Testing out Mobile (Appetiser Method) (24)
Hi guys, I've tried media buying on and off and could never get it to work but I'm determined this time. I've deposited $200 into decisive (can add in more, up to $1k if necessary) and I'm hoping with this follow along, I can optimized to a positive ROI.
Offer Type: PSafe app install, direct linking to google play. AM says it performs well (top 3 since February).
Traffic Source: Decisive, campaign setup as per the Caurmen's instructions
Network: F5 Media
Banners: 8 different ad copies, 4 banner styles (although similar, just moving around buttons, no buttons, big fonts, etc). With that, I made 8 different banners, all 320x50 with 8 different ad copies.
Targeting: Android OS, Brazil. I selected smart CPM at .5 since the payout is .35
I started running around 24 hours ago, here are my stats to date (no conversions as of yet based [even in f5 media]). My angle is just targeting a subfeature of the app and I'm not sure if that's what is causing no conversions even though the description page clearly mentions it.
Is the traffic breakdown typical? I didn't realize mobile vs wifi had such a huge difference. Any help would be appreciated!
I'm also promoting PSafe. Just started testing it yesterday, messed up budgeting and tracking though so I'm restarting today.
I'm also seeing wayyy more wifi traffic than mobile.
BTW, if you want to work together, send over your skype 
The breakdown between WiFi/Carrier and Site/App is determined mainly by what exchanges/networks the traffic source plugs into - which will be a small subset of what is truly available in that country.
In some countries, WiFi at home might be a more used option for mobile devices if carrier charges are relatively expensive.
I think you should initially focus on testing very different creatives as your CTRs are on the low side. Try to exploit psychological pressure points, e.g. fear, curiosity, etc. Iamatilla posted a good list of these recently but I forget which post they were in.
I'm going to pause these and relaunch the campaigns again. I found a few good banners and did a bit more researching so I'll keep the copy but see if radically different designs help out. The only thread recently i can find with regards to iamatilla and angles are these: http://stmforum.com/forum/showthread...EED-YOUR-INPUT and http://iamattila.com/media-buying-10...-cpi-offer.php. Are those what you're referring to @zeno?
If anything, I'm reading more and more and right now I have one angle, and eight different copies for it but in my research process I came up with 8 possible angles. I'm debating if I should take the 8 different angles, find 2 copies of it, mix it with 2-3 radically different banner designs and then run the same campaign setuip until they spend 3x the payout (in this case would be pretty much $1).
http://stmforum.com/forum/showthread...950#post160950
Haha damn it. I went too far back into his post history. I just found the same post and was coming back to post it when you did it already 
In general for mobile app installs what kind of CTR should I be shooting for? I just relaunched some campaigns and my average CTR is doubled (.5%) but I still feel like it's too low. Also no conversions yet and i spent $9 already but I just launched the campaign 20 minutes ago and the daily budget is $16. I ended up testing 2 banner styles and 8 copies for different angles for a total of 16 copies. I'm let it gather 3x payout ($1/ad) then I'll see which angle gets either the most conversions and if there is no conversions, which gets the highest CTR.
I also feel like competition is getting a bit more fierce than usual since I see 3-4 follow alongs with the same offer. Oh well, I consider this to be training.
edit: awesome, less than an hour and got a conversion...had to pay $12 for it but at least I'm getting somewhere..
edit2: sweet, 3 hours later, spent $20 but I got 3 conversions, all on the same exact angle spread across the 4 types of campaigns. If it keeps going like this, my next step will be focusing on that angle and trying other banner styles.
This isn't what you want to hear but... CTR benchmarks are nonsensical. You could have 0.01% CTR on an ad but a $50 payout and still be profitable.
In contrast, for a $0.20 app install you might need a 5% CTR for a specific country for your 10% conversion rate to be profitable at the current competitive CPM. In another country, your CPM might be 1/2 as much, conversion rate 30% higher but your ad CTRs are worse and you can't get above 2.1%.
In Brazil a 2% CTR might be normal for app-placement banner ads. In France a 0.3% CTR might be extraordinary.
CTR on it's own doesn't mean much without knowing the CPM you're paying, offer payout and conversion rates, all of which you can't determine until you test!
So, aim for the stars and just be mindful of your metrics. If you launch in a new country, you should pretty quickly be able to look at CPM and current CTRs > eCPC > look at offer payout > do I need a batshit crazy CVR to break even? If so, your ad CTR could likely improve by leaps and bounds, or your angle.
