First time in Adult Dating. Here goes nothing! (24)
Noob to Adult Dating getting started. I've done Facebook advertising for my own games which was outside the realm of Affiliate Marketing since I did my own promotion and there was no tracker or dealing with post backs or anythings. Or I've helped my partners who are affiliates with random aspects of marketing and promotion. But never done the whole process end to end myself through an Affiliate Network + offer. So here goes nothing.
Starting with my first offer
Site: BangarooBabes (Mobile)
Country: Australia
Payout: $1.75 (CPA)
I've created a few banners, and have accounts on TrafficJunky and ExoClick for traffic sources.
I have the starter VPS package with BeyondHosting along with Prosper202 ready to go.
First roadblock, TrafficJunky is closed so I can't fund my account. And ExoClick gave me this is issue: http://stmforum.com/forum/showthread...fication-issue
Once I get this issue resolved, hopefully I can continue. I have everything else ready to go and am looking forward to having data start coming in.
Haha Bangaroobabes, interesting!
Nice start, it'll be interesting to see $1.75 is quite low for AU traffic, but that offer might convert extremely well so only testing will tell.
Looking forward to the results, TJ will be live shortly and I've never seen that Exo message before :S
Haha Bangaroo, love it.
Get on it Maynzie and give us a locals opinion.
For all we know it could be a boxing/MMA dating site given how feisty those marsupials are.
And yeh, $1.75 seems quite low for AU but it is mobile. Might be worth hitting your AM up for a pay bump off the bat here, say something like "how high can your go on Offer X? I'd like to test it but $1.75 seems really low for AU traffic".
I'm super interested to see how this offer converts it could be a hidden gem, or then again a flop. If it is a flop, I've found that running web styled landers in AU (keep in mind get an approval from your aff manager first) have been some of the most profitable for AU. That is they carry really hefty payouts, and some convert even better then mobile optimised.
Keep us updated man!
Right now still trying to get the Traffic Sources sorted out:
Exoclick: Verifying one of my references.
TrafficJunky: Sent in ID and other info, waiting for the money I sent over with PayPal to show up in my account.
Traffic Force: Just approved last night. Uploaded my first four creatives.
I used the adult spy tool to get an idea of current creatives that are working, and developed some based on that.
These are the creatives right now. I'm A/B-ing putting a small Australian flag on it. And am also going to vary the contents of the animated portion. Also want to experiment with different text soon.
Creative 1A http://www.imagebam.com/image/725bb8355219399
Creative 1B http://www.imagebam.com/image/5ce6e0355219400
Creative 2A http://www.imagebam.com/image/d33ea9355221567
Creative 2B http://www.imagebam.com/image/120ed4355221569
Currently I'm running 4 different placements also.
Once I'm sure everything works end-to-end with tracking and analytics, I'm going to increase permutations and then begin work on landers.
My AM bumped the payout to $2.25. Early data is coming in. CTR is around 0.2% which I guess is pretty standard. Conversion rate is very low (< 1%), which I expected since I haven't optimized my creatives and I'm direct linking for now.
I've attached some of the breakdowns from Exoclick and Traffic Force. Once I have another day or two of data, I can prioritize certain placements or creatives. Also, I realize I need way more creatives - going to get started on those ASAP.
Ouch @ that conversion rate.
You should use a lander, really.
Also, I'm not sure the animated element is congruent with your ad copy. I'd change the imagery for the 'wait before **' ad copy or change the ad copy to focus more on the benefit(s) of the offer that relate to what they are... seeing.
Thanks zeno. I'll prioritize making a lander. I'll start with quiz format that most adult dating landers are using ("Are you over 24", "Are you looking for something casual", etc) but I'm going to experiment with some other stuff also. Maybe a comparison of various dating sites, with the one I'm promoting coming out on top in various metrics like the ratio of women to men (because we "screen men thoroughly" through the quiz)
Also, thanks for the pointers on the creative. I'll try both approaches. 1) Changing imagery while retaining the "wait before **" and 2) Retaining imagery and changing copy to be more about the benefits.
