Home > >

---

Understanding Banner-Burnout - Still getting clicks but the CVR goes down and down? (22)

---

Anyone who has done some volume in adult, knows this situation : campaign get's profitable, ROI is nice, but at some point the CVR starts to decline and nothing you do with the bids helps. Your affiliate manager assures you there is nothing wrong with the offer, but your CVR still goes down. Chances are, the campaign is dying the slow death by banner burnout. But the banner CTR is still more or less the same, you might say, wouldn’t it mean the banner is still fine? Unfortunately not, the tricky part about banner burnout is, that sometimes the change in CTR isn’t really noticeable.

I've been trying to wrap my brain around this issue for a looong time when I started with paid traffic in the adult niche. Why the hell do they still click the banner, but don't finish the registration? Why the hell did it work like charm just 3 days ago? I wasted too much time on finding the answer, so to save some of yours, let me tell you what I came up with

Affiliate marketing is like fishing, you need the right lure to catch the fish. Some lures get the fish interested so they come close to look at it, some tick it, but they don't bite. Then there are lures that actually make the fish bite. One thing to keep in mind, NO lure is universal, there is a limited amount of fishes that you will catch even with the best lure. Once you catch certain % of interested fishes in a lake, the lure will become less and less effective up to the point where it's no more viable to use it.

But how does it translate to affiliate marketing? Lake is the site you're targeting, lure is your banner, fishes who just tick the lure are surfers that will click your banner but won't convert and the fishes who bite are surfers you can actually convert with the banner. The most important part of an adult dating banner is the woman you put on it, the surfer must find the woman so attractive or attainable or whatever, that he will get through the registration process to be able to meet her or see more of her. Since human taste can vary so much, you can never find a woman that will suit the taste of everyone, just like you can't make a lure that all fishes will bite on.

To be successful in adult dating, you need to learn to spot woman who will appeal to as wide audience as possible, by doing this, you are effectively increasing the amount of catch-able fishes in the lake you are fishing at. You need to learn the difference between attractive women and irresistible women. Attractive women will get you a lot of ticks on the lure (CTR) but irresistible women will get you a lot of bites (CVR).

No matter how good looking the woman is, at some point there simply isn’t enough surfers available who still find her irresistible. There are still surfers interested enough to click the banner that’s why you can still see good CTR, but you have already converted (or shown the banner to) too high % of those that you could. You will still be getting some conversions, because you simply can't catch ALL the fish in a lake, but the CVR goes down to the point where its no more financially feasible to continue running the banner. That’s when it has been killed by banner burnout.

There are certain steps you can do to prevent banner burnout from happening too soon.

1. Fish in large lakes, in other words, target large sites or many sites at once. Avoid too specific targeting.
2. Use low cap, 1-2 in 24 hours max.
3. Don't try to be the highest bidder at all costs.
4. Always run several banners in rotation
5. Test new banners all the time, so you always have new ones to use when needed.

Questions? Comments? Agree? Disagree? ... Post a reply.

---

Awesome, mate. It's sucks really bad when it happens. What sucks more is the inconsistencies depending of the day of the week in adult :/

You're saying to test new banners all the time, but I have a question: to get statistical significance you have to wait some time, and let's say that i'll get 15% of the traffic with a DOI offer in a placement in a tier 1 country in europe (traffic junky). How do you approach testing banners? Do you test the same banner in a lot of placements at the same time or in a single placement and wait a couple of days? This is intriguing me since I started!

---

Hello, yes, the day-to-day roller coaster drive makes me crazy at times too

I'm testing banners in high trafficked spots only, that way I don't have to wait too long for them to show their potential. Since I have proven banners in the campaign already, it's sometimes literally a question of a few hours to see if the new ones have potential or not. I mean, if you run several banners head to head against each other, and the proven ones are giving you leads just fine and the new ones simply blow, then you don't really need to wait too long. In case the results are promising, then I wait as long as needed to really decide whether the new ones won over the older ones. This can be a pain with smaller GEOs, there you really have to wait a few days, but thats how it is

To limit the losses of testing, I keep the number of tested banners low. I always run with several proven banners, so let's say I have 5 proven banners in rotation, I would not add more than 2 into rotation for testing purposes. However, when all the banners start to go downhills, then I throw up a large batch of banners and I test across several placements in multiple campaigns.

