[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"patch:dev-blog-3-lessons-from-the-guildrun-demo":3},{"id":4,"slug":5,"title":6,"patch_label":7,"posted_at":8,"body_bbcode":9,"steam_url":10,"insight_md":7,"insight_published_at":7,"updated_at":11},"7834f7d9-a4df-4d11-8fc7-7c787dad89f3","dev-blog-3-lessons-from-the-guildrun-demo","Dev Blog: 3 Lessons from the Guildrun Demo",null,"2026-11-19T15:42:14+00:00","[p]Hey everyone! My name is Aleco, and I’m the lead game designer on Guildrun. With our last major demo update behind us and work on the full release of Guildrun in full swing, I thought it might be a good time to share with our community a few of the many lessons we learned about Guildrun’s gameplay over the course of the demo. When possible, I’ll also talk about how these lessons might end up impacting the game’s full release.[\u002Fp][p]We’ve mentioned this many times in our patch notes and social posts, but it bears repeating here: the popularity of the Guildrun demo vastly exceeded anything that we at Leyline had ever imagined. Most of the goals we had set for the entire duration of the demo (such as total player count and Steam wishlists) had been smashed by the end of our first week. This meant we were able to gather tons of valuable data and player feedback to help us make the very best version of Guildrun 1.0 possible, which leads us to our first lesson:\n[\u002Fp][p][b]Lesson #1: If it ain’t broke, don’t fix it[\u002Fb][\u002Fp][p]Coming into the demo, I genuinely had no idea how people would react to Guildrun. I was mentally prepared for a mediocre or negative response to the game, which would have meant fundamentally changing or redesigning core aspects of Guildrun in order to make a “shippable” game, as we say in the industry. Since the overwhelming response from our players was that Guildrun’s core gameplay was already working for them, the situation ended up being the opposite.[\u002Fp][p]The #1 lesson I can take away from the Guildrun demo is that its core gameplay loop already works, and that anything we add or change for 1.0 shouldn’t mess with that. This might seem obvious, but it isn’t so obvious when you’re actually doing the design work. As the saying goes, “Perfection is achieved not when there is nothing more to add, but when there is nothing left to take away.” This means that most of the big changes players can expect to see in 1.0 will be improvements or additions to what’s already in the game (such as Act 3, more Heroes, new builds, and improved events) rather than total transformations of existing systems.[\u002Fp][p]A direct example of how the demo’s success is informing my approach to new content is Act 3. One of the most common pieces of feedback we’ve heard is that part of what makes Guildrun feel so addictive is how easy it is to hit the “New Run” button. We very deliberately don’t have any out-of-game systems that add friction to pressing the New Run button (such as having players ban Heroes or select bonuses between games), which certainly plays a part in this. However, another major and perhaps even more important factor in why it feels so easy to start a new run is that the runs themselves feel very “punchy” and quick compared to other strategy roguelikes. One of the reasons the runs are so quick is that we simply didn’t have time to create Act 3 for the release of the demo, so the game ended up being shorter than we intended.[\u002Fp][p]Our original assumption was that Act 3 would be the same length as Acts 1 and 2, but this would mean increasing the overall length of each full run by 50%. A 50% increase in game length sure feels like something that would mess with an aspect of Guildrun that was really working for players, which has caused me to reconsider just how long Act 3 wants to be. Though we will obviously playtest, gather feedback, and iterate on the length of Act 3 before releasing the full game, my current target for the overall increase in game length from Act 3 is much closer to 25% than 50%, thanks in no small part to this lesson from the demo.\n[\u002Fp][p][b]Lesson #2: Number up good, number down bad[\u002Fb][\u002Fp][p]Perhaps the most obvious trend in our internal data was that people really liked effects that made numbers go up, and didn’t vibe so much with effects that made numbers go down. I was acutely aware that players would prefer this coming into the demo, a big part of the fun in roguelike games is making builds that generate bigger and bigger numbers, but I never expected the overwhelming extent to which this trend would show up in our data.