This is Part Two of a series of posts about 4gency’s first year in operation, including data on monetization, app marketing campaigns, and user acquisition. If you’re interested in learning more, contact email@example.com.
Node.Hack is 4gency's first released game, one with a long and complicated lifecycle. Born from a rich history of abstracted hacking games like Steve Jackson’s Hacker, NetRunner, and even my college project at Digipen “The Capricorn Document”, Node.Hack represented at once a basic and an impossible ideal: encapsulate the aesthetic of computer espionage in a fun, addictive way on mobile platforms.
The life of the game has spanned four platforms, two form factors, two engines, and three different teams over roughly a year and a half. This post-mortem won’t cover everything about the game’s history, but instead endeavors to abstract the business teachings learned from the long and sprawling adventure of taking an original IP to multiple mobile markets.
Node.Hack is, at heart, an action puzzle game. Sure, we slotted it in various App Store categories to optimize our audiences, but ultimately it’s a mix between real-time arcade action and puzzle-like optimization thinking.
The player controls a hacker character, stealing money out of various “nodes” in a randomly-generated computer grid while avoiding semi-randomly pathed enemies until a set target value is reached, at which point the entire grid begins to self-destruct, forcing the player to run for the exit before the world collapses. There are various weapons to help the player along but in general it’s a game of sector optimization to stay away from the bad guys long enough to get the loot. The original design was meant to evoke paranoia and greed in equal measure, and judging from the feedback we got, I feel we accomplished that mission well.
Node.Hack went through a number of graphical and platform changes over time, first with XNA on Windows Phone 7 in 2011, then to Unity on iOS, Android and Kindle in 2012, and finally a new, 3D graphical refresh released as an update in 2013. The 2013 update was significant - not just new graphics, but a whole new soundtrack, effects, UI, and gameplay speed and style (goodbye, randomly-generated levels) to tune for our existing fans and bring a bit more wide audience appeal.
While the game itself may have been well-developed to the design, not every decision we made was the right one; there have been some missed opportunities. It's been a good lesson, specifically around the challenges and missteps in the business and marketing side.
Tale of the Tape
Trying our hand at multiplatform development, we've released Node.Hack on four platforms (Windows Phone, iOS App Store, Google Play, and Amazon Marketplace), with the graphical refresh applied to all but Windows Phone so far. Throughout that time, we made contact with various media outlets, and cut a few deals with advertisers to drive traffic in each platform.
It's a hard road; clamoring for attention in these ecosystems is expensive and tiring business, and for us, the story has a sadly familiar geometry to anyone in mobile or indie - the initial spike and the consequent race the bottom. These well-weathered shapes showed up for us despite our attempts to inject new life into the games:
There are multiple data points in this rollup graph that serve as jumping-off points to other lessons. Advertising, updates, and rearchitecture are all part of the long and complicated story, but for those keeping conscious of their time, here are the topline lessons we learned throughout the various launches of the title:
The Multiplatform Wars - As a multiplatform title with no per-platform differentiation, Android and iOS made up 97% of our market share, but Android carried far less of the load than market share would predict; in all the Android version missed revenue expectations by 40% if you take market share as an indicator.
Marketing, Advertising and Media - On-device trials gave us 1% conversion, Web ads gave us 0.01%. Spillover from free mobile campaigns gave us 10x sustained free downloads and a 9x boost in monthly sales (from $1/week to $9/week) which appears to be holding steady four months later. Media gave us initial exposure on launch but didn't pick up on our update, which was a key missed opportunity.
Learnings, Changings - Aggressively targeted marketing, a focus on sequels (not updates), and a potential move to premium pricing in new platforms could help Node.Hack get a better return on investment.
The Multiplatform Wars
Built originally in XNA, Node.Hack was released on one platform only: Windows Phone 7. While it garnered a bit of attention and gave us a working model of mobile ad sales and trials (see Marketing, Advertising and Media, below), our user base - even free - was well below 5,000 with revenues at less than $10/mo.
With a codebase in C#, Node.Hack seemed like a natural candidate for porting to other platforms with either MonoGame or Unity. Early experiments with MonoGame in early 2012 didn't yield good results (we gather it's gotten much better since then), so we settled on Unity, and rapidly had a prototype up and working across iOS, Android, and Kindle devices, phones and tablets.
It was a rapid run from there up to release, and - in a massive stroke of heavily augmented luck we launched on all three platforms on the same day. Watching the returns come in we noted the platform differences:
While we didn't see the startling revenue difference ratio of $1.00 - to $0.24 reported in Flurry's Christmas 2011 blog post on cross-platform revenue (we saw about $1.00 - $0.70), given the massive market share advantage Android has over iOS, there's certainly a gap to answer for. Overall the difference in actual vs. expected revenue for Node.Hack on Android is roughly %40. Some thoughts on what's responsible:
Sideloaded APKs - within 48 hours of Node.Hack's release on the Google Play store, searches for "Node.Hack APK" returned pages full of links to sites where the game could be downloaded for free, without paying. That kind of thing doesn't help revenue.
Device and Feature Fragmentation - according to Google Play, our game's APK supports 2,500 different devices, but we couldn't test on them all. The first week of Node.Hack's release, a number of negative reviews were levied against the game citing issues with new versions of the operating system and niche devices. The reviews led to returns and likely to lower sales.
Asymmetrical Marketing Spend - we spent a bit of money on iOS marketing that we didn't spend on Android (see below in Marketing, Advertising and Media) but the gap existed long before the spend. It may have skewed the average but probably not by more than a percentage point or two.
In the end, I was personally glad to get the game out on Android for those folks that wanted it; our metrics still don't track Android plays specifically so we can't get a bead on the purchase-play gap and dial in a more exact piracy estimate, but at a maximum it could be 40%, and at a minimum could be 0%. Both of those scenarios seem a bit implausible, so taking the mean may be the best guess.