I've learned a lot already - and I'm excited to see where this will be in a week or two!
Okay, lander is done. Haven't had a chance to work on the creatives yet but that's coming soon.
Here's the impact on my results:
Lander click through rate is 7-8% overall. Once they're past the lander and on the offer's page, signup rate is 20%. So overall conversion rate through the funnel is about 1.5% - which is slightly better than I was doing when direct linking (closer to 1%) but far worse than needed.
Assuming I can improve the creatives to the point where CPC goes from about $0.50 to $0.25 that means I need EPC to be greater than that. EPC is currently $0.035 (0.015 * $2.25 payout) so I need around a 7-8x improvement in conversion rate through the lander for this to have a hope of being profitable.
I added MixPanel and here's my funnel:
I pasted two separate images together, so please excuse it looking a bit odd. Not as many data points as I'd like but I spent another $30 to collect this and there's probably enough there to have an idea of what can be improved.
More than 75% of visitors to the lander bounce, so either the value proposition isn't being communicated effectively, or my creatives are misleading. There's another choke point later in the funnel where less than 40% click the sign up button after answering the questionnaire. That needs improvement too, but I'll focus on that after improving my creatives and the first part of the landing page.
I will definitely follow Zeno's advice and make my creative more congruent with the images being used. I will also be using Optimizely to do some A/B testing with different text and images on the first page of the lander. I'll also give another pass through the adult spy tool and see if I can find any more common themes for inspiration.
Here is my lander: http://www.exclusivefuck.com, I based it on caurmen's sample lander, and added some of my own Javascript / jQuery for the questionnaire and countdown timer. I certainly don't mind if anyone borrows anything from it. Hardly any award winning code there. 
You should test this lander on a mobile device or in emulators (e.g. Chrome + BrowserStack).
For me, the text is very small and I don't like the buttons much. The image could take up the entire browser width and could be better - more vibrant.
Good work, and nice to see that you're already seeing improvements from a lander.
I'd also recommend split-testing quite a few images on the lander, in different styles - you'll probably find some drive much higher click-through rates.
In my experience, it's also been worth testing a single list of rules against a repeated click-through one-question-at-a-time style (as you have now). May generate higher CVR, may not.
If you can, finally, I'd strongly recommend adding some Javascript tracking into the page to see which questions you get drop-off from, and if there's one question in particular that seriously stands in the way of signups.
Yup, I've got MixPanel tracking the drop-off across each of the questions. Biggest drop-offs right now are 1st question, and the final click through once they're "approved". So I'll focus my A/B testing around these.
Definitely worth checking a list of rules type lander also.
Zeno, I finally signed up for BrowserStack so I see the issue. Safari on the iPhone auto-scales text if it thinks it'll be too small, so setting -webkit-text-size-adjust helps me resolve the issue and use a common font size between iPhone and Android for similar screen sizes. Also adding more CSS permutations for different screen sizes.
I'll do my first A/B test around the first image that they see (I can go more raunchy, or more conservative) and test out button styles. So that should give me a good amount of data. And then keep testing until that CVR is up to par.
Here's my progress over the past week. I've finally been able to devote a few days to working on this.
Ran an initial A/B test varying the first image that visitors see. There didn't seem to be much of a discernible difference.
Here's Optimizely for the first A/B test
Note that conversion rate in the Optimizely image is the "engagement" (number of users who click "yes" or "no" in based lander that I'm using)
My second A/B tested a few different things. The first variation was changing the button style as per zeno's suggestion since that was just a simple CSS change. Second variation was with a headline.
Once again, not enough data for it to be valid, but the buttons do seem to be a small improvement. Here are the 3 variations and Optimizely.
