Speaking at Develop 2023 this week, Trailmix designer Jade Ostle revealed the twists and turns that hit merge game Love & Pies took on the way to launch.
Ostle started at Trailmix in June 2020 following a stint at King, where she’d worked on Farm Heroes Saga and Candy Crush Friends Saga.
She began the talk by describing her first week at Trailmix, at which point an early Love & Pies playtest suggested it’d have day one retention of 34.6%. “This meant that, on my first day, we had to look at making radical changes,” said Ostle.
Trailmix was already working on a new meta, so the next job was to tweak the core of the game.
Trailmix created three new gameplay cores to test alongside the game’s new meta. “First, we made a match-3 with merge mechanics,” said Ostle. “We also kept our original core but made improvements, which was a tapper, a Toon-Blast-like with merge mechanics.”
“And we made a linker core with merge mechanics. We made sure that these cores had a fair amount of content for testing, with 50 levels each. We also began working on a merge-2 core, but quickly scrapped it in favour of making something brand new.”
The team tested all three cores with small groups of players, and found that the tapper came out on top with 40% D1 retention.
Trailmix also “burned the cafe to the ground” with an entirely new story meta that boosted D1 retention by another 10%. “But this still wasn’t good enough for us – and it solidified that our meta was amazing – it was the core that was bringing the game down,” said Ostle.
The target was 60% D1 retention. “There were lots of debates internally as to what was holding us back, but we determined that the core was the key suspect,” said Ostle.
“At this point we’d already made a ton of improvements to the tapper core, including but not limited to: streamlined onboarding, 500 levels, redesigned boosters, streamlined onboarding again, playtest cloud review, new blockers and constant level rebalancing. But none of it was good enough to push our retention to where we wanted it to be.”
At this stage in development Ostle says the team were struggling, and “kept trying to innovate in places that didn’t need it”
“We spent too much time making small changes that had very little measurable impact,” she said. “We kept trying to fix the symptom rather than the cause – we ignored obvious flaws in the core and retroactively tried to fix them instead of the root cause.”
“We didn’t test our cores against a proven control group, meaning that we could have made a match-3 without merge mechanics and measure our new mechanics against it.”
“At this point, we were at a crossroads: continue making small changes for our current core, or scrap it altogether,” said Ostle. “There was the path of least resistance, which would get us to where we needed to quicker, and there was the long winding road which could still get us there, but it would take longer and be more difficult. This was the path we were already on. And we simply did not have time for that.”
The team had also noticed merge-2 games like Merge Mansion picking up momentum. “We could tell from early on that these games had an extremely fun core, as well as very good early retention metrics,” said Ostle. “So, after many considerations, we decided to start working on a merge-2 game. We wanted to keep the meta exactly the same while improving the core to something thats proven.”
“Though this was objectively the best decision, it was tough for some members of the team we’d spent a lot of time and love on the old game and scrapping it was painful. This is of course the very human side to game development, and we needed to make sure everyone was on board, as we all knew it was the right path.”
Trailmix set up a 12 week plan to create a new merge-2 core with 30 days worth of content to measure D30 retention.
Love & Pies soft launched with its new merge-2 core in July 2021, with live ops and feature planning also in progress. “Additionally, we set out to create features that focused on long term retention – these numbers were really important to us, as it’s much harder to retain players than it is to monetise,” said Ostle.
These included ‘story customers’ with longer-term rewards and regular events including ‘cash dash’ and ‘date dash’.
It worked. D1 retention came in at 60%, D30 was at 22% and D180 was 10%. “This was above and beyond our expectations,” says Ostle.
Post launch, the team set about implementing proven features that make the game more sustainable, while also taking a few risks.
“We also want to ensure we have room to make awesome new stuff you haven’t seen before,” said Ostle. “This also means that the team has room to contribute ideas and build stuff they’re excited about, which is very motivating.”
There were also some parts of the game to fix, like economy balancing, a lack of reward layers and no long-term planning for content treadmills.
Rounding out the talk, Ostle said the team had spent too much time making small changes that had very little measurable impact, and “ignored obvious flaws in the core and retroactively tried to fix them instead of the root cause.” Pivoting earlier would have avoided some of those mistakes, she said.
Most of all, Ostle said that game-makers shouldn’t be afraid to completely change things if they’re not working.
“Banging your head against the wall will only make the process take longer,“ she said. “We spent a lot of time patching up the symptom of the problems we faced in our old cores instead of asking why these were happening and fixing them that way.”
“Making games is hard, and it feels very personal,” she concluded. “Letting go of what isn’t working is hard but an important skill.”