Home > General > Affiliate Marketing Forum

HOW TO optimize a campaign step by step - this is one of my methods. (59)

06-30-2016 02:59 PM #1 matuloo (Legendary Moderator)
HOW TO optimize a campaign step by step - this is one of my methods.

There has been a lot of threads and posts asking for optimization methods and what is the right way to do it. So instead of replying to the same question in multiple threads, let me explain one of my methods here and let's keep the related discussion in this thread too. I will include answers to some of the questions posted in other threads too.

This method is based on cutting the WORST elements one by one. It works the best with low paying offers, because we need a larger amount of conversions in order to really be able to spot the weak parts of the campaign.

I started to use this method when I realized it's not a good idea to cut placements in the early phase of the optimization, because it almost always resulted in minimum volume and low revenue. I was able to get nice ROI but I was making pennies.

Since I'm buying mostly banner traffic, I will also use it in this thread. You can still use this method when running POPs, just ignore the parts where I talk about banners.

The method has a few steps :

1. The first thing we need to do is to make sure we are not buying fake/BOT traffic or traffic that is simply too expensive to possibly turn profit. To judge this, you need to have some experience with the possible offer performance so you can make a ROUGH estimate of EPV (earning per visit).

Example: If I know that an offer is highly unlikely to convert better than 1:50 and the payout is $1, I cannot buy clicks that are more expensive than $1/50=2 cents. Otherwise I can't be profitable. I also know that I can increase the CVR somewhat by optimization or find a better offer, so I will not use those 2 cents as the cutting line, but I will increase it to 3 cents. I also know that some sources might perform way better than others, so let's increase it to 4 cents per click max.

The first thing I will do is run a BOT test and I cut everything that has 80% or more BOTS – only cut the WORST, this is the key, so based on the traffic profile you can increase the limit to 90% even. This is something you need to judge on a case by case basis – check what the numbers look like and get rid of the very worst.

The next thing is to look at the banner CTR and remove every banner that is giving me clicks more expensive than those 4 cents I calculated.

NOTE: this phase is not about finding a winning lander or offer, it's just about identifying BOTs and banners with lousy CTR. So don't worry about slower loading times due to the bot test script. You'll look for winners later on.

2. Now I know I'm not paying for the worst spots, so the traffic is legit from a large part and the clicks are not too expensive. At this stage, I want to figure out which of the offers and LPs is performing the WORST. I need to get enough conversions so I can spot the WORST ones again. The amount of conversions needed depends on how they are distributed across the various LPs and offers. Sometimes you see a clear winner after 10 conversions when it looks like 8 to 2, sometimes you need to run longer. You don't need to find the clear winner again, you want to identify the worst ones.

The best that can happen at this stage, is to retain at least 2 offers and 2 LP's that are showing similar performance. Once I have them, I need to let the campaign run for some more and collect data for just these “better” LPs and offers. It's perfectly fine to be still in the RED at this stage, thou you can sometimes already see green here.

3. Now comes phase 3, I have had multiple conversions for all the LPs and offers I'm still running and now I need to look at the data and look for the WORST segments once more. At this phase, you can start by cutting whatever targeting parameter you want, except for the placements, these will come as the last one. The key is to cut only the worst again and one by one if possible. Sometimes you will be able to cut 2 at once when the data tells you so.

So how do you find the targets to cut? They have to be way worse than the rest. In case you see Android gave you 25 conversions and iOS just 10 – this is NOT the signal to cut it. But if Android made 12 conversions and iOS did 0 or 1, now this is a clear signal to cut it. You need to drill down into the data quite deep here and look for interrelated signals. IOS making low amount of conversions? Drill down deeper and check wifi VS 3G, maybe you will see wifi works just fine with iOS but 3G made 0 conversions – so you want to retain iOS but you don't want 3G iOS. 3G and WIFI seem to work comparable? How about you check the browsers – maybe it's some browser type that takes down the CVR so cut it.

NOTE: In some cases, it will be impossible to set the targeting to match what you want it to do. For example, you can either target 3G for the whole campaign or not. It's not possible to target Android 3G and WIFI and just 3G for iOS in one campaign. In this cases, you will need to clone the campaign and target Android in the original one and iOS in the clone.

