The relatively new mathematics of network theory has been showing this with elegant equation after the next. That it is more important who you are connected to than to how many. In this case the who is a hyper connected distribution engine like Amazon's app store, the how many comes by hierarchical association through the store for any apps hosted and given featured placement there in. Once that brand recognition is made it is as if the company and product were validated by Amazon itself, the power of that influence...granted as quoted above for 0% rev. share could be invaluable to a company just getting started. It enables the quality of the product or service on offer to forgo the need of being absolute best of class (it just needs to satisfy the need adequately), as the network effects (especially as an early player) could lead to a spike in revenue to the featured player before any one else gets to shine their product before consumer eyes. After that spike the creators could be rich, quite a return on investment should one be fortunate enough to get the nod.
However, this leads a larger discovery, that the internet has democratized *access to influence* of other highly networked individuals, institutions or entities, it only requires the mention of those less connected by a connected node to propagate strong lines of interest through the super node to the lesser entity and changing it's influence in the graph of nodes for ever.
This latter realization means it may be optimal as a strategy to seek to bolster influence by access to super nodes over even attempting to improve the core elements of a product or service first. In many industries the differences between players and marginal, for example fashion. There are millions of people producing t shirt designs and selling them for profit but because designs are more or less the same in that they all are printed on t-shirts, it's more important to spread the idea of the designs as wide as possible rather than play up the quality of the shirts. Promulgating the designs through a super node would enable a poorly designed set of shirts to gain more sales than a really well designed set. The power of access to influence is strong in t shirt sales. In contrast other businesses must have a differentiated product in place before attempting to sell to any audience. A DVD player that fails to work or even has a high failure rate will not last long on the sales lists as it's failures will reveal themselves in the first set of purchasers...even if those purchases come through the efforts of a super node. Microsoft is a perfect example of this, they spend billions marketing their windows phone product and because of short comings of the product in comparison to the competition, it failed despite the massive marketing muscle brought to bare on the problem in an effort to use Microsoft's influence to boost sales of phones using the OS.
I think the failure of the company to gain influence came because of several key reasons unique to the service at this time: the duration of the sales period was too short, amazon just launched the app store and few people are actually downloading from it (it's a bear just to get on your phone). That is why he failed to get a good boost from the influence offered by the one day free presentation in my view. I predict things would change significantly as time elapses and more users begin using the app store and if the promotion time is long enough for network effects to take over.
Determining the proper strategy to take will then vary with the type of product or service on offer, it's level of differentiation in the market and the availability of super nodes that can promulgate those products or services and thus mitigate (to the extent possible as described above) against failings in the product or service itself.