Papermill

The first few weeks

This is an attempt to catalogue and analyse the first few weeks of Papermill, an Instapaper client for Android that I developed with the help of Matt Legaspi, a designer and friend.

Why

My original motivation for Papermill was to create a simple, minimal Android application that would serve as a portfolio piece, having just accepted my first full-time Android developer position.[1] I also wanted to have an app that I could show as being entirely my own. Having used, and admired the craftsmanship of, Instapaper since purchasing an iPad at its launch, I was intrigued when Marco Arment issued a challenge regarding its development on Android. He retracted it shortly thereafter, but it made me take a look at the Instapaper clients on offer and realise that none were what I'd hope an Android Instapaper client would be. This, alongside my urge to create my own app, was my motivation. I set out with the intention of using the compatibility libraries, the recently released design guidelines, and an open-source library[2] to create an Instapaper client embodied Android best practise and, more simply, that I would want to use. While I intended to sell the app, money was not a driving force.

Development

I began development of the app in mid-December and had a prototype by early February. I then started working with Matt to refine its design. While I already had basic interaction concepts and a vague notion of the look and feel in mind, Matt refined the concept of a psuedo-Holo app with papery textures and added a definitive look and purpose to its elements.

I worked on weekends and most evenings. I contacted Richard Dunlop-Walters, who provides support for Instapaper as well curating its editorial content, and he agreed to provide his expertise, opinion, and a degree of QA. The intention was to launch the app with the minimum baseline of features necessary to use the service, allowing for rapid development and prioritisation of other features based on user-feedback.

Launch

The app launched on Friday the 10th of March. It was intentionally launched 3 days before the launch of Readability on Android, with the hope that it would be able to garner some degree of associated coverage. It did not.

I made a point of posting links to the app and its website on Reddit, specifically its Android and Android development subsites. These forums, along with retweets by friends of its launch on Twitter, generated the majority of its intial traffic and interest, with approximately 450 page hits on the day of launch. Traffic to papermill.me from March 8th to March 15th The intial response, especially on Reddit, was mixed. While people liked the user-experience and design, they felt $4 was too much for an Android app, especially when added to the Instapaper subscription account it would require as a third-party app. 16 units were sold on the first day, 4 on the second. This fit my rough expectations for pick-up.

The app received exclusively 5 star reviews on Google Play[3] for the first few days. After this, it's lack of core Instapaper features like folder-support and the ability to delete bookmarks, and the ability to use some of Android’s core features like sharing, began to reflect in the scores. I had intentionally not included these features, hoping to hasten the launch of the app. As of March 30th, it maintains a 4.4 star review average with 45 ratings. The majority of praise is for its back-ported, ICS-centric design and the majority of its criticism is for a lack of features users are used to seeing either on the iOS client or on various Android alternatives. No-one who has already purchased the app has said it is not worth its $4 price-tag.

Media Coverage

I was contacted and interviewed by David Pierce of The Verge, a general tech blog, regarding Papermill on Friday, March 23rd. He published the ensuing review shortly after our interview. It remained the featured story on site for the majority of the weekend.

BeautifulPixels, a site highlighting design in applications and sites, published a design-centric review on March 25th.

Lifehacker, a site predominantly about optimising your life, published a short review of the app on March 27th.

While all 3 drove traffic and sales, The Verge was by far the largest influence, with new installs peaking at 107 on both the day of its publication and the one following. Daily installations since launch

Since then, a number of smaller sites have reviewed Papermill in Spanish, Portuguese, Russian, Chinese, and Dutch.

Financial data

As of Friday March 30th, Papermill has been on sale for 3 weeks and has sold 411 units at ~$4.00 USD on Google Play, resulting in $1630.12 gross sales. The net revenue after market fees is $1140.00. 79 orders have been cancelled or refunded.[4]

Gross revenue since launch
Gross revenue since launch
Sales split
Sales split

Papermill’s incurred costs consist solely of web-hosting, a domain name, and Richard’s contracted services. These account for approximately $550, resulting in approximately $590 in profit thus far.

It’s worth noting that one of the most called-for features is more fonts. While there are many freely-licensed fonts, those most commonly called for are sold, justifiably, at around $200-$300.

While it’s hard to estimate exactly, I believe I’ve spent at least 280 to 320 hours developing Papermill in one way or another over the last 3 months. At a freelance rate of $90–120/hr, my expectation as a freelance Android developer, this would equate to somewhere in the range of $30000. To be reimbursed for my time working on Papermill at freelancer rate, assuming the current rate of installations/week and that I stopped development entirely, would take approximately another 5 years.[5] While this is besides the point of its development, it’s worth noting. Given that the media coverage and user interest has already far exceeded what I expected, I find it hard to believe the sales rate could increase, let alone maintain their current rate.

