Scenario

We have a soon to launch app which includes a rewards system. These rewards are real money, so a direct Rewards Balance: $xxxx should suffice (in theory). Basically put, rewards are a real payment for some task completion

However, our first tests shows that there are 2 parts to the story. Since these rewards may decrease there are 3 scenarios: Balance grows, balance shrinks, Balance stays the same

In the first scenario, tests obviously shown users have a very good reception for the amount shown in $.

However, on the second scenario (balance shrinks) the friction went to the roof, based on loss aversion bias. In turn, this lead to test groups become static and stop doing any tasks at all.

For third scenario, we didn't find negative signs (at least nothing important) in relation to the $ amount shown, but we found a curious threshold pattern: after some sum, they froze, which means loss aversion started to play even before decreasing.

Proposed Solution

So we decided to go with points. So far, test users didn't react well or bad, they seemed quite uninterested. The task completion was more or less on the above mentioned threshold pattern, maybe slightly higher, but this happened for all 3 cases: growing, shrinking, stalling. On the brighter side, this allows us for a marginal profit

Current conclusions

While showing money amounts has a lot of friction, it has more tasks performed until that friction stage. Points have a lot less friction, but they're not enticing enough to make users perform more tasks.

Question(s)

Since we're close to pack this up, times are really tight for more testing, so my question is: is there any study showing that either points method or $ method is definitely better?

Assuming points is the way to go, is there a way that works better than others to avoid the above mentioned problems? For example: 1 point = 1$ OR 10 points = 1$ OR 1 point =$10

or ranges that may lead to less money for users but a higher completion rate since they will need to get to the next step for a higher sum. Example: 50 to 99 points achieved = $50 . 100 to 199 points achieved = $100

EDIT: Just in case it is of help for the answer, sums may go from a few dollars (well, another currency which has more figures, but for the sake of example) to 5 figures in extreme cases. Not likely, but possible. However, most amounts will be on the 3-4 figures range

EDIT 2: This is not a gamification model. This is real money for real tasks performed. Due to very strict NDA, I can't provide more info, but as a totally unrelated example: let's say the client wants users to find yellow, blue or red flowers with 5 to 8 petals. So many users will send flowers. The client will want only pne yellow flower with 5 petals, one with 6, one with 7, one with 8. Same for red and blue. Everything else will be rejected.

But then we'll find there are many users sending yellow flowers with 5 petals, many users sending yellow flowers with 6 petals and so on. We have an algorithm in place for this , but for the sake of example, let's say client will approve only the first flower of each type.

Due to this huge waiting time, we need to reflect tasks were performed and as long as the specs have been followed, temporarily approved, or users won't have any way to follow their progress and see the result of their efforts.

So we're using temporary figures, and clearly specify these figures are subject to final review, so we have no problem with the process itself, as most answers are pointing out, the problem is with the figures: money amount or points amount, with the problems we've found for each one