Another Chef Doing The Appetiser (first campaign) (15)
Hey,
after trying several campaigns here and there to get the feeling for each type (pop, ppv, mobile, etc) I decided to follow the appetiser and I'm pretty happy, learning a lot!. Let's start:
GEO: Argentina
Traffic: go2mobi
Offer: adworkmedia - whatsapp emoji pin submit ($0.18)
Tracker:
It's been already a couple of days, but I'll try to explain everything so far. First, the global numbers:
spent: $38,58
revenue: $4,86
roi: -87%
From the original banners, there were 4 winners, and 4 losers. If we sort by conversion rate, I killed the last 4 (no hope for them):
After creating 4 more banners and letting them run for 1 day, I'm here right now (check out the first one, might be a killer!). Also notice that some of the new banners don't have much data yet, and that I've manually adjusted the bidding:
http://i.imgur.com/vSQgKIY.jpg (sorry img too large)
Then, I moved to placements, and blacklisted, as per the guide, those without conversions with 3-4x the payout. I killed 9 that were eating up to 70% of the money!
Note that this is the best placement so far:
Then, I moved to OS. You can see that MOST of the conversions were from windows phone... on the windows page! LOL
http://i.imgur.com/9qTdOHv.jpg
Now moving to carriers, pretty clear here too: (I'm starting to think all the conversions were the same person from the same phone...)
http://i.imgur.com/OAuPzuc.jpg
After cutting, this is the new campaign that I have today:
COST: $9.60
REV: $1.62
ROI: -83%
Things to consider:
- the autobid is pretty expensive!
- I think I've cut too much too soon and I've killed the campaign
- I have the feeling that the payout is way too low
- should I run away?
Your screenshots seem to be in the wrong order. 
But everything's there so all is well.
Very nice data analysis! And you're getting conversions which is always a good sign!
Some thoughts below:
1)Test multiple offers. Yeah your offer payout does look pretty low. Although payout is only a part of the equation, testing multiple offers from the start can't hurt. I've seen in the past Whatsapp offers in Argentina that pay a lot more than 0.18 - although I don't know what kinds of offers are out there right now. You can do a search on CPA offer aggregators such as odigger.com and offervault.com, and sign up to some aff networks for those offers if you need to.
2)Cut banners using stats tools. Just in case you didn't - here's the method suggested by caurmen:
http://stmforum.com/forum/showthread...ats-Calculator
3)Be careful when cutting for multiple variables at the same time. Yes it would seem that Claro and Windows Phone have the most potential, and it seems like a good idea to cut everything but these in order to get green faster. But have you drilled down to BOTH variables one after the other (e.g. carrier -> models) to see how they're interrelated, and how the cuts will actually affect the campaign as a whole?
For example: What if the carrier Personal is the one used by most Windows Phones? If that was the case, then cutting Personal from carrier targeting may not be the best decision, as Windows Phone is your best-converting model. I just pulled this example out of my arse, but you get what I mean.
4)Autobid is expensive. Yup - as you've mentioned! Starting with a low bid like you did is wise. Doing so can allow you to test for a LOT cheaper than doing autobid.
Very nice start so far! Looks like you have some nice banners - by testing and finding a better offer you may start to see green.
Amy
Hey, thanks for the comments Amy. I haven't blacklisted everything yet!
Changed the bid to manual ($0,12 CPM) and getting a win ratio of 70%, so I think I can still lower it more. With the new properties, this is what I have for today:
COST: $3.91
REV: $2.16
ROI: -45%!
I'm pretty happy with the results. Also upped some more banners with the 'You need to update WhatsApp' angle, that seems to convert pretty well. I've learnt to do gifs and it's fun :P Also, there are a couple of banners with positive ROI, yay!!
http://i.imgur.com/BxwbH0A.jpg
Next steps:
- gather more data
- add more banners
- wait
-45% is very exciting! And you have a couple of green banners already!
I wouldn't suggest lowering your bid - you're already not getting a ton of traffic. You won't want to slow down your traffic further since you're wanting to test more banners.
Once you have a solid angle, and corresponding banners that do well, DO TEST MORE OFFERS!
The better your funnel is, the higher you'll be able to bid, and the more traffic you will get - which will result in more profits overall.
Very excited to see what you can do with this - there's so much potential!
Amy
Hey, thanks for the quick reply!
Do you think I should add more exchanges? I'm only working w/smaato right now
Originally Posted by thedrone
A good strategy is to target all exchanges first, see which one will give you the most bang for your testing buck, then JUST use that exchange to test all your stuff.
Then once you've optimized your funnel (i.e. have an offer + angle combo that works well and is sufficiently profitable), you can always target all the exchanges again to see which ones will be profitable.
Alternatively - since Smaato is already working out well for you, you could just continue to use that to test banners and offers - once you have a "best" offer and angle, expand to other exchanges.
