Albert Reed is the CEO and cofounder of Demiurge Studios. Follow the entire series here.

In the previous parts, we touched mainly on things that had worked well from an ARPU perspective for Marvel Puzzle Quest. That’s a bit disingenuous, actually. To date, we’ve had many attempts that haven’t moved the needle in the ways we originally hoped. For example, one approach we at Demiurge Studios felt would be fruitful was adding additional items to the game for sale. The logic is strong: Some people out there would spend money on the game, but they just haven’t seen something they perceive as valuable. In Part 5, I’ll go into more detail about both Shields and Boosts and share some of our key learnings.


Versus events in Marvel Puzzle Quest run for two to three days. At the end of the tournament, we give out prizes on how well you ranked compared to others in the game. Like many games, we award points based on Elo. To prevent players who hyper-engage with the game from running away with the win, we keep the scoring roughly zero-sum, and under Elo, highly rated players are worth more points than lower-rated ones. One side-effect of this system is that players who are ranking well get pummeled, and there is a race-to-the-finish that happens in the final hour of a tournament. That’s not great for gameplay and forces our players to schedule their lives around the game to be competitive. To combat that, we’ve long wanted a mechanic that enables players to protect themselves from attacks.

[Left] The UI for purchasing shields.. [Right] Players who lose in Versus are breadcrumbed into the Shield mechanic.

Above: [Left] The UI for purchasing shields.. [Right] Players who lose in Versus are breadcrumbed into the Shield mechanic.

Image Credit: Demiurge Studios
One nice side-effect of being so generous with our hard currency is that we can use it for gameplay while keeping the game free-to-win. Players can save up their Hero Points and exchange them for an easier victory (and fabulous prizes!) in a versus tournament if that’s what they choose to do.

Over the past 30 days, shields have been 1.8 percent of the Hero Point economy. At the time of launch, we didn’t have A/B testing functionality, so we did some crude experiments when they were rolled out. In the first four versus tournaments, they were priced at 50/100/200 hero points (with no volume discount Hero Points are $0.01 but they discount down to $0.005 when purchased in bulk), and for the next events we doubled the price of shields to 100/200/400. We sold almost exactly half as many per participant.

We think Shields are a fun gameplay feature, and we want as many people to engage with the mechanic as is practical, so we settled on the lower price.


The state-of-the-art in free-to-play is increasingly focused on a combination of consumables and very deep durables hooked into the economy through speed-ups. You can see this most clearly in Boom Beach as compared to Clash of Clans; publisher Supercell is clearly moving away from durables that allow players to make a fixed purchase (builders) to save time and asking them to spend money per minute saved instead.

Marvel PQP5 BoostPurchaseUI

Above: The Boost Purchase UI.

Prior to a fight in Marvel Puzzle Quest, players can draw from an inventory of boosts that help them get through particularly tough fights. We opted for this approach rather than the “saving throw” that’s in Candy Crush Saga and Puzzles and Dragons because we felt this is a more fun mechanic — players are wagering that that they can beat a fight with a little bit of help. Boosts started out being sold only for soft currency. Late last year, we began experimenting with boosts sold for hard currency with mixed results. These days, they make up 2.1 percent of our hero point spend — a bit better than Shields.

Our existing boosts offered players bonus damage to particular subsets of enemies, so we started by trying to distribute boosts that were more effective than the ones that were already in the game. These boosts were tied into a specific event that was running at the time. As with everything in Marvel Puzzle Quest, we give players ways to earn the consumable as well as offering them for sale. At their best, these worked out to around 1.5 percent of daily hard currency spend.

We theorized that these boosts might be not successful because they were all-or-nothing. Players who used their boosts (regardless of if they earned the boosts or bought it) found a fight easy. That’s a great carrot to offer as a reward, but we had to be stingy with them and price them high enough that most folks wouldn’t incorporate them into their regular play. To correct that, we “borrowed” again from Zynga’s Gems with Friends and made it possible for players to stack the same boost multiple times. We also theorized that the existing boost UI was too noisy and intimidating, so we stopped offering several of the old-style boosts.

Stackable boosts.

Above: Stackable boosts.

Image Credit: Demiurge Studios

Stackable boosts yielded some very interesting results in terms of pricing and positioning of the upgrades. One of our hard currency boosts increased the starting “AP” of a team. This has massive gameplay advantages, but it’s not as easily comprehensible. Implementing stacking for these boosts actually dropped the total Hero Points being spent on them by a significant margin.

The other hard currency boost was simpler to understand — increased damage from basic attacks. The same change to these boosts, increasing the granularity and control players had in upgrading them. When stacking was implemented for these boosts, the cost

Another theme that we’ve explored in previous posts is the impact tap friction can have on unit volume. Enabling players to purchase 10 times the cards at 10 times the price resulted in a massive increase in how many hero points were spent on covers. In investigating prices of boosts, we also discovered that packs of things that sold for 50 hero points ($0.25 to $0.50, depending on bulk discounts) sold nearly the same unit volume as those sold for 100 hero points with twice the efficacy.

One highly unexpected side effect of stacking was the amount of total boost our players were taking advantage of in each fight. They can stack the match damage boost five times, but most players weren’t actually taking advantage of this.

Stack size for match damage boosts

Above: Stack size for match damage boosts

Image Credit: Demiurge Studios

This graph was surprising and led us to believe that the UI was confusing or simple tap-friction was causing poor performance. Our next improvement is going to be default to a full boost stack.

After these changes, most of our revenue still comes from sales of cover packs. We’ve seen large swings in the economic performance of our consumables as we adjust pricing, but we think the largest improvements are going to come from updating our consumables to provide greater value to the player. More on that as we update the game!

In the next part, we’ll take a closer look at the launch of Alliances, the biggest feature we’ve added since launching Marvel Puzzle Quest six months ago.

Demiurge CEO Albert Reed.

As cofounder and CEO, Albert Reed has held the reins for over a decade at Demiurge Studios, a game development house focused on innovative, high-quality games for all platforms. Find him on Twitter at @almnop.