Nokia's stock price has been soaring over the past month after reports of sell-outs of their latest flagship device, the Lumia 920. TheNextWeb and WMPowerUser fanned these flames by estimating anywhere between 4-7 million Windows Phone sales so far this quarter (Oct-Nov), based on the Monthly Active Users (MAU) of Windows Phone's in-built Facebook application. Let's take a look at the veracity of these claims.
Windows Phone SalesLet me start out by stating that any sales estimate based on the usage statistics of an app is inherently flawed, but it is especially so in this case. For any new platform (even one that has faced adoption challenges), early sales are driven by enthusiasts and as the platform ages, the user demographics regress to the mean and hence, usage patterns change as well. This means that any attempt to estimate sales by assuming similar usage scenarios is incorrect. I have to credit TNW for picking up on this, but they still haven't taken it into account.
But for the sake of argument, let's ignore the first major flaw, and look at their figures. The estimate is essentially based on one figure, i.e. the ratio of Windows Phone device sales to the increase in Monthly Active Users (MAU) for the Facebook app. Instead of considering the average of this figure over time, the estimate assumes that a single data point from Q4 2011 (approx. 6.7 handsets sold per active user of the Windows Phone Facebook application") would be enough to estimate sales in Q4 2012. Their rationale for using that figure is that it correctly predicted Windows Phone sales in Q4 2010. In addition to this, the Windows Phone sales figure from Gartner that is used in the estimate includes sales of Windows Mobile.
So instead of taking this estimate at face value, let's correct their errors and gauge the impact on the Windows Phone sales estimate. Let's begin by looking at the historical MAU trend:
|Source: AppStats, Gartner, Kantar|
As we can see in the chart above, the ratio of Windows Phone devices sold per MAU added shows a completely random pattern before March 2012, with no cyclicality. Thus, individual data points cannot be considered to be meaningful inputs as the random spikes seem to be generated by a low MAU added figure for that month, as opposed to high Windows Phone sales. The assumed input of 6.7 WP devices sold per MAU in Q4 2011 is part of these random spikes. If we take a historical average of this figure to estimate Q4 sales, we end up with a figure of 5.8 WP devices sold per MAU added, while a 2012 average gives us an even lower figure of 4.4. Using a median, instead of an average gives us 5.3 and 3.9 respectively. Using medians and 2012 data makes more sense as it minimized the impact of outliers (spikes) in the data and takes into account the changes in demographics, but I've taken the highest possible figure in the chart above to drive home the point. Here's the estimate listing all possible methods I've mentioned:
As we can see, the estimates show that Windows Phone sales totaled about 2-3 million so far, and may reach just 3-6 million by the end of Q4. Please bear in mind that this is not my prediction of Windows Phone sales, as this is still a highly flawed approach. This just serves to highlight the large drop in the estimates if more realistic assumptions are made.
Nokia Lumia 920 SalesSo if Windows Phone sales haven't totaled 4-7 million so far, how are they doing? At this point, we have very little data to come up with any hard numbers. However, numerous analysts have chimed in with their opinions on the state of Lumia 920 sales based on retail checks, which we can use as a barometer for Windows Phone's performance. Normally, I'm not a strong believer in retail checks as the sample size tends to be small, but if multiple retail checks point to the same conclusion, it could mean something. Here's a list of analyst opinions on the Lumia 920's supply-demand situation:
Our checks indicate that retailers in Germany say they are only now beginning to receive the 920 across normal sales channels, and the volumes being received are still very small. We believe there is some initial pent-up demand that is resulting in stores selling out of initial shipments in a few days. Nevertheless, we believe this is largely to do with the low shipment volumes rather than surprisingly strong demand.
We believe a somewhat similar dynamic is likely going on at AT&T for the 920. Based on the inventory on hand, we believe AT&T is selling only 10,000 to 15,000 Lumia 920 devices per week at the moment. We believe stores are able to sell available stock in a few days; however, we found most stores getting only a handful at a time.
Despite press reports suggesting strong Lumia demand, our retail & industry research suggests limited supply and muted consumer interest in Nokia’s latest smartphones.
On a global level, relative Google search interest appears to be only at levels similar to last year’s Lumia 800 launch.The comment about Google search interest seems trivial until we take a look at Google's research, which indicates that search interest has a strong correlation with smartphone sales.
The firm adds that while Nokia’s stock has rallied on reports of Lumia stock outs, this likely has more to do with conservative component orders than strong demand, and notes that Verizon sales reps are not having much success selling “inferior hardware and mobile platform that is unfamiliar to the average user.”
Limited initial supply was cited as the reason for early post-launch stock-outs at some carriers versus overwhelming demand.
The inputs above lead me to believe that the Lumia 920 sell-outs are a factor of supply, rather than demand. Based on this, I don't believe that Windows Phone sales are particularly different from previous quarters, as the theme seems similar to the previous Lumia 900 launch.