|Bitmap Frogs said:|
While I agree that all your stores being off has a low chance of happening, with just a 2% of direct data the chance that you have a chunk of stores off the country average screwing your estimations is higher. Historically, this site has missed many unexpected trend changes which suggests your spider sense has more of a weight on the total numbers than the direct data.
I am aware of how much of a challenge is increasing your direct coverage without the resources that come from selling a monthly report but it's clear that maybe the site team should focus more on that.
I agree 100% that the data would obviously be better with more sources but I don't think we do a bad job with what we get. The actual market coverage is only pretty small but it doesn't all come from one store or type of store, rather a spread which is intended to be a good sample of the market as a whole. It's not like we are taking Amazon sales, say, and using those to estimate for the whole market. We also work closely with a number of analysts and manufacturers who give us their thoughts on whether things are up or down in a particular week based on their sources who feed them snippets of top-level trends. Historically, April is a pretty slow month, and I did take factors like this into account when putting the data together, but to have seen the weekly average drop following an enormous software release is pretty illogical. Sales in the first few weeks of April were down on March for sure - our data showed the same, but the week before GTA and then GTA week itself saw large increases - MS said +54% based on their own figures (with a rise the previous week as well) and retail press releases were indicating that PS3 had an even larger increase - pretty much in line with what we had shown. Also, the 6 million GTAs sold worldwide suggests that 2.8 million in the US is too low (GTA is much bigger in US and UK than continental Europe and anywhere else - same goes for Halo).
But anyway, like I say this isn't intended as a slagging match against NPD. Inevitably people will say we need to change our data and we may well have been a little high on PSP / PS3 / 360 but I don't believe to the extent that NPD is suggesting and I believe the difference will be reversed next month which will hopefully prove my point on this. If VG Chartz is tracking a game too high or too low we will go back and adjust in the weeks we were wrong (or just across the board - drop everything by 10% or whatever) so any adjustments should correct the mistakes where they happened rather than post-correcting them in later data - which I have no proof of but suspect, looking at previous trends, that most tracking services do. What other choice do they have other than constantly release corrections, which will just confuse people and make them look bad. There will always be a degree of tweaking figures to make up for past inaccuracies - I would do it if we were in the same situation.