Mobile POP - How to calculate budget to test campaign correctly? (7)
Mobile POP - How to calculate budget to test campaign correctly?
Hello all,
After reading a lot of threads, I have question regarding how to calculate budget to test campaign correctly. There are most common scenario where people say test it with x3-x10 payout of offers, but that's to vague to me. I would like to get some more info regarding this, what would be good budget to test campaign when you take everything in consideration (number of offers, landing pages etc.)
I have found Amy's calculation where she said if her campaign has 3 offers, 5 LP's and Avg payout of $2, the formula would be 3x5x$2x2=$60 (worth of traffic).
So by my calculations, if we buy traffic worth of $60 and for example let's say CPM is $1, we will get 60.000 views. Let's say that average CTR of our landing pages is 5%, that means offers will get only 3.000 clicks and if we have rotation od 33%, every offer will get 1.000 clicks or $1 of traffic. If offer has payout of $1 then we tested this offer with only x1 payout.
Can we say that this campaign is tested correctly?
Thank you for answers! 
For 3 low payout offers and only 1 LP, you don't need that much traffic, if you go by AMy's calculation, which is a solid baseline ... 3x1x1x2 = $6 should be enough, or in other words 2x offer payout ... this is the most frequently suggested rule of thumb to go by, 2x offer payout per funnel variation.
Hi Sachin! Yes your calculation is correct. But there are a couple of important considerations I can point out to clarify things further.
First of all, just because "only" 1000 clicks "makes it" to each offer, and the cpm is $1, doesn't mean we're testing the offer with half a payout.
Every campaign decision we make, should be aimed at maximizing profits - or minimizing losses in this case, by not spending too much on testing, should the offers turn out to be duds.
From the cost viewpoint, we are paying $60 to test 3 offers. So $20 was spent testing each offer. If a $2 payout offer doesn't convert on a $20 spend, then it would be less likely to end up profitable. (It wouldn't be IMPOSSIBLE, but if we want to make 100% certain of that we'd go broke in the testing phase. Would be difficult to recoup testing costs with any profitable campaigns later on.)
The second point I need to make, is that the previous point will only be valid IF the landers are reasonably good. This is probably what's tripping you up as well.
If the landers you're using are considerably WORSE than the winners you'll ultimately be using, then the results would not be a fair judge of the offers' performance.
Let's say that that the average popular lander available on Adplexity - for that particular vertical and geo and other targeting - has the potential of helping your best offer convert at x%, but the landers you're using can only convert your best offer at a small fraction of that, then the inferior landers you're using are dwarfing the potential of your offers.
Also: We can apply the same reasoning to landers as well - if the offers suck, then we won't get conversions even if we were using the best landers in the world.
This is why in the beginning, it's so crucial to make sure that:
1)You're using popular landers from Adplexity that many people have run good volume to, that you have thoroughly tested and optimized to make sure they function correctly, display correctly on major devices, and load fast.
2)You're using offers that have shown good track record. Either a positive trend in a spy tool, or recommended by your AM as having converted well for at least a few other affiliates.
3)You're using for testing, traffic that is of good quality. This means using a traffic source you've had the most success with thus far, and bidding at least average to avoid getting crap traffic.
4)You're testing multiple landers and offers to increase the probability that at least some of them are decent performers!
A good summary: In the beginning when you're breaking into a new vertical and geo, you're faced with 3 major variables that are all unknown: Offers, landers, and traffic, and if one of them fails you won't see good results. So what you need to do, is "sure up" those variables with the 4 points listed above. This will establish a benchmark based on which you can do further testing to improve the campaign - by testing more offers and/or more landers and/or more traffic networks.
Once you have some benchmarks it would be easy to test other stuff. E.g. when you have a winning lander, you can use that to test more offers. And when you have a winning offer, you can use that to test more landers.
That was a long explanation, but hopefully it helped to clarify some doubts!
Amy
Sent from my SM-G930W8 using STM Forums mobile app
Thank you so much matuloo and Amy for answers, it definitely helped me to clarify my doubts!
So if I use most popular landing pages from Adplexity and recommended offers from AM then if I use your budget calculation (number of offers X number of landing pages X avg payout X2) should be more then enough data to conclude if this campaign has chance of profit?