Original:
Variation 1 (iPhone styled buttons):
Variation 2 (with headline):
Optimizely results:
The cost to get data is so high right now, I'm going to switch gears and try to optimize creatives. Paying $20 to get 80 clicks is rough since I might need several hundred clicks to get data resembling anything valid.
Following the suggestions above with some landers, so I'll know how those are going in the next few days.
For my lander, I'll try the 1 page rules based lander as suggested, as well as redoing the layout to be more vibrant and eye-catching. I also have a few other things to try before I ditch this offer. This is my first offer so I'm not expecting anything close to even break even. Indeed, so far at -90% ROI.
Good progress so far - and yep, switch it up, test more, drop if it's performing at -90% ROI after that.
And sorry -- I was testing the IP header spoofing for spying the other day and used an Australian IP. I actually saw your lander after clicking on a bunch of ads. I should have converted for you. But I'm not so easy, try again :P
Haha...I will work harder at converting you, zeno! 
I took a look at my day-parting / week-parting graphs and saw that the bulk of my conversions happen on the weekends and between 8 pm - 11 pm PST and 8 am - 1 pm PST.
It would be cool if Prosper could do some sort of geolocation for time zone, but assuming all of these were in Sydney (I realize Australia has several time zones), this corresponds to 2 pm - 5 pm and 2 am - 7 am. Most of the conversions are also on Friday and Saturday. So makes sense.
So I spent most of the week working on another offer and am saving my A/B testing for this weekend.
Here's what's on tap:
- New creatives. I'm trying one with an Indian girl since I heard (maybe here?) that Indian girls are considered exotic in Australia and could appeal more to some men.
- A/B testing. I'm testing out a different lander that has another color scheme that matches the offer itself, and changing up some of the copy.
- One more test with a one-page lander with a bunch of rules.
I'll update this thread with any data that comes out of the A/B tests. I wasted today's data since I overzealously copied/pasted code from my other lander and forgot to update MixPanel, Optimizely, and Prosper202 links in my page. So I had no analytics and was directing visitors to the wrong offer! Ah well, lesson learned. Will have to add that to the SOP.
I haven't updated this in a while since I've just been collecting data. And learning some PHP on the side to help out.
On the plus side, I've managed to crack that -90% ROI I'd been seeing consistently. Last few days have been inching closer to the -50% ROI mark which was a goal of mine for a while.
Here's what's been going on:
Large A/B test with the following variations:
Original: 3 question lander
Variation #1: Use color scheme, font, style matching the offer
Variation #2: Using geotargeting. I follow caurmen's guide, zero issues and very easy to follow (http://stmforum.com/forum/showthread...our-Own-Server)
Variation #3: Single page rules lander
Here are some screenshots:
Original:
Variation 1:
Variation 2:
Variation 3:
And the corresponding two parts of my MixPanel funnel through the landers:
Funnel 1: 
Funnel 2: 
Note that Variation 3 is the single page rules lander, which has a somewhat higher lander click-thru than expected. Unsurprising, but still a quite high bounce rate.
Nevertheless, conversion is what matters in the end. Here's what Prosper202 has to say:
I realize we're so deep in the funnel, that I have very few data points left. 343 clicks to 35 being presented with the offer - and 5-10 for each lander. But that's how it goes with my testing budget. :/
Nevertheless, looks like "Lander - Blue" (Variation 1) and "Lander - Geotargetted" (Variation 2) have some promise.
So the next step would be try merging those two into one variation. That is, the same style as the offer along with geotargetting.
Costs for the above were $113.07. Revenue was $20.25. -82% ROI. So inching up but still bad.
Ultimately my cost to get data is still too high. About $0.20 CPC, so I decided to focus on the creatives again. I'd been working with two angles so far. The usual "Stop Jerking Off", and also "Bored of your wife?". I'll be coming up a few more angles as well.
Here's my Exoclick report from the above test:
Those variations should probably be named more descriptively. :/ But looks like some definite losers (9474158) which also had no conversions - so I'll cut that.