Hunt for the WORST, keep the average ones running for now, that's the name of this method. Every time you cut something, make sure to run the remainders for some time and look at the data again. And only when you used all the targeting options, then look at the placements and cut those that are simply not working for you. At this stage, you should be starting to see some green or at least be very close to it, otherwise the offer/banners/LPs simply won't cut it and you need to improve them or find better ones.

This is it pretty much, this method isn't the only one I'm using but it's one of my favorites for low paying offers. Any questions?

06-30-2016 03:10 PM #2 onleads (Member)

Hey Mat(can I call you Mat?)

Thanks for the info. This guide really save us a lot of money and time. At least it is a guideline for us to learn and build up our own system.

If I know that an offer is highly unlikely to convert better than 1:50 and the payout is $1, I cannot buy clicks that are more expensive than $1/50=2 cents.

So the ratio 1:50, is something we only know through experience?

NOTE: this phase is not about finding a winning lander or offer, it's just about identifying BOTs and banners with lousy CTR. So don't worry about slower loading times due to the bot test script. You'll look for winners later on.

Sound like you implement the bot test code into the lander? or add it as 1 of the lander in Voluum?

At this phase, you can start by cutting whatever targeting parameter you want, except for the placements, these will come as the last one.

This sentence sound a bit weird.... I dont understand

06-30-2016 03:11 PM #3 Mr Green (Administrator)

"only cut the WORST." This is so key

A lot of people hope for one big winner or two off the bat, the better strategy is actually just paying attention to the tail end.

06-30-2016 03:22 PM #4 tpm767 (Member)

This is great and extremely helpful for me in the early stages of my affiliate career, so thank you!

One question - is there an easy way to run a "BOT test"?

Thanks!

06-30-2016 08:30 PM #5 matuloo (Legendary Moderator)

Originally Posted by onleads

Hey Mat(can I call you Mat?)

Of course

Originally Posted by onleads

Thanks for the info. This guide really save us a lot of money and time. At least it is a guideline for us to learn and build up our own system.

So the ratio 1:50, is something we only know through experience?

Yes, you will figure out what to expect once you run enough campaigns. Just follow what the data is telling you.

Originally Posted by onleads

Sound like you implement the bot test code into the lander? or add it as 1 of the lander in Voluum?

I have a BOT testing lander that I use in the first phase when running on a new source, once I have an idea about the bot %, I replace it with a normal LP.

Originally Posted by onleads

This sentence sound a bit weird.... I dont understand

I wanted to say that you can start optimizing all the other things like OS, Carrier, WIFI ... in the order you see fit based on your data. And keep the placement cutting as the last step.

06-30-2016 08:32 PM #6 matuloo (Legendary Moderator)

Originally Posted by tpm767

This is great and extremely helpful for me in the early stages of my affiliate career, so thank you!

One question - is there an easy way to run a "BOT test"?

Thanks!

Read this : http://stmforum.com/forum/showthread...n-Any-Campaign

I'm using a slightly modified variation.

06-30-2016 09:24 PM #7 glennb (Member)

Hey Mat,

In phase 3, when you start cutting the WORST placements, how does this affect your bid? With every cut, do you up the bid?

06-30-2016 10:24 PM #8 simon_89 (Member)

Originally Posted by matuloo

So how do you find the targets to cut? They have to be way worse than the rest. In case you see Android gave you 25 conversions and iOS just 10 – this is NOT the signal to cut it. But if Android made 12 conversions and iOS did 0 or 1, now this is a clear signal to cut it. You need to drill down into the data quite deep here and look for interrelated signals. IOS making low amount of conversions? Drill down deeper and check wifi VS 3G, maybe you will see wifi works just fine with iOS but 3G made 0 conversions – so you want to retain iOS but you don't want 3G iOS. 3G and WIFI seem to work comparable? How about you check the browsers – maybe it's some browser type that takes down the CVR so cut it.

