe Amazon is 10% out on sales versus NPD, which doesn't show in rankings tables where +/-10% of an item can move it 10 places up or down. If that's the case, Amazon is a good barometer. However, if Amazon is more like 50% versus the rest of NA retail, then Amazon's figures are worthless and shouldn't be posted any more than photographs of the entrails of fish predicting the future.
I can't believe how many words are being wasted on here to discuss a certain point, when all is needed to say is that no one in their right mind would ever use one single source to produce a clear idea of real and current sales figures - be it Amazon or anyone else.
Preach it, Brother!
Even more so considering Amazon has NO figures apart from a best selling list. And especially considering that we don't even know the size of Amazon business, due to their not publishing sales figures. We do know they represent a small fraction of the entire market.
And I love Amazon!
Arwin said:
I would only say that this whole discussion has already been on GAF, and someone there did a 'proper' polling and tracking of data on Amazon already last gen, and that correlated surprisingly well with NPD reports if I remember correctly.
That'd be a rally valuable link! If a proper investigation determines Amazon's viability as a reference (or not), it'll end this bickering and we can make better use of the most available stats tracking available to us.
You can't always connect 2 things that seem to correlate.
Was there not a month where the Xbox1 outsold the PS4?
Because, in that month, on amazon never was the Xbox1 above the PS4.
Case in point.
You're looking for a 1:1 tracking which isn't going to happen for reasons already described. However, overall trends and movements should hopefully match between Amazon and retailers by and large over a decent sampling period. In your case, the difference in sales between XB1 and PS3 could have been all of 2 thousand units putting the consoles very close. Or PS4 was supply constrained and Amazon had a larger shipment of PS4s, meaning greater availability and sell-through and higher ranking for that item. That doesn't make Amazon completely wrong as long as you're looking at the data in the right way.
