- ARPDAUPosted 4 years ago
- What’s an impressive conversion rate? And other stats updatesPosted 4 years ago
- Your quick guide to metricsPosted 5 years ago
The Spilt Milk Live Post-Mortem: Week 3 – Beta path to your door.
How the devil are you? I’m Andrew J. Smith and welcome to part 3 of the development diary for what will – until its release – be known as ‘Project #2’ (or Angry Lines, or Line Em Up). So until now we’ve been over how the project got ‘green-lit’ (mutual disregard for competitor’s products between myself and coder-extraordinaire Nicoll Hunt), the reasoning behind some of the key design decisions and the processes we used to develop it from simple prototype to addictive time-waster. This week we’re going to take a look at the challenges we faced adding tutorials to the game, audio production and the Beta process – how do we get strangers to play our game and help us improve it, for free?
One of the first things we noticed when handing the game over to new players is they struggle with the controls for that crucial first second or two of play. We’d initially had the player’s avatar controlled with a simple left/right choice, much like driving a radio controlled car. It seemed to confuse a few people, especially as the turning was dependant on the direction their line was facing, so we implemented a tap-based input. Tap in the top half of the screen, it’ll move up. Tap in the left half, it’ll move left etc – with some obvious tweaking in the overlaps we felt this was as precise as the previous controls, but handing it over to friends and family proved we hadn’t solved all the issues.
The problem was particularly pronounced with people who’d played games on the iPhone a lot – they’d start by trying to swipe to control which way the avatar turns, and the resulting ‘odd’ turns would confuse them. Inevitably they’d figure it out or get familiar after a simple explanation, but we’d need to implement a tutorial system, and perhaps even the swipe controls. Of course the broader game rules needed explaining too, and so something had to be done.
One Saturday afternoon the two of us were having a bit of a game development session and this problem got solved in the most peculiar of ways. Starting with ‘why not just have a tutorial play mode with instructions’ we um-ed, ahh-ed, and generally voiced our dislike of any solution we came up with until it morphed into ‘why not have the lines talk to the player?’. Likely quite high on mutual-congratulation (and that peculiar feeling you get from ‘jamming’ some game development with a mate) we thought it was the best idea ever, implemented it, and immediately forgot the whole point of the system. Rather than specifically help the player, we just made the lines say fun things.
Quotes from popular movies, tips on the game, snarky comments, jokes, and even simple exclamations of joy – all of these things, triggered on player (and enemy) spawn and death events meant that the game all of a sudden went from an addictive-but-dry score chasing game to something brimming with character and incidental humour. Players would naturally read two unrelated comments from the AI and turn the little lines – nothing more than 1 pixel thick coloured lines, remember – into fully fledged characters. This was a huge success, and something we plan on exploiting to its fullest.
The biggest single effect on the game is that it has an immediate appeal to more casual players, drawing them in and extending the period of play they’re willing to spend on it. ‘Defeat’ matters less (there is no real failing, just getting a lower score than you wanted) and we believe this will lead to better player retention and conversion. If they play for the humour and ‘fun’ of these little characters, and then pay (or stay) for the score-chasing then we’ve really hit on something. And that something was never planned, never would’ve been scoped out, and took about an hour and a half to realise its potential on a whim.
The next step was to prove our game’s worth in the crucible of public play-testing. Using Twitter alone, we ran a ‘first come, first served’ campaign for about 2 weeks to try and get as many people on board as possible. Seeing as I was at about 500 followers at the time, getting 30 people on board was not a bad result – 6% of the audience were onboard. Not a bad conversion rate for a game that’s not even out yet! In all seriousness, we’d released a few screenshots of the earliest version, and people were aware that we were working on something, but we were very pleasantly surprised with the reaction we got.
We got the lucky people to send in their UDID numbers, and built a version of the game with Flurry support (Flurry is a lovely piece of free-to-use stat tracking software) and fired it out to them. A few unfortunates had trouble installing it but the majority ended up playing quite a lot and we even got a bit of lively discussion and feedback on Twitter (the hash tag is #AngryLines if you want to join in – a temporary name at the moment) and email.