So, I have this one Indonesian campaign that pays on the pennies per every lead generated. I have ran this offer for about 2 weeks now, and narrowing down data here there I always end up with these variations. This is targeted differently between Android and iOS. But, Android definitely has more volume. Here is a list of combos I considered

1) Connection Type(Mobile or Wifi) ---- OSName ---- Browser Name
2) Mobile Carrier --- OSName --- Browser Name
3) ISPName --- OSName --- Browser Name
4) OSName --- BrowserName --- ISPName

All of these combos kind of lead to the same results. One of the highest volume carriers is Telkomesel with the highest OS volume being Android 4/4 and highest volume browser being Android Stock Browser. None of these combos are generating a positive ROI or nowhere close. The payout is less than $0.10. In this case it's probably sensible to improve upon my winning lander in hopes to get a better conversion rate. Because that's the only thing I see that I have in CONTROL of. I've already tested the similar offer from multiple affiliate networks. Here are some of the results from the combo below. This is only targeting Android.

Originally Posted by matuloo

I have a BOT testing lander that I use in the first phase when running on a new source, once I have an idea about the bot %, I replace it with a normal LP.

Does this mean you implement the bot script on every single lander you test in the beginning and run a sample test of say 10x offer payout and then you start blacklisting placements. Then on the next test you get rid of the bot script and start finding testing offers/landers.

Plus when you're done testing landers/offers, you don't look at the past data to optimize because it's skewed due to the offers/landers/bot testing that you're doing. Rather your approach will be to start looking at the placement/targeting option performances once you get winning lander/offer.

06-30-2016 10:59 PM #9 matuloo (Legendary Moderator)

Originally Posted by glennb

Hey Mat,

In phase 3, when you start cutting the WORST placements, how does this affect your bid? With every cut, do you up the bid?

No, I don't up the bid with every cut. It depends on what the campaign looks like, I will of course try to play with the bids at some point if I feel higher bid might help. I usually try this before I start to cut the placements.

06-30-2016 11:05 PM #10 matuloo (Legendary Moderator)

Originally Posted by simon_89

This Indonesian campaign of yours looks like something that's not gonna work with the current setup. Hard to say what's wrong with it, maybe the offer just sucks or you need to improve the creatives ... maybe it's simply a flop and you need to pass.

Originally Posted by simon_89

Nope, I don't run the BOT script on every lander, that would be an overkill. I'm only testing a source for BOTS when it's new for me or I haven't been running there for a while ... then from time to time again or if there is something suspicious. Once I catch the worst placements I simply blacklist them on new campaigns I start on the source straight away.

And yes, part of the data collected during this bot test is worthless so to speak.

06-30-2016 11:23 PM #11 simon_89 (Member)

Originally Posted by matuloo

When you're using a lander with the BOT script, is it actually a lander or is it just a simple redirection html code.

07-01-2016 04:16 AM #12 onleads (Member)

I have a BOT testing lander that I use in the first phase when running on a new source, once I have an idea about the bot %, I replace it with a normal LP.

How much you usually spend for this purpose? I only spend 5 dollar for this step. Do you think this is enough?

07-01-2016 05:54 AM #13 eddiezhan (Member)
HOW TO optimize a campaign step by step - this is one of my methods.

Originally Posted by matuloo

Nope, I don't run the BOT script on every lander, that would be an overkill. I'm only testing a source for BOTS when it's new for me or I haven't been running there for a while ... then from time to time again or if there is something suspicious. Once I catch the worst placements I simply blacklist them on new campaigns I start on the source straight away.
And yes, part of the data collected during this bot test is worthless so to speak.

Hey Matuloo, I've been running the bot lander and actually changed the 300ms to a 50ms redirect to speed up load speeds. I tested the 2 against each other and they give almost the exact same results so I'd recommend changing to 50ms just purely for faster load times.

Also I've noticed that the bot script works very differently on different traffic sources. For example, on adCash average CTR (human traffic) is 80ish% while propeller is 50-60ish%. On adCash I even have some placements with over 100% which leads me to believe it doesn't do too great a job of catching the bots...