---

You can offset a great deal of banner blindness by dynamically rotating presells. It prolongs the campaigns by a factor of multiples.

Some banners are more prone to blindness than others - some have great initial result then drop like a bomb - especially on mindgeek sites; others are a lot more stable they can run for months and still deliver consistent results. It makes sense to look at the data sepearately. 1st week of running and data after 1 week of running. The intial winners might be in a death spiral of loss which you are unable to spot unless you discount the 1st week of data.

Most of the banners can be resurrected by just changing the image on the creative- while keeping the text and cta part the same. Its a cheap method to keep them running for longer.

Statistical significance ? LOL

---

Yup, image is the main element when it comes to banner-burnout, irresistible chick is the queen

Rotating pre-sells - I assume you mean landing pages - didn't really prove to be of much help to prevent banner burnout according to my own tests. Sure it will help some, but didn't seem to have such a huge impact as you say.

---

Thanks a lot for the explanation matuloo, I can understand more clearly the banner burnout now.

I've been experiencing this since many years on my websites, without understanding it and and even without imagine it can be this.

As you already said, the key is to always test new banners to get ready when the best ones are dead.

But when one banner is burned out, do you think the template/headline can still be used with another woman's picture ?

---

In adult yes, its about giving them the woman they simply need to go after from a large part.

---

great post, was thinking about something similar and remembered this post.

and guys matuloos wisdom doesn't just apply to dating banners but to everything in marketing. just substitute woman for angle/appeal

to add my 2 copper: it is known that people only see and read what they want to see ie. what is on their minds already. let's say we go back in time and we start a campaign saying "ur battery is damaged by viruses. download this app now". now just like with matuloos woman example, people who have issues with their battery ie those who have this topic on their mind will notice that ad and respond. but eventually we catch them all. even in big markets. in germany every day 2000 people or so are being born but millions live already. so even if you target really broad like whole germany, eventually everything will burn out.

so how would we promote our antivir app? luckily there are still TONS of people who have very different things on their minds aka fish who like other lures. an example would be to target gamers and saying "ur hearthstone account may be vulnurable to attacks". different fish, different lure and because this stuff is already on their minds it gets their attention and hits them emotionally.

I stopped thinking in terms of markets and started to think in terms of segments. even if you show ur ad to the whole "market" it will only reach a certain segment ie only the fish that like that lure. this is my theory based on the campaigns I have run so far.

thanks for posting this.

---

Nice, Im glad to see someone with the same way of thinking when it comes to this issue - I really like the example I used - fish/lure, I think it really relates to AM in a lot of ways

---

Absolutely agree, we should aways remember or banner burnout.
Couple banners in rotation during the same period or total replacement of ads will be great idea.

---

This is interesting, I hadn't thought of this. Seems like a really good idea.

How many times do you guys re-test your banners with new images?

Do you only re-test banners that performed bad on the initial test?

Say one of your banners did well, would you take the image on the banner and test it on multiple banner designs with new headlines, CTA's, colors etc... to see if the image was the reason for getting good CR, Clicks etc...?

Hope you guys don't mind all the Q's, I'm just super curious about this since right now I'm testing banners and if they don't do so well I just toss them aside and forget about em. I go into Photoshop and create entirely new ones, with new images, headlines, CTA's, design, color etc... Which kinda sucks because I think the headlines or the banner design itself was decent and it sucks to just forget about it once I test it and it tanks on the first run.

Look forward to a reply!

Thanks & as always, great post Mauloo.

---

Do you only re-test banners that performed bad on the initial test?[/QUOTE]

There isn't really a limit or a set amount of times you should do this. However, this write-up was geared towards campaigns that once worked well, but stopped to. So it's not about retesting banners that failed initially.

Yes, this is one test we are frequently doing. When we come up with a banner that rocks, I want to know what's the key element for such great performance. If it's the image, we also use it on LPs in many cases and obviously make many different banners with it.

This is not the best way to approach banner testing actually, you need to be more systematic. There are basically 3 main aspects that determine how good a banner will perform : 1. angle, 2.imagery, 3.layout

The way you are doing it now, is that you change all 3 aspects at once, which makes it hard to determine where you need to improve. Imagery and Angle are the most important, the layout comes after. So try to sort out the first 2 aspects first.