Amy
Well, added another exchange, a couple more banners and adjusted the bid a little lower.
Fucked up the campaign, no impressions yesterday. Let's readjust the bid 
If I get this right, I've been moved to the bottom of the bidding queue because of the bid change and now this will have to 'warm up' again?
Thanks
I did suggest not to lower the bid further - although I'm not happy I was right. No harm done though. Did you increase the bid again? Did you start getting traffic again?
Try to gradually up the bid until you start getting traffic again.
If you still get no traffic, contact support (of the traffic source).
Worse comes to worst you could always create the camp again...
Amy
Yes, I did experiment without consideration for your opinion. Upped the bid again (to the previous value) and getting traffic back again. Changed the offer to whatsapp wallpapers instead of whatsapp emojis (pays 5x) so I have to adjust all the parameters again. Will get back with results!
Thanks again
Originally Posted by thedrone
Thanks again
Amy
Wow, yesterday was a pretty strange day. These are the "final" stats:
http://i.imgur.com/aPqEM2T.jpg
As you can see, it's pretty good. But I've almost run out of banners. I received an email that some banners were no longer valid (I think they approved them without translating them and later checked the meaning, etc) so I just have two banners now. And a little after that it showed 60 conversions!!! (
- Am I banned?
- Also think that this offer is pretty crappy. I cannot find another emoji offer, so moved to a whatsapp wallpaper offer (not converting as good as this) and I'll try another offer totally unrelated (read this on a success thread..)
- I'm going to start more campaigns with different offers or maybe countries, wanna polish this 
Thanks!
- Am I banned?
Regarding the 60 conversions: Could there have been a delay in conversion reporting at your aff network? Worth asking your AM about. Also: Your screenshot shows 13 conversions - I assume the rest of the conversions came in later? Would be interested in seeing a screenshot of the stats that include all 60 conversions.
Are you targeting multiple exchanges? Perhaps the sudden drop in traffic is due to a high traffic exchange having banned your banners and is therefore no longer sending any traffic? Worth checking for sure.
It may also be a good idea to message your mDSP's support to ask them whether you've been banned (or your banners at least).
Amy
Ok, so the banners got banned for being too aggressive. Gotta change the angle (and the offer). I'll start again (and create another thread). Maybe try pops as it's easier bc/no banner involved. Thanks for the replies, I've learn a lot here!
Cannot explain the phantom conversions though.
a) stick to the plan, don't improvise unless improvisation is planned
b) don't cut too soon (or too late). Math is your friend.
c) measure everything, keep all data
d) work the angle
Pop is a very popular traffic type for newbies, which is why it has gotten very competitive especially in recent months with the massive influx of newbies into the AM field. Generally there will be a lot more volatility in performance etc. compared to display - although after your rocky start with display I'm not sure I can make this general statement for your particular situation.
One major difference between offer-testing approaches for pop vs. display, is that for pop, you don't even need to brainstorm angles at first: Just rip all the different landers you see from adplexity for that offer type and test them and you'll likely find at least one that will work half-decent or better. Then you just use that winning lander to mass-test offers. You can always come up with unique angles and original landers, but they'll get ripped in 2 seconds unless you take measures to protect them from being ripped.
I'd suggest to use ripped landers for initial testing, and then when you find a good offer and see profits, to THEN spend the time on coming up with original landers, or variations of the best-performing ripped landers.
Another major difference between display and pop, is that for pop, the bid tends to affect traffic quality a lot more. Based on personal experience, there's more bot/fake traffic on pop sources than on display sources. Actually this may only be true for display apps and not sites - the reason may be because it's much easier to create bot traffic for sites compared to apps. Anyone (including you and I) can send a bunch of junk traffic to any target website/placement we know the url for, using a simple software; whereas many people don't have the know-how to send bot-traffic to an app (unless the owner decides to do this themselves - which I know for a fact does and has happened).
Anyways - my point is that because there's more bot traffic on pop sources, it's important that you bid high enough to avoid all the junk at the bottom of the barrel. Sure enough, sometimes you can bid really low and still see green. But when testing a new geo and/or offer, it would be better to bid above-average to ensure you're getting good traffic quality that will do your offer justice. Another common approach would be to set up multiple camps that are the same except for the bid, so that you're testing multiple bids at the same time to see which one will give you the best ROI.
Also - try to test at least 2-3 offers from the start (for your chosen offer type and geo), that are recommended by AMs from preferably more than one aff network. Doing so will increase your chances of finding a good offer compared to just testing a single offer. You need at least one converting offer to help cut landers to find a winner - which you could then use to test every offer you can find on all the networks you're an affiliate of.
_______
That's a primer for your next follow-along in a nutshell - please do send me a PM if you do start that next one. Would be glad to provide some feedback there also.
Have fun and best of luck!
Amy
Oh my Amy, thanks! This is gold. Will surely start a new thread when ready
Home > Mobile > Follow-along Campaigns