And @onleads, medium sized geo on average volume source you gotta blow $20-30

Peace

07-01-2016 09:29 AM #14 matuloo (Legendary Moderator)

Originally Posted by simon_89

When you're using a lander with the BOT script, is it actually a lander or is it just a simple redirection html code.

The lander just redirects yes.

07-01-2016 09:31 AM #15 matuloo (Legendary Moderator)

Originally Posted by onleads

How much you usually spend for this purpose? I only spend 5 dollar for this step. Do you think this is enough?

I spend more, with broader targeting and larger GEO, $5 is not enough. I would say 20-30 is about the minimum to spot the biggest budget drainers.

07-01-2016 09:39 AM #16 onleads (Member)

Originally Posted by matuloo

I spend more, with broader targeting and larger GEO, $5 is not enough. I would say 20-30 is about the minimum to spot the biggest budget drainers.

Wow, that is a lot.

Are you testing using the bot test method on this forum? As i know it is a plain page and not a offer lander. Spending on a plain page like this, ouch....

Am i have some kind of misunderstand here?

07-01-2016 09:59 AM #17 matuloo (Legendary Moderator)

Originally Posted by eddiezhan

The script is not a perfect solution, but it's one that works to certain extent and it's very easy to use. Bots are designed in many different ways, some can pass this detection, but some can't. That's why it's always important to look at the whole picture and look for placements that simply stand out from the rest too much and cut those.

07-01-2016 10:05 AM #18 matuloo (Legendary Moderator)

Originally Posted by onleads

I have something similar yes, the blank page is a redirect, you can then send the users to a real LP or directly to the offer. So I can actually test whether it can work directlinked at the same time

You're not throwing the traffic away.

07-01-2016 10:13 AM #19 onleads (Member)

Originally Posted by matuloo

I have something similar yes, the blank page is a redirect, you can then send the users to a real LP or directly to the offer. So I can actually test whether it can work directlinked at the same time

You're not throwing the traffic away.

I see. You are smart. Lol

So you will redirect a visitor after 300ms?

07-01-2016 10:15 AM #20 caurmen (Administrator)

To add to that - what the bot test does, if the user is not a bot, is simply add about 500ms delay (300ms in Javascript plus connection time) before forwarding them to a real LP or offer.

That will definitely cost you some ROI so you shouldn't do it forever, but it won't be enough to make those clicks worthless.

07-01-2016 10:23 AM #21 matuloo (Legendary Moderator)

Originally Posted by onleads

I see. You are smart. Lol

So you will redirect a visitor after 300ms?

300ms is a value you can use or change it, it's a value Caurmen came up with

07-01-2016 10:34 AM #22 caurmen (Administrator)

The reason for 300ms is it's just below the average human reaction time. So it gives the bots enough time to disconnect, but shouldn't lose any humans.

07-01-2016 10:38 AM #23 onleads (Member)

Originally Posted by caurmen

The reason for 300ms is it's just below the average human reaction time. So it gives the bots enough time to disconnect, but shouldn't lose any humans.

I am going to use your setting, 300ms.

07-01-2016 11:00 AM #24 eddiezhan (Member)

Originally Posted by caurmen

The reason for 300ms is it's just below the average human reaction time. So it gives the bots enough time to disconnect, but shouldn't lose any humans.

I use 50 ms and I split tested it against 300ms. No difference

07-01-2016 04:24 PM #25 vyapari (Member)

A great post! Must be read by all.

07-01-2016 08:15 PM #26 matuloo (Legendary Moderator)

Originally Posted by vyapari

A great post! Must be read by all.

Thanks, glad you liked it

07-04-2016 11:41 PM #27 affpayinggao (Veteran Member)

Thanks for the great post! learned a lot!

