Sunday 18 November 2012

The Amazon FAotD Experience

If you look on the web for posts by developers regarding Amazon's Free App of the Day feature, you'll find a lot of criticism for it.  Amazon didn't do themselves any favours when they started their App Store; an early-doors about-face on their FAotD policy cost them a lot of developer goodwill.  When originally announced, developers were told they would receive 20% of their app's list price on every download: a very appealing prospect.

However, before they rolled out the facility, Amazon changed this: developers would have to opt-in to the FAotD, under the condition that there would be no payments made at all.  The user would download the app for free, and Amazon would not be paying the developers anything for that day.
That coloured opinion fairly strongly, and other faults with Amazon's App Store were viewed harshly.  In it's original form, user reviews were completely sacrosanct from the developer: they could not contact the user who wrote it, or give any feedback that other users could see.

Amazon also take a very strong view on maintaining the lowest available price for every app it stocks; a developer is not supposed to offer their app on a different store for a lower price.  This means that the developer will be admonished by Amazon if they were to, for example, have a sale day on Google Play where they sold their app for half price.  Note how this contrasts with the FAotD: Amazon say it's OK for the version on their store to be sold cheaper, but it's definitely not OK for the reverse to happen.

This is the viewpoint I read about when researching the Amazon App Store.  Why was I looking at the Amazon App Store?  Unlike on the iPhone, where Apple's App Store is the only option available, on Android there are multiple App Stores, and it's recommended that you get your app out on as many as you can to maximise user exposure.  There generally is no cost to doing so (Amazon have a yearly fee, but it is currently being waived), but the downside is that every store your app is available from is another store you have to update with patches, create graphical assets for, maintain descriptions and version histories... basically, the more time and effort you have to invest.

In the end I decided on three stores: Google Play (the official Android store), SlideMe (a store available in lots of countries that, at the time, Google Play was not), and Amazon.  Amazon's user base was big, and seemed like a good place to go as the Kindle Fire was about to emerge (and the Fire would be locked into Amazon's App Store much like the iPad was locked into Apple's.)

Ask any Android developer what the hardest part about selling apps is and they'll tell you: getting your app noticed.  Google reports that Play stocks around 675,000 apps as of September 2012.  675,000!  In that morass of games, utilities and novelties, how do you get people to pay attention to your app?
You tell your friends, and get them to tell their friends. You email all the gaming websites you can think of that might want to review your game.  You could take out some kind of advert, but unless you have a lot of bankroll, that's too expensive an option.

I told my friends, and they told their friends, and that got me about thirty sales on Play.  I never heard back from any of the review sites I emailed (this may be because my email sucked, but I think it more likely that they get swamped with emails just like mine, and it got spam-filtered to purgatory).  Meanwhile, the version on the Amazon App Store sold a single copy, which generated no feedback.  And over on SlideMe... no copies have ever moved, ever.

Enter the FAotD!  Going into it with my eyes open, I didn't care about not getting paid. This was clearly a chance to get Pigment on a huge number of handsets, into players' hands - as with pretty much all game developers, what I want more than anything is for people to play, to enjoy playing, my games.
Having said that, it's impossible to log into Amazon's developer console, see the massive number of downloads, and not do the math:

48,756 downloads * $0.99 * 70% * 0.62 ($=>£) = £20,000

That would have made a nice addition to my bank account!  However, moving beyond that momentary pang: it's a massive win for me.  50k downloads is 50k people who have at least tried the game.  Some of them will like it, and play it.  They'll recommend it to their friends & family, or people will see them playing it and ask what it is.  In short: the FAotD is an impressive piece of marketing.
After the FAotD the app is exposed to Amazon's sales-driving machinery: the "Customer's who bought this also bought..." panel, the "Recommended for you..." panel.  Pigment would now start popping up in these; a continual source of advertisement.

The second thing it has given me is exposure to lots and lots of handsets.  As a lowly indy developer, I don't have a vast Q&A department to do strenuous testing.  I tested on all the phones I could get my hands on (i.e. all my friends' mobiles, as well as some ninja testing on display models in the Three store), and in fact Amazon have a  pretty good quality control system in place for their App Store: they test apps on a few handsets, too, before allowing them to go online.
Unfortunately, this small amount of testing available to me wasn't enough to catch a big problem: Pigment doesn't work on some Motorola handsets (Droid 4, Razr, Razr Maxx).