Originally Posted by sachin_roy
So if I use most popular landing pages from Adplexity and recommended offers from AM then if I use your budget calculation (number of offers X number of landing pages X avg payout X2) should be more then enough data to conclude if this campaign has chance of profit?
But you will likely need to do further testing based on the benchmarks.
In the very least, you should take the winning lander and test more offers.
To clarify further: In the initial testing, as long as you're not losing excessive money, it would be good to keep running to generate enough conversions to help you cut down to a winning lander, just so you can use that to test more offers.
And testing lander variations or your own innovations can help a lot too.
Also, for the initial test, if you're sure you have some solid landers, and still not getting conversions, you can consider replacing the offers from other networks for a retest. It's easier to find good landers than good offers.
Amy
Sent from my SM-G930W8 using STM Forums mobile app
Hi Amy
How do you determine what a good lander is if you aren't getting any conversions? Do you go by the CTR from the lander, since that's the landers main task...to generate the click through?
I'm just starting with pops landers and am testing multiple verticals (sweeps for phones, sweeps for vouchers and AV) in a single geo and so far am seeing significant differences between CTR's in each vertical over 6k impressions in each vertical, with iphone clickwheel type landing pages the worst performers (<0.3% CTR from all 6 landers) with voucher sweeps landers and AV landers the best performer (over 1% CTR) from 6k impressions in each. I think over that number of impressions, it's OK to cut the vertical for iphone sweeps, even though I haven't had an conversions in any of the other verticals as yet?
Originally Posted by bustedflush
Truth is, CTR doesn't mean much - I've seen too many winning landers that have the lowest CTR of the batch. The opposite is true too: I've seen landers that have the highest CTR end up getting cut first.
The purpose of the lander is not to generate the click through, it's to generate the click though that will result in a higher probability of conversion (compared to just sending the visitor directly to the offer).
So many factors can determine CTR, that when you see high CTR, you CANNOT assume it will fetch high CR either - and it's the latter that makes you money.
For example, I can add code to my lander to automatically redirect all visitors to the offer after 2 seconds - CTR would be very high, but CR would be low, because visitors wouldn't be pre-sold properly.
Another example: I can promise the moon on my lander to get high CTR, but if the information on the offer page does not back up my claims, then CR would suffer.
The only time I would even make decisions based on CTR is if it's so low that it would take an impossibly-high CR for the campaign to even break even. e.g. CTR is 1%, payout is $0.50, bid is $1CPM. For every 1000 impressions, only 10 visitors will see our pop ad, and at least 2 of them would need to convert (or 1 in 5) just for us to break even. For pop traffic this just isn't likely to happen.
When you see CTR that low, you can basically eliminate that lander (to double-check, you can drill down into placement stats and other stats to make sure it's not caused by one big bad placement or something similar).
I lied about that being the ONLY time I'd make decisions based on low CTR. Just thought of something else: The SAME lander(s) is/are going to result in different CTRs on different placements. If the majority of placements have CTRs that are in a certain range, and you see placements that have CTRs WAY outside that range, then those placements probably have issues of some sort (e.g. bot traffic) and should be blacklisted. For example, if for the same set of landers, most of the placements have CTRs of 50%+, then you see a few placements that have CTRs of <10%, then you can probably blacklist the latter. This spreadsheet may shed more light on this:
https://stmforum.com/forum/showthrea...Rules-of-Thumb
Other than these scenarios, CTR doesn't mean that much. Campaign-optimization decisions should mostly be made based on profit amount (which is partially driven by CR, not CTR).
If sweeps aren't running being run very much there, it may mean that there are offers available in that geo that are easier to make money with, where profit margins are high so your competition can afford to bid higher than you.
If you DO see a lot of traffic in that geo being run to quite a few different sweeps offers, then chances are it would just be a matter of testing more. (Although it IS possible that the large traffic volumes are being run by the offer owner/advertiser themselves, if you see quite a variety of sweeps offers then chances are some of those may be affiliates.) Test more landers and offers (also consider testing offers you see on Adplexity that exhibit good trends - uptrending graph, good volume over a few days, etc.). Make sure your landers are optimized for speed and display, and function correctly. Test until you find a lander+offer combination that is turning a good part of your total traffic green (e.g. resulting in enough profitable placements, to generate good daily profits), then cut the unprofitable parts until the campaign is in profit overall.
Hope that helps!
Amy
Home > The Newbie Zone > Questions and Answers