Home > Adult Traffic (NSFW) > Follow-along Campaigns

---

Watch me when I'm reaching to $XXX in 2016! (21)

---

Hey guys,

I believe in systems and goals. So, I want to start my upgrowth with "Follow Along" threads. Here is my first Follow Along.

DETAILS
GEO: UK
Offer: Adult dating (Milf) - SOI - Desktop
Payout: $2.80
Traffic source: Traffic Junky

I've spent $250 on RON to get some data with 1 LP, 10 banners, 4 offer LPs. And here is the result I got in 3 days:


With -25%, I thought this is it! I'll get this one profitable then scale it up. These were metrics I can improve:

• Better AD CTR, so better CPC
• Better LP CTR
• Conversion Rate
• Offer CPA (maybe)

Banner and LPs
Then, I got banner which has best CTR and lander which has best CR. I've put 4 new banners that similar to my best banner, put 1 new banner that looks-like my best performing lander.
(5 Banners & 2 Landers)

Targeting
I have selected my best performing GEO, best performing site and spots. Capped 1 view per user in 24 hours.

Here is the result for 2 days:


I know it's very low data, but it seems it's green now! After vicious optimization it got 50% ROI, but volume is very low. I think it's about low CPM bidding and tiny targeted audience.

What is next, how can I get some volume and still be in positive? Some ways, in my head:
1) Since I know GEO / Site / Placement performing good for my campaign, I increase my CPM bidding to get more volume. Then, I can focus on trying optimize that traffic with banner and lander tests.
2) I can change cap to see if my ROI stays same. So, I can get at least 3x-4x impressions.
3) I can add a few more sites, that is not possitive but still have potential. After volume, I can try to get better CR from them.

What do you advice? What is next? How to get more volume with positive ROI?

I'll update thread everyday, check it out!

Thanks,
R.

---

Off to a great start! Well done. Scale it bigger!

---

No. 3 makes more sense. When you're just starting out (I presume) on a new vertical, new geo, you need to first optimize your funnels. You need data. Don't cap off your traffic too early and throttle your whole progress !

---

Nice start and progress! -25% ROI almost right off the bat looks like the campaign has a lot of potential.

Adult dating isn't the easiest game to play anymore. I gathered up some tips in this post which may help:

http://stmforum.com/forum/showthread...l=1#post259000


Some comments about your process:


1)Testing more banners would be good. Try to come up with different angles, and create several banners based on each angle so you can see a trend, and give each angle a reasonable chance to prove itself. Once you see a trend as to which angle(s) seem to work better than others, you can create more banners for the same angle(s).


2)To optimize your funnel, you need to test multiple candidates so that you have several things to compare to each other, and then cut the inferior ones. Then you add more candidates to the test to begin another round of split-test and again cut the inferior candidates. Repeat a few times with offers and landers and banners and you'll have a much better funnel!

Right now your single offer and single lander seem to work OK, which makes it the perfect time to test banners. Once you have better banners, it would be great to test offers next, by adding other UK Adult Dating offers into rotation. After you find the best offer you may want to test landing pages next. Have you done the spying necessary to identify the most popular offers and landers being run for your geo yet? If not - this post will help you (scroll down to "Cherry Picking the Best Offers" - note that you can also cherry-pick landers while scouting out the best offers):

http://stmforum.com/forum/showthread...l=1#post196820


3)Cutting anything based on CTR is not a good idea. After all, CR is what makes you money, not CTR. And I've seen many, many low-CTR banners and landers make more money than high-CTR ones.

Here's info on how to cut landers: http://stmforum.com/forum/showthread...Banners-Part-1

As for banners, it would be better to test several per angle as mentioned above. Once your campaign is in profit, and you want to cut any banner that isn't likely to be profitable, you can use the process described here:

http://stmforum.com/forum/showthread...Banners-Part-2

You can try all 3 and see which one is more cost-effective. Basically it boils down to this: If you want to save money, you can still do your testing with the best site + best placement but it will take longer. If you want to save time by paying more, you can buy more traffic (by increasing CPM or including other sites/placements).

Regarding frequency cap: Typically the more you open up the frequency the lower the conversion will be. By all means play with that and see what happens. Personally, I start with 1/24 and will only open it up after I already have a profitable campaign but want more volume at the price of lower conversion rates, and hope that the overall profits will still result in an increase.


Thanks for starting this follow-along! Can't wait to see the next installment!



Amy

---

Hey guys!

@meathead, thank you very much! I'm trying my best to scale it.

@adsflo, thanks a lot for your recommendation. (I'm really happy to coming across to your inspiring thread. It has boosted me.)
After, getting some data with more GEO. My next volume action will be absolutely adding more sites.

@vortex, hey vortex! this was very very helpful. THANK YOU SO MUCH. I read all of your comments & Finch's posts & all statistical work. Looooots of reading and lots of taking notes. WOW!

Thank you all,once again for encouragement. Here is my updates about my status.