You can read more about the technical details of the problem here.  For this post it's enough to say that I only found out about the problem because of the FAotD.  More, I'd say that it's unlikely that there will be another surprise like this down the road:  50k is a lot of handsets, and should represent a good cross section of the entire handset market.  Anyone who had the game not work is likely to have posted a negative review saying so (check Pigment's reviews and you'll see what I mean), so I'm confident if there were another handset it didn't work on, I'd have heard about it.

Which brings me to the third gain from being the FAotD: Amazon are very good at getting their customers to provide feedback.  This feedback is aimed primarily at other customers, to inform other potential buyers about products, but it also gives me, the developer, a load of useful information.
Of course, it isn't all rainbows & unicorns: Pigment got it's fair share of 1-star reviews.  This arrived as quite a barrage, because the negative reviews come early:  if you play a game and don't like it, you stop playing then and there, and if it really ticked you off you go review it.  If the game simply doesn't work you do the same.  On the other hand, if you like the game, you're probably going to play it for a while before you report back.  This means that early on Pigment was getting as many 1-star reviews as 5-star, and for the most part I think the 5-star reviews were prompted by people who liked the game seeing the 1-star reviews and wishing to contest (for which: thanks!).

This echoes back to my earlier point about testing:  while finding bugs and fixing them is good, doing so in a live environment where it affects your sales isn't.  Finding out about the problem on Droids and fixing it is something I'm happy about, but it would have been much better to have found that out beforehand: roughly two-thirds of the 1-star reviews were because the game wouldn't run on Droid handsets.  Currently, Pigment is sitting at an average star rating of 3.3, and has a spread from 5-star to 1-star of 37, 24, 12, 11, 27.  Clearly that 27 doesn't fit the curve.  If Pigment had run OK without crashing, and we assume that those reviewers didn't feel the need to comment on it, then the spread would have been 37, 24, 12, 11, 9, with an average of  around 3.8, which would round to 4 for Amazon's star rating image.  I imagine this would have been better for sales.
More damning than the numbers; Amazon displays 3 snippets of text at the start of the reviews section, representative of the whole body of reviews.  These are good guide for a customer wanting a quick overview, and Pigment is stuck with one of them stating (now incorrectly) that it does not work on Droid handsets - those early reviews are unlikely to be edited.

Star ratings are what jump out at Amazon customers, so of course I want to try and up Pigment's.  However, better than the star rating was all the explicit feedback provided by customers.  Hearing that someone was having fun playing is fantastic, but reading a detailed description of why is even better.  Finding out what people thought worked, didn't work, how they were progressing... this was all excellent.  A good star rating is only one part of the Amazon experience: some of the reviews were very detailed in describing what the reviewer felt about the game, providing a potential customer with a much better understanding of what they would be buying than just a number ranging from one to five - Amazon's interface, with it's most helpful reviews being highlighted, best good review / best bad review...  it all gives the reader all the info they could want.

As well as that it gave me feedback on what I could do to improve the game.  The fiddly nature of the Kindle Fire's emulated Back button made trying to Undo in game extremely cumbersome; hearing that from Fire users I could implement a feature to compensate. I've also added some minor hints to the interface - how much to help the user in a puzzle game is a fine line to walk, and something I'll go into in a later post - but the feedback from the reviews allowed me to attenuate where, exactly, I was aiming.

My earlier research into the Amazon App Store actually came back to bite me, in a way: in what I'd read there was criticism of the developer's inability to respond to customers.  Because of this I thought the best way I could respond to everyone would be to write my own review.  This proved popular to many, but to others it looked like I was trying to cheat the system: I gave Pigment 5-stars in my 'review'.  This generated some more negative feedback.  I do believe Pigment is worth 5-stars (I may be biased), however, to some it looked partisan, and on reflection I chose to change the review to the average three stars: one 5-star review here-or-there wouldn't really make any difference, and I didn't want to appear to be gaming the system.

Of course, I could have avoided all this if I'd only realised you can, in fact, reply directly to reviews!  Amazon must have taken note of it's critics and allowed it's users to comment on reviews.  I could simply have responded directly to each customer.  Something to bear in mind for my next game, but I think that one big post (my review) actually has some merits:  It keeps all my feedback in one place, easily viewable by everyone. I can update it as I release patches, keeping everyone informed.  Next time I'll probably leave out all the little @soandso comments - I'll post those directly to the relevant parties, but I'll still make one big post as a review, covering the major topics being discussed.

Being the Amazon FAotD has been, for me, overwhelming positive.  While I prefer Google Play from a technical point of view, for driving sales the Amazon App Store is worlds ahead.  Hopefully both stores will improve themselves toward the other, with Amazon providing some more developer friendly features, and Google coming up with some way(s) to push it's apps at users more effectively.

No comments:

Post a Comment