Damn, I figured as much, just was curious if there was just a standard number to see how off I would be. Although, like you said it probably doesn't matter if I back out. Anyways, since relaunching on Sunday with the 8 ad copies duplicated with 2 banner styles (16 unique banners), my stats are as below for the last two days
I just noticed since I didn't use a tracker, it's hard to see which angle out of all the campaign does well. I might use
All 16 banners
Grouped by Design
Grouped by Angle
Not sure if I have enough data to move on with angle 7 or should I try another batch but focus on angles 5, 8 and 7 (different ad copies, same banner style). Or continue with the appetiser method and start cutting out targets/placements/sites/apps. Those two angles are sorta loosely related in the sense that they do not really focus on the app itself. Thoughts?
edit: Holy crap. Just looked at my reports and noticed for wifi site and mobile site, 1 placement was eating 75-80% out of the spend with no conversion so I just paused that real quick. Hopefully the CPA will creep down a bit more. I also wanted to ask, if I make a new campaign, as long as I am testing the same angles/banners, is it safe to apply the same blocks I placed right now? Can I assume that for the rest of the time I'm promoting this specific offer (as in, would I start fresh if I wanted to test new angles)?
I'm not sure I understand your last comment - doesn't the first image show that you did get conversions from the site traffic?
I think you should cull some placements first and get more data - angles 1-4 look pretty miserable but you don't have enough data to make any calls on angles 5-8.
In any case, you need some major placement/banner optimisation within angles 5-8 to get closer to profitability.
If you do relaunch the campaign, then yes - transfer your blacklists across - assume new banners aren't going to turn the dud placements into superstars.
here's an up to date look at my numbers. I'm hoping by doing this follow along I can slowly see progress and also hold myself accountable.
Updated numbers on the angles:
And these are the daily numbers for this campaign
Day 1
Spend: $35.83
Revenue: $2.97
P/L: -32.86
ROI: -91.71%
Day 2
Spend: $23.89
Revenue: $3.69
P/L: -17.87
ROI: -84.8% (ROI increased due to placement exclusion of one site)
I meant that one site (or placement) in the site campaigns were the around 75-80% of my spend and they had no conversions. I'll go through the all four campaigns today and see about cutting some placements and let it run for another day or two. I'm assuming no but if anything I will look at it to see if it's a pattern. if I start cutting some placements (I'm looking for those that are 3x payout with no conversions), should I increase my smartcpm to get more traffic?
Would it be helpful if I start tossing up more banners based off angles 5-8 to see if the stats are similar or better?
edit:
Went through all 4 campaigns and cut out placements spending $1+ and any category that had no conversions.
Yes cull some placements first and get a bit more data.
You are not winning many of the bids so there is clear room to scale both in budget and smartCPM bid.
Get rid of angle 1, 2 and 4.
Hm - I'd recommend slowing down a bit. You're trying to test 8 angles at once, and the side-effect there is that it'll be tricky to tell if it's the angle or the copy that's really working.
At the very least, make sure you spend the full $15 on each angle test. Mobile has a LOT of factors influencing the amount of data you need for statistical significance: make sure that you're cutting placements rigorously and giving your angles enough data that you can distinguish a genuine trend from just random noise. A coin might land heads-up twice, but that doesn't mean it will always do so.
@zeno: I've went through and cut out a lot of the placements that haven't performed, hopefully today will be a better day as well as remove angles 1,2,4 from the mix. I also updated day 2 cause I posted it when it wasn't a full day, ROI went back up (F5 media took back the conversion that I had so the revenue earned went down).
@caurmen: I originally was following your method but noticed the particular angle I chose sucked (1 conversion out of $28 total campaign spend). I was hoping with this one I can see if I can gather enough data to see if a particular angle is doing well, then go back to your method of building 8 copies for that angle and optimize from there. When I do move to next set of copies, do I just continue on the same campaign and add in new banners or would I want to start over? When starting over, I don't know if there's a way for me to transfer placements I blacklisted.
Today looks like it's doing better, ROI is still in the negative but has improved a lot since last two days. I don't expect to make money off this particular campaign but I'm hoping I can narrow down to 1-2 promising angles and also collect a lot of crap placements so my next campaign I can focus on ad copy as well as move over a lot of the placements that didn't work.
Day 3
Spend: $25.07
Revenue: $5.28
P/L: -19.79
ROI: -78.9
Yesterday was a better day, ROI is not increasing as quick as I'd like. I ended up cutting angles 1,2,4 as well as 6. I also cut out a lot of the placements with no conversions and over a $1.2 as well as any CPA over $1.5. I also blocked out a category eating a lot of the budget and producing no conversions. Hopefully it'll look better today.
Below is up to date overview of campaigns and angles.
I've also spent last night making 8 unique banner designs based off angle 7 which will let me go back on track with caurmen's tutorial. And if that doesn't work, I'll move down to angle 5. I want to see if i can find a banner design that does better than the rest.
Today's ROI is better due to the cuts yesterday, I'm at -60%.
