Puzzle Retreat, their next big game, garnered several notable awards and positive reviews, but the Agents' extensive blogging and critical analysis of their own work has shown many areas where it could've been improved, and they bluntly lay those on the table in a post-mortem which has been posted to Work In Beta.
On hand to discuss the process were the Agents' Henrik Pettersson (designer) and Sam Wong (programmer).
There were several key takeaways, both good and bad, on making Puzzle Retreat:
- The game wasn't immediately accessible to random strangers on the street in a round of early testing. As a control experiment, the team then tried playtesting in a similar fashion with other successful games which had already been released. They found similar barriers when people were first exposed to those games - sometimes it just takes time for people to get used to your game, and that's ok.
- Computer-assisted design paid off for the team in terms of rapid prototyping of new puzzles. They had a program which plotted the correct course from the beginning of each puzzle to the end, which 'frees up the designers mind to think of the different parts of the puzzle rather than having to keep [the whole thing] in working memory', according to Wong. 'Good tools make good games,' adds Pettersson, who noted that only 20 out of every 100 puzzles created made it into the game, necessitating speed in their design pipeline.
- Players got more enjoyment out of solving a less challenging puzzle if it was prettier than a similar assymetrical or messier one.
- When you're working with a small team, you can't monitor absolutely everything through metrics. The Agents' advice is to ask specific questions of the players and use the metrics as a scalpel rather than a broadsword.
- The robustness of the menu in which people make their in-app purchases is directly correlated to how much value the players will place on the content they're about to buy. Much like a dingy shop front in real life, if your in-game store appears flimsy or fragile, the intergrity of the content you intend to purchase will be sub-consciously called into question by your would-be patrons.
Far and away, however, the primary takeaway from the team was that their freemium model had problems. The free game gave players 60 puzzles with subsequent puzzles available in packs thereafter.
Speaking to Develop, Voxel Agents Creative Director Simon Joslin said:
I feel like the core gameplay itself, the polish and the content all worked pretty well. I'm happy with the game's design itself, but if we did it all again, the free-to-play and monetisation would have to be drastically reconsidered.
Selling content is a mistake, at least in the way we were doing it. The typical player will play through 60 puzzles for free, but that's actually quite a nice experience and quite enjoyable in itself. Once you get to the end of that, you have to be quite motivated to then spend money to continue, and we didn't really build any hooks or major motivational tools to encourage the player to complete the whole game.
It's not like you're on some kind of leaderboard or have some other social motivation showing you how far you've gotten in the game, so really you had a quite complete experience just playing through the 60 puzzles.
Joslin explained that for a game like theirs, a model where moment-to-moment hints to help people play through the content, rather than charging for more content itself, might have been a preferable strategy.
You can check out the post-mortem video here.