OVERALL RESULTS
The only actions I've taken is adding new 3 GEO. And, here is the total result

Nothing special, I just got more volume and I think we got good data now to take action about banners & landers and offers.

METHODOLOGY
I have read lots of good stuff like Finch's premium post, statistical methods, caurman's tips, vortex's tips. I'm nearly clear about my framework to turn my campaign profitable and get volume.

• I'll use Finch methods to find best offers, landers and banners.
• I'll use statistical methods (Thanks to caurmen & vortex) to cut banners, landers, offers, placements.
• I'll master on some placements.
• I'll try to add competitive advantages to make campaign work long.

WHAT IS NEXT?
Since I got 300.000+ impressions, I think I'm ready to cut some banners, landers, offers or placements whatever looks OK to cut.

• to Optimize:
  
  • Try to cut banners
  • Try to cut landers
  • Try to cut offers
  • Try to cut placements
  
  
• to Test:
  
  • Pick and add new banners
  • Modify best banner to create new banners with borders etc. to test
  • Pick and add new landers
  • Modify best lander to create new lander with flags etc. to test
  • Pick and add new offer
  
  
• to Scale:
  
  • Add second best placements
  • Increase bid for best placement
  • Increase freq. cap

So, in this thread I'll cover "to Optimize" part for now. Let's dive in...


1. Which banners to cut?


Overall metrics says KILL banner-3 and banner-4. This is total results of two campaigns. One is for 1 GEO, the other one is set for 3 GEOs. Let's look two campaigns to see if they validate separately:




So...

• Banner-1 is OK for total results and Campaign-1, but it's red for Campaign-2. So, it'll be KILLED for Campaign-2.
• Banner-2 is OK for total, Campaign-1 and Campaign-2.
• Banner-3 have to be KILLED for both campaigns.
• Banner-4 is RED for total. It's RED for Campaign-1 too. OK, it's gone. But, for Campaign-2; if we look total numbers it tell KILL it, if we look Campaign-2 stats it tells KEEP RUNNING. What do you think? I didn't decide if I should KILL it or not.
• Banner-5 is OK for total, Campaign-1 and Campaign-2.

Here is other information about banners. CTRs:
Banner-1: 0.6476
Banner-2: 0.4583
Banner-3: 0.3898
Banner-4: 1.3334
Banner-5: 1.5023

Here is the ROI of the banners:


Bottom Line # My questions:

1. What do you think about Banner-4?
2. Do you think I need also do other tests statistical significance and Bayesian test?
3. I haven't look for CTRs. Is it OK to only optimize with "Impression-to-Conversion" perspective?

I'll continue cutting tomorrow with...
2. Which landers to cut?
3. Which offers to cut?
4. Which placements to cut?


Blood for oil.


Thanks,
GREENzone

---

Your posts are a pleasure to read - you organize the information so nicely in such a logical fashion, and use color and spacing so meticulously. Also - not everybody will bother to find out how to display images in full-size - I know I didn't before I was assigned moderator.

Your methodology is great - you'll no doubt modify it and add more details as you gain experience, but it's an awesome start!

With regards to which banners to cut: You're definitely on the right track! However, during initial stages of testing banners, it would be much better to focus on testing angles first to find out which angles perform well, instead of fixating on individual banners.

If you create multiple banners for each angle, you'll be able to see a trend on how well each angle performs, based on the performance of all banners for that angle. This is more accurate than basing conclusions on a single banner - because the banner may have just gotten "lucky" with a great image or something.

Once you know which angles do well, you can produce an endless stream of banners from the same angles. Whereas if you only have a "lucky winner" banner, you'd have less of an idea on why it converted well.

Also - once you know which angles convert, you won't need to be super-hesitant when cutting banners - because you'll be making so many banners from the good angles that you won't care. Using the spreadsheet THEN will make the experience a lot more satisfying.

One of my favorite tricks: Take the best banners and make combinations of their headlines and images. This can often make best banners even better!



Looking forward to your next installment!


Amy

---

@vortex, thank you very much for your help. You always give best tips and your words are very motivating.
Grouping banners as angles. is a great tip. So, I'll do test 3 angles with 3 banners. Let's see which angle wins!

Here is what I did for today:

2. Which landers to cut?

Result: I cannot cut one of them.
Next: Absolutely add a few after my spy tour.

3. Which offers to cut?

Result: I cannot cut one of them.
Offer converts with both landing pages well for now. I'll continue with it.
Next: Maybe I can test Double Option of same offer. I'll also try to detect leading offer. Maybe, then I'll give a test to increase CR. I've %11 CR (Offer Page) at the moment.

4. Which placements to cut?

Result: I'm saying good bye to Spot 2 in Camp 1.
Next: Maybe I can test that spot for that GEO later after I've optimized my funnel better.

Next actions for next hours:
- Cherry pick 1 king offer, 2 landers, 2 banners
- Add one more new website target (goal: more volume with similar similar roi)

Thanks for follow,
GREENzone

---

After my first in-depth cherry pick, seems like one network is really dominating here with similar offers. I got disappointed seems like top affiliates on my spot are very special for that network, so i think I can't get same offers. And more, I will never get same payout so I think It will really be hard to win the battle with those big boys. Let's think more about this one.

