Hey All,
BACKGROUND
Read a bunch about AM and dived in with POF at the end of December 2013. Learned the fundamentals of setting up a campaign, creating banners, Lp’s, learned about tracking and a basic testing process. We were advised to cross over into Adult as you can learn the same key skills you can on POF however Adult has scalability as an advantage. So here we are in Adult running Campaigns through Traffic Junkie. Let’s dive in!
CAMPAIGN
Offer: Xdates
Source: Traffic Junky
Hosting: Beyond VPS
Tracking: CPVLab
INITIAL TEST DETAILS
Angles: We spoke to our AM’s and found out what the best offers were and ran with that. Researched what competition was running through Adult Spy tools and hidemyass direct to source. We then brainstormed 4 Angles and created banners and Landers for those angles.
Ads: (Static) 5 per angle, totalling 20 (varying: young, old, nude, partially nude, fucking, sucking… you get the picture.)
LPs: 2 per angle, totalling 8 (varying: images, headline, design, colour, static and dynamic pages, body copy, CTA etc.)
CAMPAIGN STATS
Geo: Tier 2 after reading a blog post from Dr. Ngo.
Spot: Footer Square. (after trying NTVA and but bids were to extremely high)
Budget: $25/day
Frequency: 3/day (would like to hear any thoughts re. frequency, as there seems to be some community divide on whether it’s worth it or not.)
Bid: 0.040 CPM
GOALS To break even and to get to $100 /day for 7 consecutive days, then breaking into $1,000/day campaigns.
DAY 1
Spent: $26.61
Conversion: 1
Rev: $3.30
ROI: -87.6%
Revelations:
Banner CTR became the obvious first concern with most ranging roughly between at around 0.03-0.06 @ 25,000 impressions. Some feedback on what the average footer CTR for a TIER 2 country is would be great. I have read somewhere that what we have is relatively consistent with others… thoughts???
Need more data before we reach a point to evaluate statistical significance.
LPs: Off the bat, we probably have too many. However, 2 pages per angle seems consistent as a start across most AM forums and would make sense when split-testing elements. Highest LP CTR was 34%, lowest 0%, with others dithering around either 6% or 20%. However, with the relatively low number of visitors per lander, it becomes difficult to speculate.
Any ideas what the base amount of visits per lander would be adequate before you begin making changes? Some people recommend 3x the payout?
Now we’re relatively new to AM, so if you see something we’re doing wrong or some of our testing is ineffective please let us know.
Anyway, let’s get the ball rolling.
Will update you all with day two soon!
Good work! The variance in CTR that you're seeing on your landers is particularly good - shows you're testing widely not just testing tiny variants of the same thing.
For lander split-testing math, I heartily recommend this approach: http://stmforum.com/forum/showthread...Data-Read-This! That'll tell you exactly when a lander's CVR can be trusted. I'd strongly recommend cutting on CVR, NOT CTR - see http://stmforum.com/forum/showthread...y-Don-t-Matter
Thanks Caurman. We did design the LPs with wide variance. Images, layout, colours etc. We wanted to test what worked generally first, then hone in on split-testing specific elements. Thanks for the links too... will check them out. Update coming soon.