Though it may not need to be said, Windows Phone and Kindle remained low in sales. We had high hopes for Kindle Fire at the time, and may still apply to the Indie Game Store on Amazon if we refactor for PC, but even at 59% of the US Android market, we just haven't gotten a foothold on Kindle Fire and the Amazon Marketplace.
For those wondering about Tablet vs Phone usage, our metrics for iOS show it's roughly 25% tablets, 75% phones that have installed Node.Hack. It was released as a Universal binary, meaning the same price and features across phones and tablets. It may have been slightly better to differentiate, but considering the extra work it would have taken, it probably washes out.
Marketing, Advertising and Media
You've heard it before: have a dedicated marketing budget for your game. While we played it relatively cheap on Armored Drive, we invested in a few marketing experiments on Node.Hack's various versions, including trial versions, web ads, and in-device ad partnerships to try to drive eyes and conversions. These met with various levels of success.
First, with Node.Hack on Windows Phone, the game launched as free with ads that showed in the top bar. Two months later, "Node.Hack EX" was launched: an ad-free version of the game with expanded levels and new weapons. The free version was modified to upsell (though not annoyingly so, just a menu option). The numbers tell an interesting story:
Mobile ads are a tough business - free downloads swamped paid downloads at roughly 10:1, and yet the paid version at $0.99 made twice the money the free version did in the same amount of time. Given the ad ECPMs of $0.44 and 19,000 impressions (roughly 5.4 impressions per installed user), this is unsurprising but a disappointing reminder of the difficulty of mobile ad performance.
Few conversions - trial conversions at 1% are well within the understood window of 0-5% for game trial conversions (at least, the historically understood window, search "game trial conversion rate" on Google and your top results are from 2008 - talk about a sailed ship). We could have boosted it with more annoying upsells, but in the end it's a funnel, and we just weren't getting enough folks in the funnel for it to matter much.
When the game launched on multiple platforms, we did get a bit of media pickup. This was great news and was responsible for the early pickup in sales numbers. We scored 4 major Android stories and 2 iOS stories, including video spots on Gamespot and a great writeup in PocketGamer.
In March 2012, Node.Hack won the Pocket Gamer Silver Medal award - an achievement to be proud of! We decided to follow up with a skyscraper web ad as well as a Facebook campaign to drive eyes to the game and hopefully inspire more sales.
For about $525 we reached 1.3M people across a number of different demographics and walked away with overall lower sales than before the campaign began:
There's a good chance this isn't inverse correlation, just ineffectiveness combined with a natural downturn in sales. The lesson: if you're paying for eyes, get the right eyes. With our shotgun approach, we lacked the kind of aggressive targeting we needed to grab people who were prepared to spend money on a game like this, on their device. (Remember, 63% of new app discovery is searching on the App Store). Expensive lesson!
When we did finally get our act together with on-device marketing, we chose to partner with FreeAppADay; their install base of 8M on-device users and extensive Twitter notification network would be our best opportunity to get folks discovering and loving our game. Hopefully, good ratings would boost us a bit when we got back to paid after the end of the campaign:
What we found after the smoke cleared:
It got us users - the campaign, which lasted for a week, gathered up more than 30,000 new users, and increased our paid revenue after the campaign to 9x the amount we were making before.
They may not have been the users we wanted - for all the users, reviews and ratings on the app store were few and far between, median session length went down and the number of users who played 1 session and never returned went up, suggesting these folks weren't inclined to deeply engage on our game. Again, a mis-targeted marketing effort.
It continues to pay, but slowly - our 9x boost to sales continues to this day, but at that rate, paying back the $3,000 is going to take quite a while.
One more critical thing to note: while our team sweated and fought to put forward the 2013 graphical update to Node.Hack, delivering a brand new audiovisual experience across three platforms, our efforts to get it noticed by the media outlets that praised the original Node.Hack were met with complete silence. The update did not get even one story in any publication. This was extremely painful even as our initial customers praised the update.
In the end, it may have been a better business decision to have released the update as a sequel, even though it would have been a hard sell for our existing customers.
It's easy to look back and say if we'd tweaked this value or that metric we'd suddenly be millionaires, but in the end our lessons are driven by much larger hydraulic systems we need to get better at navigating:
Marketing: get the right eyes - we don't know how many potential users our game initially got exposed to, but I know it wasn't enough and it wasn't the right set. An aggressive marketing strategy to identify a targeted, spend-willing set of potential users with a super-simple funnel (i.e. on-device) is something we should have spent more time and more money on.
Media: be new, be bold - mobile games begin to die the moment they are birthed; my decision to release the new Node.Hack as an update and not a brilliantly shiny "Node.Hack 2" killed a potentially huge PR beat that may have breathed much more life into the game.
Monetization: consider premium - while we didn't discuss it specifically above, it may have made more sense to price the game higher, especially the update, along with targeted marketing to draw in hardcore gaming users in a deeper spending niche. Node.Hack is a unique game and may never reach that million-mark needed to drive freemium or low-price revenue. It may mean a transition for the sequel to a PC format and more premium feature set. You'll hear about the sequel one day soon.
Node.Hack represented 4gency's first foray into the amazingly complicated, highly competitive mobile games market. Having brought the game to four platforms (and more to come, we hope!) and over 50,000 users, we've been delighted to hear the feedback from our fans, even if the revenue numbers weren't up to hopes-and-dreams levels. We'll keep at it, and hope you'll stick with us.
Charles N. Cox has worked on games for Sierra Studios, Microsoft, and Sony. He is now the Developer Education Manager at Xbox, and runs his own mobile/tablet game company, 4gency. Charles blogs at http://www.charlesncox.com.