What do you advice?

---

A quick update. Overall stats:

---

One more quick update. Testing 2 LP of my offer. I'm %88 sure one is converting better, but still need +2 conversions to statistical proof:

---

Offer LP A vs. Offer LP B (Last 7 Days)



Offer LP B goes to HELL!

This is overall data for 4 GEOs and 2 Landers. Do I need to wait all sub dimensions will be statistically significant?
Like Country -> Offer LP (4 x 2 = 8 dimension)
Aff Lander -> Offer LP (2 x 2 = 4 dimension)

---

Do you mind sharing what tools do you use to compare landers and banners?

Great progress!

---

Actually, they've all mentioned in optimization posts of Vortex and Caurmen.

• Statistical Significance for A/B Split Test: http://getdatadriven.com/ab-significance-test
• Sample Size Calculator: https://www.optimizely.com/resources...ze-calculator/ (Use this when you plan to add a new banner, lp, offer, targeting; to know how money views do you need to see true results.)
• Test Ads/Banners with CR Data: http://www.cardinalpath.com/resource...-testing-tool/
• AB split test graphical Bayesian calculator: http://www.peakconversion.com/2012/0...al-calculator/ (Use this when you need to decide with lover volume)

---

When I see numbers like these, I dont bother checking if its significant or not its too much of a difference. Almost double the leads with the same amount of traffic.

If you want to stay in adult, you need to develop some kind of a "sense" to be able to make decisions fast. If you wait for everything to reach statistical significance, the funnels will be dying on you before you even reach it.

---

Good tip You're right, I need more speed than competition. So, yes, as you say I need to develop a SENSE. thanks matuloo!

---

Everyone has a different way of doing things - but I wouldn't be so quick to ditch the split-test calculator.

For those of you who have been doing adult for more than a few months, you'll notice that landing pages haven't changed that much - the ones I'm seeing today are similar to the ones I was running nearly 2 years ago. So I serious doubt that by making decisions before stat sig is reached, it will allow you to get ahead of competition.

Stats will allow you to stay objective. Without this objectivity, we'll start making decisions arbitrarily based on emotions. One of the most well-known super-affiliates in the industry told me recently that knowing how to cut stuff based on statistical significance is a skill that if mastered, can give you a significant edge over your competition (pun not intended).

If you're a very experienced affiliate marketer, you MAY be right more often than you're wrong by using your "sense". But as people relatively new to AM, not using stats in your process is not recommended.

In the end, everybody develops their own method. But I've seen too many new people make the mistake of cutting things arbitrarily without using statistical methods. Thousands of dollars in the hole later (or worse), they realize their mistake, learn to use the split-test calculator, and finally start to see their ROI improve - all because they're finally cutting the right stuff!

Just my 2 cents.



Amy

---

If your offers are very similar - for example all of them are adult dating for straight folks - then I would just look at the overall lander stats (total across all offers) and cut using the split-test calculator.

Each geo will react differently to different landers, so I would suggest to test landers for each geo separately.


There will always be bigger players than you. You don't need to be the best to make money. You only need to be better than some. So don't feel discouraged! Also, after finding a good offer and sending some leads, you can negotiate a better payout.

You can bid lower than they do and hussle harder. Test more angles for example. By the way - have you found some good angles yet? In adult, banners are the KEY (landers don't change much). And because banners burn out really fast, you need to know what the good angles are so you can keep churning out new ones - rather than relying on a random good banner.

Once you have enough good banners and doing more traffic volume, try to rotate them so you're using a different batch every day. This will make them last longer.



Amy

---

I said the A and maybe forgot to add the B, so of course I dont recomend this "sense" based decisions to new people. The sense is basically just a skill that allows you to spot statistical significance sooner than it happens - or in other words, when after 500 clicks, you see one option killing it, even tho its not significant yet, after some time you will realize that in 95% cases (when certain conditions are met) its gonna stay that way and you dont need to wait for statistically significant 1000 clicks to prove it

---

@vortex, @matuloo; thank you very much about putting your valuable opinions. It's a great pleasure read your different viewpoints.

I think, for myself actually, I need significance test til I build a sense. However, I need long time to learn that sense thing. Because I won't feel myself in comfort if I act without scientific ways. This cost more to get every other thing approved by statistic. It makes harder, but hey I'm in.

@vortex, Amy, again great tips.

• I'll also check offer LPs GEO based to see if results are still same.
• I've prepared top 3 angles and 3 banner for each angle. Adding new 9 banners today.

---

This thread has dead...

There will be a better Follow-Along soon.

---

All your posts was very interesting, I can't wait to see the next one.

Anyway, vortex and matuloo are full of awesome informations, so good to read and learn.

---

Home > Adult Traffic (NSFW) > Follow-along Campaigns