Questions:
1) I'm not even close to optimizing/giving up on this offer am I? I guess it's discouraging to see the slow increase in ROI but at least it's slowly getting better and I am cutting a ton of crap
2) Should I use a new campaign in decisive or just add it to current campaigns and pause all the other banners? I know Adwords has such things as campaign history or they take into account historical performances so I wanted to find out what everyone's opinion is.
3) I'm tempted to pause mobile/site campaign cause of it's poor performance but I'd like a second opinion.
4) My smart CPM is currently .5 and I use no where near my daily budget. If I increase that, I should win more bids right?
5) GIven how my stats are, would it be ill advised to just drop all the campaigns and focus on wifi site since it seems to be performing the best?
Looking better, but are any banners/placements giving CPAs even close to the offer payout?
Be mindful of this... If CPA is slowly going down but still 3-5x the offer payout, then there is some massive optimisation to do - and it may never get to profit.
2) As for adding creatives, just do it to the current campaign - might as well run them on the better placements from the start and campaign history won't be an issue.
3) How many placements have you culled through? If there are a lot doing poorly, I would cut the entire mob/site campaign.
4) Yes, possibly. You aren't winning a large % of the bids. To get more traction it would be better to go to CPM and manually force a higher bid. At the moment though you aren't paying CPMs close to $0.50... not sure. Nuances of Smart CPM I guess.
5) There is a lot of traffic there above what you have bought, so yes, focusing on that is a sensible option. The others 2 campaigns might be equally profitable, but focusing your funds on that one has its benefits.
Day 4:
Spend: $24.29
Revenue: $6.60
P/L: -$17.69
ROI: -72.8%
Thanks for the input Zeno, for the most part I've cut out less than 10 placements per campaign, the rest of the placements don't have enough data to come to conclusion.
I've noticed that for both mobile/wifi app, this one app is driving the conversions. The both campaigns have 39 conversions and 23 of them are from the app. The only issue I see is I'm winning a lot of the bids so I can't imagine I can earn too much from just focusing on that.
mobile-app
wifi-app
Angle breakdown to date
Focusing on angle 7 for now, will move to next potential angle if I can't get it to work (or just scrap the offer).
Next steps:Pause all campaigns, up bid to .55. Also removed mobile-site and focusing on the three campaigns. I'll pause the app campaigns if they aren't working or just whitelist that particular placement so I can get closer to profitability. Using new banners on current campaigns while pausing the old banners. Pray that ROI goes higher.
Looks like a solid plan! Angle 7 is definitely the one with the most potential, although you've still got a way to go with it.
You might also want to consider testing another offer for a couple of weeks and coming back to Brazil later. From what I saw, Brazil's CPM spiked hugely just before the World Cup - I'd expect it to drop back down again shortly afterward (13th July and on), which is likely to make running campaigns there a lot easier.
I was thinking about that too (been watching the cup all week), specifically how much my info might be skewed due to the immense amount of foreigners/tourists in the country now. Might run this for a few days, if it still does not work, move onto another offer (I'm eyeing an indonesia one right now which decisive has a decent amount of traffic in).
I'm pausing the AV offer. Angles are below over last three days (i've been cutting out any bad placements) but can't get the ROI to increase. For the campaigns, nothing really popped out. I'm going to take a look at decisive's planner and find countries with decent traffic and lower cpm averages and test AV offers in that country. I keep reading that it's highly recommended to run the banners to a lander for AV offers so I'm thinking about doing that.
How my new creatives worked. These are banners based off angle 7, 8 different designs. I guess it's not unique enough since the CTRs are so damn similar.
Day 5:
Spend: $20.35
Revenue: $6.65
P/L: -$13.7
ROI: -67.3%
Day 6:
Spend: $26.16
Revenue: $10.15
P/L: -$17.69
ROI: -67.6%
Day 7:
Spend: $24.26
Revenue: $7.35
P/L: -$16.91
ROI: -67.7%
Good call. Change country, change offer, and test again!
Thanks. All I've learned last week was test everything. I noticed my angles/banners seemed like they were too similar since the CTRs were closely together so I'll try to make it a bit more...drastic next time around. Also, spent a few hours yesterday building a fake scanner AV landing page (something I'd need to ask my AM about since i know not all advertisers like it). I made sure it never said they had one but it focused on putting the fear of god into them and implied they may be in danger. Also took the time to make it look like Android (downloaded a few GUI PSDs and I've specced out Android apps in the past).
Drastic differences are good! Try to sit and view all your banners in a single screen after you've created them - that's a really good way to see if they look different enough. Also, try viewing them from further away from your computer than you usually sit - another way to see if they blur together.
Yea, I think my biggest problem was while they are "different", the issue was I followed a thread here where they listed out the best ads based off their observations so mine followed that. Not enough color or variation (all banners were red/white/black).
Home > Mobile > Follow-along Campaigns