What's the deal with bidding? (8)
Haven't seen this talk about much but I'm curious how this works.
I have been randomly placing bids with no strategy and get the feeling I'm doing it wrong.
I've noticed that some traffic sources such as propeller ads give you a recommended bid or a chart that displays the average bid, the highest bid and your current place in the bids for your traffic (popads). For example popads will say something like average bid for your traffic is 0.001, highest bid 0.026 and my bidding queue 18.
Obviously we want good traffic. To be at the top it would cost $26 per 1000 popunders which is a insane cost for popunder traffic. Especially if you're dealing with payout offers that are in the $2 range.
Is there some kind of formula that is used when launching a camp on a traffic source to determine how much you should bid? Or do you just randomly place a CPM/CPC bid for testing and hope for the best?
http://stmforum.com/forum/showthread...hlight=mapping
http://stmforum.com/forum/showthread...-The-Landscape
http://stmforum.com/forum/showthread...ing-Techniques
http://stmforum.com/forum/showthread...he-Competition
When in doubt read anything written by caurmen the man turns words into gold. Literally.
Definitely read caurmen's posts like the guys recommended above!
I like to bid average or higher when testing offers, so that I KNOW the traffic isn't crap. When you're testing something, you need to make sure the other variables are proven or you'd get into a catch 22 situation, where you don't know which variable is the problem when you're not getting conversions (or enough of them).
But once I've found a good offer+lander+targeting combo I would test different bids to see which one works better - either by cloning the orig camp and assigning different bids to see which one does the most profits, or just vary the bid on the orig camp every 1-2 days to see which bid works better.
And if you want to play with placements - there are a lot of routes you can take. Some examples:
-Set a higher-bid camp and a lower-bid camp in additional to your original camp. Obviously cut placements in the orig. camp that are hopeless. Then you can put good placements into the higher-bid camp to get more traffic (and hopefully more profits - if not you can put it back in the original camp), and put marginally-profitable placements into the lower-bid camp.
-Set up multiple camps with different bids, and cut placements for each. Can also compare the same placements across different camps to see which bid is best for each placement. (This approach could get expensive though.)
-In addition to the original camp, set up a higher-bid camp, with all placements that are getting good traffic volume in the original camp blacklisted. This way the higher-bid camp will trigger better-quality placements that your competition is bidding higher for. Can repeat with another, even higher-bid camp, and blacklist all placements getting good traffic volume in the lower-bid camps.
If you're running pop traffic though, it probably won't be worth it to bother messing with placements like this because pop camps die quickly. However, if you're planning on continuing to run in a certain geo again and again (for example, whenever new offers come out), taking the time to sort placements using one of the approaches above could really give you an edge over an affiliate who's new to the geo and having to start testing from scratch.
Good luck with testing! 
Amy
In an ideal world you want to bid per domain/placement based on EPV and market rate.
Means, a bid that
(i) is competitive and provides you an good amount of traffic of an certain placement
(ii) provides you an positive ROI / is in line with your target ROI.
Since that is not feasible in most cases, EPV baskets are a good second best.
Once you have some data, you create several campaigns with placements that have about the same EPV.
For those baskets, set your bid based on market rate and your EPV.
For a new camp, you can start with and medium bid to get a first picture of the market environment. Then proceed as described above.
To unlock placements which are not sending traffic with your medium bid, you can create a new blacklist camp with every single placement that is sending good traffic (medium bid) so far blacklisted and set a high bid to unlock the hidden placements.
All of that, only if you have a winner. No need to polish a turd.
Somehow I missed those caurmen threads. Kudos for pointing those out. After going through those threads. It looks like it's about optimizing your bidding AFTER you have seen you have a winning campaign on your hands. I'm talking about bidding for the testing phase of a new camp to see if its a winner.
I have no clue if I'm bidding a respectable amount to get data and just throwing money away.
@vortex, @sebastian_r you both mention a "medium" bid. How do you determine what a medium bid is on a traffic source?
Here's what's going on in my head. Maybe you can point out the issue I'm having.
Lets say I have 1 offer with payout of $2.00 with 5 landers. If we do the offer x 3 for every landing page. We have a testing budget of $30 to see if this camp has potential. I go to a traffic source like popads. For my targeting I see the average bid is $2.80 CPM with the highest being $30.
I can bid $2.81 to be slightly above average. I estimate I would get 10.6k traffic for this price with a $30 budget. But if I was to increase this to a $5 CPM my traffic would reduce to 6k traffic (maybe higher quality) with the same budget. So I would be paying more but with less impressions for data.