947246 has the highest CTR, but no conversions. So that can also be cut.
In fact, based on Prosper, there are three standouts:
Creative4, 2B and 3A have some potential. I know there's low data (especially for 3A) - but there could be something there. 4 and 3A use the "Bored of your wife?" angle. Creative4 uses an Indian girl. Creative2B is the original angle.
One thing I'm doing on another campaign (no follow-along thread yet) is matching the first girl they see on the site to the creative that they clicked on - so I think that could be worth trying. If they click on the ad expecting "exotic" girls, and then just see the usual girls, I expect that'll cause some to bounce.
The other experiment is on creative sizes. I've been using 300x100 placements for now. I also want to give 300x250 a try - and I just tried that yesterday. I'll post some stats on that shortly.
They didn't help with LP CTR, but they had a significant impact on conversion - over 4x improvement to conversion based on very limited data. Last image in Post #16 has the numbers.
I'll be adding a more detailed post with my latest tests soon. But Tuesdays typically have a poor conversion rate for this offer (Thurs - Sat are the best days) for me. In fact, I haven't had any conversions at all on Tuesday or Wednesday. But using the GEO lander gave me an almost 2% conversion rate for Tuesday which is still not good, but better than before.
Good testing and writeup bud.
I would go straight for the geo-targeted and blue versions and test more from there. Try placing the geo-info in a more noticeable position, highlight it.
Thanks. I just completed a version that incorporates both the blue theme and geo targeting. I'll play with highlighting the geo. Will run a few A/Bs about putting that into a header, top of the image, calling it out in the CTA after the 3rd question, etc.
300x250 placements seemed to yield visitors that had better LP CTR (17% vs the usual 10%). So hopefully that will translate into improved CVR as well.
Making progress...slow, but making progress.
Thursdays have converted well for me, but admittedly I don't know at what times since Prosper202 cannot give me simultaneous Day and Week Parting information.
As an aside,
I ran the campaign this morning and results were very bizarre. 85% bounce rate for my lander (which is normally around 50%). 4% CTR through the lander (normally around 15%). This was over 50 - 60 clicks so not an great amount of data but worrying nonetheless.
To be fair, Thursdays may have done well in the second part of the day (7 pm onward, which is 1 am on a Friday in Australia). Which is why I need day parting info filtered by day-of-week. So we'll see how it does this evening.
I'm trying a new angle, "F*** a fat girl beats j**** off". My lander isn't updated to reflect that and I won't have time until tomorrow. So I'll simply be evaluating CTR for that angle to gauge its potential - and then updating the lander as appropriate.
In the spirit of collecting more data, I've separated my Low and High CPM campaigns with a custom Prosper202 Variable and passing that to MixPanel also. Unfortunately Exoclick doesn't support a bid token so I have to do it that way. Since I'm trying a few separate campaigns with lower bids, I'd like to see if the quality of those visitors is worse, and by how much.
I think it's time to abandon this campaign. CVR is basically at zero, LP CTR has dropped to a third for some reason. Offer might be saturated at this point. I've put about $700 into this overall and made back $76.
I've learned a lot and it's been a great vehicle for getting the right mentalities and technical know-how. But it makes no sense to keep pouring money into it. At this point, I think to grow as an aspiring AM, I'll need to spend time with other offers and learn about the differences - and ultimately how to judge a good offer from a bad one.
I have three other offers I'm working on. Adult dating, and two others are traditional dating. I'll keep this thread alive with the new adult offer. More info coming soon.
Originally Posted by blofeld
Knowing when to drop an offer/campaign is important.
Testing other offers and campaigns is the most important part of your workflow - we don't get handed money by clients to run campaigns on a performance basis so it's up to use to seek out and attack the campaigns that will bring home bacon.
Good share
Home > Adult Traffic (NSFW) > Follow-along Campaigns