Handset data

The most common handsets
The most common handsets

It’s interesting to note that, despite the fact that the app is listed as supporting 2.2+, the top 3 firmware versions are all Ice-cream Sandwich variants, accounting for a little over 40% of all sessions.[6] To my mind, this trend can be explained by the general lack of apps that adhere to the ICS guidelines (or, for that matter, any design principles at all) and the willingness of those who do have a high-end handset to pay for such an app. I believe that this is also due in part to Instapaper’s user-base, who I imagine as being quality-conscious, technically minded, and willing to pay for a good design experience.

Conclusion

Please note: I've clarified some of the points I mention below here.

Media coverage and sales of Papermill have already far exceeded my expectations and I can’t envision either increasing. With this in mind, it’s easy to conclude that the the application, with its current price and need for a subscription, will never generate a profit, especially when costs like fonts still need to be incurred. If I were to create a 'freemium' or ad-based version, the app’s profitability would almost certainly increase but I believe that this would decrease the quality of experience that the app offers and that is rare on the Android market.

I think this unhappy end-scenario - of applications that either compromise on quality or have not had the necessary time invested in their design - is as a result of Android users not being willing to pay for an apps whose focus is quality and whose price reflects this. Instead, these users opt for a free but less refined experience. This has led to a race to the bottom, with independent developers creating applications are de-facto free instead and relying on ads for profit. The quality of the design and user-experience are subsequently not a factor in their creation, as there is both no great impetus to provide it nor any expectation from the user that it will be forthcoming.[7] Applications for larger companies that are developed for multiple platforms are also not developed with quality in mind. The Android version in this scenario is often turned out as quickly as possible simply so that company can advertise as being on the platform. In the majority of such cases, the Android is either a direct port of the iOS app or is a copy of it, with UI elements and UX principles from iOS clearly present.

I think this at least in part due to the nature of the average Android user. To my mind, the majority of Android users want ‘a smartphone’ and end up with whatever Android model fits their price range, rather than what iPhone users want (‘the iPhone’).[8] (If you’re an Android user who’s reading this, though, you’re probably not that person.) This leads to users who, by definition, aren’t willing to pay for expensive/high-quality anything, the opposite of the user at whom the iPhone is unabashedly marketed. While “cheaper smartphones” is an entirely valid core market to target (and one that is actually Android's strength - while device manufacturers will always be creating mid-range Android handsets and can edge into the high-end market, Apple is highly unlikely to create anything but a high-end smartphone), the resulting user expectations, and subsequent race to-the-bottom app development, is reflected in the current general quality of Android apps.

This is also not helped by the total lack of moderation on the Android Market. Each day, at least one of the top trending apps is a data-theft/ad-spamming application that’s usually nothing more than a live wallpaper of a small animation. These applications are free, have incredibly high download rates, and any a large number of misleading or clearly machine-generated ‘reviews’. They can be featured on the trending apps for numerous days. This has caused me, and I'm sure others, to largely ignore trending apps.

I would like to believe that both the size of the userbase and range of handset quality is starting to reach a point where it will support higher-quality apps. With enough owners of higher-end Android handsets who are willing to pay for apps, development of apps that focus on quality rather than mass-appeal would be be rewarded. In combination with backwards-compatibility libraries and design guidelines, it would be nice to believe that these high-quality apps, intended to appeal to the Android user with a higher-end handset, would also appeal to someone who bought a lower-end phone 'just for email'.
Perhaps, though, the market size, phone range, and developer-attitude simply aren’t there yet.

- Ryan
(@secretsquirrel)


  1. I had already been developing Android applications as a freelancer and as co-founder of a limited company for 3 years.  ↩

  2. ActionBarSherlock, a truly fantastic back-port of the Android 3.0+ ActionBar pattern.  ↩

  3. ‘Google Play’ is a ridiculous name, by the way.  ↩

  4. 5 orders have been placed on the Amazon appstore but are ignored for the purposes of this document.  ↩

  5. All calculations on this page was undertaken at approximately 1am and their degree of accuracy should be considered with this in mind.  ↩

  6. It’s also interesting to note that both Android 4.0.4 and Android Jellybean, the hinted-at-but-not-announced next version of the Android OS, are both listed in the device logs.  ↩

  7. That you have no right to complain about something because you got it for ‘free’ is one of those self-inflicted fallacies I will never understand.  ↩

  8. In part, I think this isn’t helped by the ridiculous naming schemes various companies have adopted, Samsung being by far the worst culprit. With no clear, advertised naming scheme and wild re-use of ‘Nexus’, ‘Galaxy’, and ‘S’, it’s genuinely hard to keep track of what handset has what specifications even as a developer, let alone a potential customer looking to purchase a new handset once every few years. To my mind, Nokia used to manage this issue of a naming convention across a wide handset range problem well - by naming handsets using numeric groupings that at least vaguely denoted speed or capability.  ↩