When testing a camp for a winner. Where is that line between how much data you should collect with the amount you should be paying for traffic? So with $30 to test a camp. Is there a certain amount of traffic I should receive before moving on? $30 I should receive 10k, 15k, 20k impressions?
Hopefully I'm asking the right question. This part just has me stumped right now.
I'm working with mobile pops/redirects.
Originally Posted by gameandwatch
Here's what's going on in my head. Maybe you can point out the issue I'm having.
Lets say I have 1 offer with payout of $2.00 with 5 landers. If we do the offer x 3 for every landing page. We have a testing budget of $30 to see if this camp has potential. I go to a traffic source like popads. For my targeting I see the average bid is $2.80 CPM with the highest being $30.
I can bid $2.81 to be slightly above average. I estimate I would get 10.6k traffic for this price with a $30 budget. But if I was to increase this to a $5 CPM my traffic would reduce to 6k traffic (maybe higher quality) with the same budget. So I would be paying more but with less impressions for data.
When testing a camp for a winner. Where is that line between how much data you should collect with the amount you should be paying for traffic? So with $30 to test a camp. Is there a certain amount of traffic I should receive before moving on? $30 I should receive 10k, 15k, 20k impressions?
One thing I could say for sure, is that I would not set a rule of thumb based on the number of impressions. There are other metrics that are way more important. Traffic cost and conversion rate of the offer+lander for example.
Like you've pointed out, trying to determine what the "medium" bid is, is not always straight-forward. The point here is that when you're testing unproven landers and offers, you NEED to ensure traffic quality (and including lots of landers will help as well, to maximize your chances that SOME of the landers will work well; and also only using offers recommended by AMs at the start will give the best chance of having at least one that will convert well enough to help cut landers down to a winner). If you bid too low and end up with all the junk traffic, you can't test your offers + landers adequately, i.e. if performance is bad you won't know whether to point fingers at the traffic or your offers/landers.
So basically just avoid bidding too low in general instead of fixating on finding the "medium bid". Some traffic sources have bidding charts or graphs. Zeropark, popads, and propeller for example all have them. ZP and popads will tell you what the average bid is. For propeller I try to bid higher than the first "bend" in the graph to make sure I'm not the lowest bidder. For sources that don't show bid data/suggestions, I would just estimate using the sources above.
Then, when you're running the camp, keep in mind that once you find the best offer and lander and targeting, you can potentially decrease the bid to increase profits. But that would be the second step. The first step would be to bid high enough to give the offers and landers a chance to convert, by ensuring that traffic quality is there.
As for setting a budget - I don't have a one-size-fits-all approach. I would often run the first test campaign at a loss just to test a lot of landers to find a good one, so that I could mass-test offers in the second round of testing. And since I wouldn't know what the CR is for each offer and lander combo, or at exactly which point statistical significance would be reached for each lander to meet cutting criterion, there'd be no way to decide on a budget from the start.
In response to your question "Where is that line between how much data you should collect with the amount you should be paying for traffic? So with $30 to test a camp. Is there a certain amount of traffic I should receive before moving on?" As mentioned above, I wouldn't look at impressions. I would either look at profits potential or check for statistical significance, depending on what my aim is for that particular round of testing
If you do what I do, by testing lots of landers in the first round to find a good one, then basically you'd want to keep running and cutting landers as they reach stat sig until you're left with a winner. The aim of the first test would be to find a winning lander, not end up with a winning campaign, so unless the test is in such a serious loss that you'd rather invest the money on testing another geo/vertical/etc., budget shouldn't be the main focus here.
However, say you already have either a proven lander and are mass-testing offers, or have a proven offer and are mass-testing landers: In that case I would definitely assess profits potential. This is when I would drill down into Offer > Lander > [every major traffic segment] (major traffic segment = one that has enough traffic that you could make enough profits by targeting it exclusively, for the camp to be worth your time running it) to see if I see any green. In this particular case (i.e. when I have a proven offer or lander) I would keep testing and cutting, and would only abandon the camp if I've spent 2 or more payouts on each offer+lander combo and no longer see green on any remaining combo.
Hope that helps! The main point I'm trying to make is that the number of impressions is not a consideration here, but either profits potential or how much it's costing you to cut stuff as it reaches stat sig. Sample size is automatically taken into consideration when you're using stats calculators so you don't need to worry about impressions.
Amy
Home > General > Affiliate Marketing Forum