The way I'm doing this, and it works fine, is to make banner variations by changing just one aspect at a time. So lets say I start with 10 banners, I pick some angle and some layout, but I chose 10 different images. 2 or 3 images will stand out, so I test several angles with these images, let's say 3 or 5 angles per image, depending on how many images I'm testing. At the end I should have a set of good images and good angles. Then I can start playing with the layout. Try to make a system similar to this one.

---

Good to know, I will definitely keep this in mind for future campaigns.

Damn. I had a strong feeling I was going in the wrong direction with my current way of creating / testing banners.

Okay, so I should be more methodical with how I create & test banners. So I'm guessing no one on STM is willing to share their complete way of creating / testing banners so I should come up with my own system. Would something like this be okay....

Phase 1: Find a good angle. - (all banners in this phase will look the same, only thing diff would be the image / headlines, but they will be relevant to the angle)

So I would begin by testing multiple angles to find one that works. Once I have my angle I would move onto phase 2.

Phase 2: Test images. - (all banners in this phase will look the same, only thing diff would be the images / headlines, but they will be relevant to the angle)

So I would look for say 10 images to start. I would run them alongside each other in one campaign using the same banner, everything in the banner will be the same only thing different will be the images / headlines. Once I've found the best image/s I move onto phase 3.

Note: I have now found the most important aspects of my banners, which are the angle and images. Also, at this time, my banners have ran with the same layout for about 2/3 days so banner blindness should start to kick in, I think.

I'm unsure about which should be Phase 3. Can't decide if layout or headlines should be the next phase. (You mentioned layout as 3rd aspect, but I think headlines are a little more important, am I wrong?)

Phase 3: Test the layouts.

I create multiple layouts to test, I will use the images I gathered from phase 2 and test like 20 diff layouts. Grab the best then move onto phase 4.

or

Phase 3: Test headlines.

Brainstorm multiple headlines, all headlines related to my angle of course.

Phase 4: Test petty shit like CTA, Fonts, Color etc...

Am I going in the right direction here with this way of testing? Is there a better way of systematizing the process. Please let me know what you think.

As always, thank you so much for the help!

---

That's pretty solid as a testing strategy. You're testing the right stuff in the right order.

Only thing I'd change is I'd probably recommend testing headlines before layouts. But tbh that's subjective. Either can work.

---

Hey Caurmen,

Thanks for your input.

Reason I think layout is also important to test in the early phase is because when I begin testing the banners, they'll all be similar in layout, (color, image positioning, headline positioning, etc...) - by the time I'm done testing the angle and image phase if I move on to headlines that would mean the layout stays the same. At this point I'm already 2-5 days in with the same banner layout (depending on amount of traffic I can get), wouldn't banner blindness settle in by that point and mess up my data?

Or am I safe to run the same banner design for 3-4+ days (24/7)?

If you do think it's safe to run the banner with the same layout for that long, then I will definitely prioritize headline testing over layout testing.

I know this is something I could just test and figure out for myself, but it doesn't hurt to ask you guys which are more experienced at this.

Thanks for all your help, really do appreciate it!

---

I'd be very surprised if banner blindness set in that fast, given you're varying images and headlines. You should be fine to run them for longer than that before testing layouts.

---

Thank you so much for your response, Caurmen!

---

Banner layouts don't tend to burn out fast, if at all. I'm using same layout in many GEOs and verticals for many many months, just replacing the images and playing with the color schemes - I mean big color changes here, from yellow to black and then to blue ...

---

You mentioned to test several banners and always test new banners.
I have a small doubt here.
Let's say we have 1 campaign and it has 5 proven banners.When testing new banners,do you clone the existing campaign to test new banners or else you will use the same existing campaign url to test new ones?

---

One method you can use, is to dedicate small part of your profits to testing new stuff ... so in your case, you could let the 5 good ones run and add one more for testing. With sources that allow weighted distribution, you can simply dedicate 10% or so to testing new banners. Just make sure you don't blow too big part of your profits on testing new stuff.

---

Thanks for replying matuloo.
Based on your answer,you are telling me to use the existing campaign.
Looks great.

---

Yes, because the campaign itself already has a proven funnel - so use it and find more banners that would work with it.

---

Home > >