[\u002Fp][p]Let’s take a quick look at our overall Hero popularity data from the demo, which shows all the Heroes in the game sorted by the total number of fights they participated in across all game modes and difficulties. Though many things factor into why someone would pick one Hero over another (including the Hero’s visual design, theme, and perceived power level), it’s hard to ignore a few very clear trends in overall Hero popularity that likely stem from the Heroes’ mechanics and keywords:[\u002Fp][table equalcells=\"1\" colwidth=\",,,\"][tr][td][p align=\"center\"][b]Popularity Rank[\u002Fb][\u002Fp][\u002Ftd][td][p align=\"center\"][b]Name[\u002Fb][\u002Fp][\u002Ftd][td][p align=\"center\"][b]Popularity Rank[\u002Fb][\u002Fp][\u002Ftd][td][p align=\"center\"][b]Name[\u002Fb][\u002Fp][\u002Ftd][\u002Ftr][tr][td][p]1[\u002Fp][\u002Ftd][td][p]Dragomir[\u002Fp][\u002Ftd][td][p]14[\u002Fp][\u002Ftd][td][p]Grace[\u002Fp][\u002Ftd][\u002Ftr][tr][td][p]2[\u002Fp][\u002Ftd][td][p]Tilly[\u002Fp][\u002Ftd][td][p]15[\u002Fp][\u002Ftd][td][p]Funke[\u002Fp][\u002Ftd][\u002Ftr][tr][td][p]3[\u002Fp][\u002Ftd][td][p]Rowan[\u002Fp][\u002Ftd][td][p]16[\u002Fp][\u002Ftd][td][p]Gustav[\u002Fp][\u002Ftd][\u002Ftr][tr][td][p]4[\u002Fp][\u002Ftd][td][p]Pimenta[\u002Fp][\u002Ftd][td][p]17[\u002Fp][\u002Ftd][td][p]Ming[\u002Fp][\u002Ftd][\u002Ftr][tr][td][p]5[\u002Fp][\u002Ftd][td][p]Nyx[\u002Fp][\u002Ftd][td][p]18[\u002Fp][\u002Ftd][td][p]Pollen[\u002Fp][\u002Ftd][\u002Ftr][tr][td][p]6[\u002Fp][\u002Ftd][td][p]Zuri[\u002Fp][\u002Ftd][td][p]19[\u002Fp][\u002Ftd][td][p]Reyna[\u002Fp][\u002Ftd][\u002Ftr][tr][td][p]7[\u002Fp][\u002Ftd][td][p]Rip[\u002Fp][\u002Ftd][td][p]20[\u002Fp][\u002Ftd][td][p]Logan[\u002Fp][\u002Ftd][\u002Ftr][tr][td][p]8[\u002Fp][\u002Ftd][td][p]Kai[\u002Fp][\u002Ftd][td][p]21[\u002Fp][\u002Ftd][td][p]Ratna[\u002Fp][\u002Ftd][\u002Ftr][tr][td][p]9[\u002Fp][\u002Ftd][td][p]Yuuna[\u002Fp][\u002Ftd][td][p]22[\u002Fp][\u002Ftd][td][p]Sal[\u002Fp][\u002Ftd][\u002Ftr][tr][td][p]10[\u002Fp][\u002Ftd][td][p]Irini[\u002Fp][\u002Ftd][td][p]23[\u002Fp][\u002Ftd][td][p]Fiona[\u002Fp][\u002Ftd][\u002Ftr][tr][td][p]11[\u002Fp][\u002Ftd][td][p]Hoyoung[\u002Fp][\u002Ftd][td][p]24[\u002Fp][\u002Ftd][td][p]Karsu[\u002Fp][\u002Ftd][\u002Ftr][tr][td][p]12[\u002Fp][\u002Ftd][td][p]Niklas[\u002Fp][\u002Ftd][td][p]25[\u002Fp][\u002Ftd][td][p]Aria[\u002Fp][\u002Ftd][\u002Ftr][tr][td][p]13[\u002Fp][\u002Ftd][td][p]Skorn[\u002Fp][\u002Ftd][td][p][\u002Fp][\u002Ftd][td][p][\u002Fp][\u002Ftd][\u002Ftr][\u002Ftable][p]Our two most popular Heroes, Dragomir and Tilly, are the only two Heroes whose signature abilities generate Shards at C-Rank. These Heroes also happen to be fan favorites for their visuals, but Shard-related specializations, relics, and items almost always ranked among the most popular choices in their categories.[\u002Fp][p]The data also revealed that players loved permanent scaling effects. Interestingly, the three Heroes with permascaling on their C-Rank abilities don’t exactly tell that story: Zuri, Logan, and Ratna came in as our 6th, 20th, and 21st most popular Heroes overall. However, almost any effect that offered permanent scaling ranked among the most popular in its category. Of the 75 Hero specializations in the game, 9 of the 15 with the word “permanent” in them are in the top 20 most popular overall, joined by 3 more specs that offered permanent scaling but used the words “this game” instead. These specs were also usually the Hero’s most popular spec by a wide margin.[\u002Fp][p]Conversely, effects that reduced enemy stats instead of buffing Hero stats were much less popular than I would have predicted. Frost is all about reducing enemy stats, and Guildrun’s three Frost-based Heroes (Gustav, Sal, and Karsu) are the 16th, 22nd, and 24th most popular Heroes out of 25. The pick rates for Frost-related items and relics tell a similar story, and it was abundantly clear from the demo that Frost excited players much less than Poison and Burn did. There were also a number of Warrior-related effects that reduced enemy Defense, and these effects were similarly unpopular within their categories.[\u002Fp][p]Most importantly, these “number go up” and “number go down” popularity trends persisted throughout the entire demo regardless of the effects’ actual power level. Tilly and Dragomir have never not been in the top three most popular Heroes of a patch, despite regular periods where their power level was far from the top. Rowan’s “The Friend” has been his worst-performing spec of the three in nearly every patch, but has always been his most-picked by a mile. Sal and Karsu have consistently been among the top five performing Heroes patch over patch, yet they have steadily maintained a low play rate. When the power data and the popularity data tell opposite stories, that’s when you can be sure you have a strong signal that players either enjoy or don’t enjoy a specific mechanic.