How unconventional setting affects the success of a hyper-casual game in an occupied niche

We got a puzzle game for publishing with the familiar yet unconventionally implemented mechanics of pouring water into flasks: first of all, it was completely in 3D, and second, it featured little men instead of paints. We decided to test it.

Color Sort Puz has now crossed the threshold of $1.5 million in total revenue and continues to grow every month.

First metrics

The gameplay is simple: there are several lanes with colored figures, you need to populate these lanes with little men of the same color. The prototype was single-handedly developed and submitted to us by Sarthak Madan, CEO at Pizzart. We started testing it right away and got the first metrics:

  • CPI US — $0.5
  • СPI WW — $0.05
  • R1 — 27% WW / 29% US
  • R7 — 9% WW / 12% US
  • R30 — 3% WW / 3,6% US
  • Playtime — 15 minutes daily
  • Ad views per player by Day 7: 12 interstitial and 1.5 rewarded

Relatively good CPI, long retention, and good playtime confirmed that the audience is interested, and the mechanics have not yet become boring.

Why we decided to keep testing:

  • The project exhibited good metrics and player interest after the first test.
  • Familiar mechanics at the peak of popularity and non-standard setting.
  • An overall good impression of the product and belief that the metrics can be improved.

What was done

We started with the little men animation and made them move more smoothly from lane to lane (both visually and in terms of optimization).

Smooth animation

It turned out that the lanes were too close to each other and interfered with the animations. We increased the distance, and it revealed a level design problem — there were fewer lanes now, so the levels with a large number of lanes became impossible to implement, and this has affected the players looking for a challenge. This is how one change opened up other game design problems.

That’s why we began to balance the levels and the location of the little men. Subtle at first glance, these improvements allow to significantly increase retention and playtime.

One of the first iterations

We tested a lot of different difficulty curves. Sometimes the outcome was positive, sometimes it wasn’t, and we had to roll back the changes. As a result, we settled on a funnel where the complexity increases by levels but has peaks instead of being linear.

As part of working on the balance, we had dozens of experiments with the number of figures on the screen. Larger men in one line performed best, so we made it less crowded:


It barely affected the retention and playtime, but the number of ad views increased by 15-20%. Later, we tried to bring the crowd back to make the visuals more familiar to hyper-casual players. Those iterations weren’t necessarily successful, but we still left the opportunity to choose the number of figures in the settings.

The animation we’re currently using

When we added sounds to tapping the lanes, the metrics also increased — first, the retention and playtime, and then ad views. We’re currently planning to add background music and update the lane sounds to blend in with the background.

We made a store where players could get “cosmetics” for watching rewarded ads, as well as make an in-app purchase to turn off ads. As a result, rewarded ad impressions grew by 30% (on average, there were five views per user by Day 7), and retention didn’t decrease.


Now we regularly release content for the store and balance item prices. One of the interesting changes we introduced was moving color scheme options for the game from the settings to the store. We plan to remove the unpopular options in the future.

Separately, we were working on monetization and ads — experimented with interstitial cooldowns, monitoring access to the Internet, and restricting access to the game without the Internet connection. We also tried offering hints for viewing rewarded ads. Previously, there was a hint to add a lane and endless move undos. Now we give 2 free undos in total for each level, and then +3 for watching rewarded ads.

Release and plans

Approximately six months later, we started scaling the project. The metrics at the time were as follows:

  • R1 — 37% WW / 42% US
  • R7 — 10% WW / 15% US
  • R30 — 7% WW / 8% US
  • Playtime — 16-18 minutes daily
  • Ad views per player by Day 7: 14.5 interstitial and 7 rewarded

The total revenue and spend have already surpassed $1.5 and $1 million respectively. CPI in the US is now about $2, and $0.20 worldwide, while the project has been holding the metrics steadily for a year now. We haven’t hit the ceiling yet, there are many points for further growth and things that can be polished.

Now we continue to balance the levels and try to add new variations of mechanics, quests, and bosses to them. We’ve already tested the bosses — in the form of lanes where the colors of the men at the bottom are hidden — but there’s still some pushback from the old, so we’ll try to add it again later on. UI improvements are also in the plans.

I developed Color Sort Puz as a fun personal game prototype that I wanted to play myself. From it’s early stages, the talented people I worked with at Azur Games recognized the potential of this puzzle game and were determined to bring it to the market with the utmost care and quality. The development process was a collaboration between myself and Azur Games, with regular feedback from the market to make gradual improvements. The result is a polished and highly enjoyable puzzle game that has received a warm response from players. I believe that this is only the beginning for this game and am excited to see where it will go in the future

Sarthak Madan, CEO at Pizzart
Back to blog