The nicest surprise of all was the fans – we actually have a couple of people who were so into the game they were ranting and raving about it as much as possible on Twitter. It was really morale-boosting and encouraging that complete strangers with no investment in the project would so enthusiastically talk about our game all of a sudden; we can only hope the same will happen when we release and start courting the forums and reviewers! As a company I’ve tried to come across as friendly, open and fairly light-hearted; the Beta interaction (as well as the Facebook page, complete with ‘Caking of…’ features and other ‘nonsense’) has hopefully communicated that rather well.
A Datum to remember
The best thing about Flurry is that you can tag anything in your game code to be tracked and monitored anonymously. We went for a simple and useful implementation with this being our first outing – so we can track how many times a mode is played (bear in mind we have 6!), and for how long. This allowed us to easily judge which modes are the most popular, and by how much – or so we thought. But as with any statistics, we’ve learned to be a bit more critical.
The key is remembering that the modes themselves all handle ‘plays’ differently. Most of them are defined by player skill, but Gauntlet is a mad rush against insurmountable odds and so sees a lot more plays (read: ‘retries’) than most modes, while Deadline is limited to three minutes long and so the player will likely get a more satisfying experience in one play than they would in another mode (seeing as it’s a guaranteed 3 minutes and the player can’t ‘lose’). We believe this led to fewer replays and so the mode being under-represented in the stats. With that in mind, here are a few (unedited) figures from the system at the time of writing:
Of course we didn’t want to artificially skew the data so we point-blank refused to discuss tactics and play styles in the game with the Beta participants, nor did we tell people how to control it. To get a true and honest reaction to the game we needed people to come to it fresh and find the fun (or not!) themselves. So talking to the new fans was great fun and really rewarding, but pretty much limited to thanking them for their feedback and enthusiasm, while gently encouraging more play (my highscores have been thoroughly and embarrassingly thrashed, sometimes by double!) and talking about the potential for alternate control styles. Predictably enough our existing mode ’tappy’ suited some people fine, but enough suggested a ‘turny’ version as well as the ‘swipey’ version – both of which I discussed earlier. Even if we only got that useful insight out of the Beta, it would’ve been worthwhile as this is the kind of thing we cannot predict and can easily accommodate.
The final part of the puzzle is audio
The final part of any game’s puzzle is the audio. I often find this gets left till the end of a project, and in this project I can ashamedly say I did not get on top of this aspect early enough. I left it too late for ‘my usual guy’ to help out but luckily I was able to fire out the scope to a couple of different potential contractors and, while one did not manage to get anything back to me by the deadline, the group who did manage a pitch have delivered wonderfully. The budget is tiny, and the schedule a little tight, but through careful references, a half-decent scope document, and close communication with the chaps involved we have been able to get some really great tracks that fit the style perfectly. As an insight into this process, here is an un-edited excerpt from the document:
The game is an old-school classic (Tron) brought bang up to date via a few modern classics, like Pacman: CE DX and Geometry Wars 2.
We’re looking for an upbeat, blood-pumping and dub-ish drum and bass kind of soundtrack, ideally with a ‘theme tune’ to run on most of the modes (plus the menus) and a couple more tracks, with sfx too.
We also want to hark back to chip-tunes a touch, with elements of Daft Punk, The Prodigy and the like."
Again we lucked out in that my original choice – the man who did the music and sound effects for my previous iPhone game (Crunch – The Game) – wasn’t able to fit us into his very busy schedule but is always open to lending an ear and some choice professional advice. He was invaluable in picking the samples we used as reference and, much like a mood board for artists, this sort of thing is invaluable in communicating the desired end product. As a treat, here’s that section of the document too:
^ probably too dub-step-y, the dirty ‘wub-wub-wub’ would be (if chip-tuned up a bit) special.
^skip to 2:20, this shit is hot
^nice balance of chip-tune versus actual tune. Skip to 00:28s”
So that’s it for part three. Next week we should be gearing up for launch, so we’ll delve into some details on our plans for courting public interest, plan of attack for the reviewers, how we made the trailer, and the world-conquering plans we have brewing for the weeks and months beyond the game’s launch. Exciting times ahead!
Early Days Fun facts:
- Fan’s Favourite Line (most mentioned in correspondence) – Chuff!
- Most ‘Popular’ Mode (Beta) – Gauntlet, with 1,007 plays
- Word-count of Email Feedback from the Beta – 1,806 words
- Title Music Track Length – 2:42