[\u002Fp][p]So, given the relative popularity of Shards and scaling, what’s the plan for 1.0? When it comes to Shards, it’s clear that we should support more Shard-based gameplay and builds (while keeping things balanced, of course). I’m planning to expand the number of classes that have Shard generation as part of their identity, and to create more effects that let you spend Shards outside of the shop. As for scaling effects, I’ve already begun putting this data into practice as I design new Heroes and redesign old ones. The current plan is to give as many Heroes as possible one permascaling option among their three specialization choices.[\u002Fp][p]When it comes to Frost and other stat-reducing effects, the data is clear that we need to be doing something more exciting. Even though the overwhelming consensus among top Red Rift players was that Frost was the most powerful of the three debuffs, our goal is to make things that are fun. Making powerful stuff is easy; making fun stuff is the challenge. I’m still working out exactly what changes to make to Frost, but one thing I know for sure is that we could do a better job of creating exciting hooks to pull players into Frost-based comps, and of offering cooler rewards for doing so. I’m also planning to cut all of the Defense-reduction effects from Warrior and replace them with Frost effects, since Frost reduces Defense anyway and the Warrior class could use more keywords to help it hook into a wider variety of builds.\n[\u002Fp][p][b]Lesson #3: Devil in the details[\u002Fb][\u002Fp][p]One of the biggest challenges in designing an autobattler is that every piece of content in the game is highly interconnected. Every Hero, spec, item, and class modifier can appear in any run and be combined with anything else to form tons of different builds. This is a big part of what makes autobattlers so fun, but it’s also why it’s so important to nail the balance between those builds. If just one build is significantly more powerful than the others (especially when you have leaderboards like we do), what was once a dynamic drafting game that rewards flexibility and creativity can warp into a minigame of “How can I most effectively force this one build?”.[\u002Fp][p]As a hardcore fan of autobattlers dating all the way back to the Dota Auto Chess days, I was very aware of all this when I started designing Guildrun. Consequently, I spent a ton of time in many different spreadsheets meticulously curating the numbers behind every system in the game, doing my best to make as many Heroes and builds as viable and balanced as possible. At each item and relic rarity level, every keyword needed to appear the same number of times, provide inputs and outputs, hook into other keywords, link into the game’s classes the correct number of times, and so on.[\u002Fp][p]Not to toot my own horn, but I’m confident that all of the extra hours I spent on this largely invisible kind of design work paid off and can be felt in the demo’s gameplay. The reason I’m so confident is because every system where I [i]didn’t [\u002Fi]apply this same level of obsessive tweaking and tuning came back to bite me in some way during the demo. There were a few major areas of the game where I neglected to maintain this balance, but the one I want to highlight here is the balance of keywords across Hero specializations.[\u002Fp][p]Simply put, certain keywords (Burn, Stall, etc.) appeared more frequently than their counterparts across the Hero specializations in the game. In some cases, much more frequently. The most glaring example was Shields, which appeared more than twice as often in Hero specs as Poison did. Each keyword appears an equal number of times in the items, relics, and rank modifiers, but in arguably the most important category of all (specializations), I let the balance of keywords slip through the cracks! It should come as no surprise that Shields ended up being more powerful and playing a much larger role in the demo than intended. Conversely, underserved keywords like Poison ended up being less powerful and playing a smaller role than I would have liked.[\u002Fp][p]I’ve been putting this lesson into practice as I design the new Heroes and rework some of the older ones. At present, Stall and Rush each appear exactly 18 times across the Hero specs, and all of the other major keywords appear exactly 12 times. Thanks to some reshuffling of the old Heroes and careful planning of the new Heroes’ classes, there is now also an exactly equal distribution of primary classes, secondary classes, and class combinations in the game. Every combination (such as Warrior + Duelist) appears exactly twice, and every class appears six times as a primary class and five times as a secondary class. It took a lot of spreadsheeting to get the keyword and class balances exactly where they needed to be, but one of the big lessons the demo has taught me is that this effort will be well spent.[\u002Fp]","https:\u002F\u002Fsteamstore-a.akamaihd.net\u002Fnews\u002Fexternalpost\u002Fsteam_community_announcements\u002F1849827110428865","2026-11-22T12:25:01.261253+00:00"]