When testing offers to see which one work, should I bid high or low?
For example, if Propeller suggest bid $4.5, but I will still get traffics for $0.6, which one should I bid?
Should I bid high, find the working offer, then lower the price, or should I bid low, find the working offer, then increase the price to scale?
Why?
Thank you.
So for me its always good to start small, I bid low to the point I'm sure wouldnt cost me if traffic isn't converting, if you can get traffic at $0.65 , I can place my bid above at $1 and watch it, With these I get enough data to optimize and can then bid high to get all available traffic.
Depends on geo, offer type, targeting and payout.
Generally, it's best if you can afford to bid high, learn placements (cashcows and complete crappy ones), and then play with the bid up and down, adjust best on placements, etc.
I always try to hit a point of the bidding curve where the increase in traffic is still exponential. The propeller recommendation is for most campaigns out of the world.

If you cannot afford paying high for the whole campaign, you can try running a basic blacklist and breaking out profitable placements into a whitelist with different bid. You can as well break out different targeting option (carrier) into separate campaigns and match the CPV to the EPV.
The graph that Sebastian posted is a really good one, there is always a point that going below doesn't make sense, because the traffic will stop. And there is also a point at the other end of the scale - when bidding higher will not give you a significant boost in volume, because there simply isn't more traffic for having. You need to oscillate between these two points and find the level that gives you the best results.
There are obviously exceptions, for example, some sources strongly prefer the highest bidder, so you need to be the 1st one in order to get solid volume. But usually it's ok to stay a bit behind.
I would like to ask.
If you decide to change your bid , does this mean that the data you have already collected is no longer valid?
Thank you