07-06-2016 07:15 PM #28 grandslam ()

Haha matuloo you just answered 70% of all conundrums here... If you keep this up, this place is gonna be crickets No seriously though this is a great and well thought out column. This is great not jsut for newbies but I learned something too

07-06-2016 08:35 PM #29 matuloo (Legendary Moderator)

Originally Posted by grandslam

Haha matuloo you just answered 70% of all conundrums here... If you keep this up, this place is gonna be crickets

No seriously though this is a great and well thought out column. This is great not jsut for newbies but I learned something too

I'm surprised this thread didn't get more traction to be honest, but I hope people will find it eventually, sooner or later

07-07-2016 03:40 PM #30 grandslam ()

Originally Posted by matuloo

I'm surprised this thread didn't get more traction to be honest, but I hope people will find it eventually, sooner or later

Just link this thread to a post asking about this

05-23-2018 09:51 AM #31 silvertongue (Member)

Originally Posted by matuloo

Yup, there is a lot of info at first and I understand it helps a lot to get a direct answer for a specific question, so keep em coming

Will do sir, will do ...

Btw, didn't know you had blog. I've just read your latest post on cryptocurrency and really enjoyed it. Looking forward to reading more. Well done!

05-23-2018 10:15 AM #32 matuloo (Legendary Moderator)

Originally Posted by silvertongue

Will do sir, will do ...

Btw, didn't know you had blog. I've just read your latest post on cryptocurrency and really enjoyed it. Looking forward to reading more. Well done!

Yup I do, but didn't post much there lately ... have to force myself to start posting again soon

11-09-2019 02:54 AM #33 j80montes (AMC Alumnus)

Hi Matej,
This is the best optimization method, I have read it again and again.
I am running adult campaign, still confuse in a few things.
1. When to optimize banner? I think should be optimize banner on phase 3?

2. How many banners and how many types of banner we are looking for?
We are looking for 2 offers and 2 landers, how many banners we a looking for? I think should be greater than 1. But due to the blindness, if the answer is greater than 1. What types of banners we are looking for? 1 image with diffirent angles or 1 angles with diffrent image? Or the image and angles are totally diffirent?

3. How to optimze banners?
Even though every decision base on Statistical Significance, But it's very tricky in optimize banner, due to situation such as:
a.The best banner today, tomorrow will be the worest one, and several day later, it will be the best one again.
b.How should we rotate the banners? base on CTR or CR, or Evenly? Diffirent method will impact the data directlly.
c.In the most case, how many days can a banner live before be kill by blindness?
d.the banners works for this site, but maybe will perform suck for the next one.

11-09-2019 07:33 AM #34 cpa2020 (Member)

Originally Posted by matuloo

Since I'm buying mostly banner traffic, I will also use it in this thread.

Please name some good Traffic Sources for banner traffic.

11-11-2019 09:44 AM #35 matuloo (Legendary Moderator)

Originally Posted by cpa2020

Please name some good Traffic Sources for banner traffic.

Right now, the best one is probably tTrafficfactory... best overall quality imo. Trafficjunky, Trafficstars and Trafficforce are also good, then there is Exoclick which has the largest volume and largest amount of publishers, which makes it a bit harder to optimize as the quality varies from one publisher to the next. You can also try Traffichaus for flat buys. And there is also a number of smaller ones like juicyads, ero-advertising etc ...

11-11-2019 10:16 AM #36 matuloo (Legendary Moderator)

1. When to optimize banner? I think should be optimize banner on phase 3?

Yup, try to sort out bot traffic and the offer first, then move on with banners or LPs. Bot traffic won't convert and even if it does it's fake signups, so you don't want that. Bad offer wont work well even with the best banner, so you need converting offers first. Then you can move on to the other parts that you can control.

2. How many banners and how many types of banner we are looking for?
We are looking for 2 offers and 2 landers, how many banners we a looking for? I think should be greater than 1. But due to the blindness, if the answer is greater than 1. What types of banners we are looking for? 1 image with diffirent angles or 1 angles with diffrent image? Or the image and angles are totally diffirent?

You can do this in several ways. You can decide to test a specific angle or a specific image or mix it up a bit, I don't think any of the approaches would be the best.

When testing an angle, keep the same copy and make several banners with it, just use different images for each. In case you believe an image is the right one, make several banners with the same image and use different copy on each of them. In case you have no idea what could work, just make banners with different copy and images and if after some initial traffic, something seems to work great, focus on it and make variations with either the same copy or image.

At the beginning, it might be the best idea to simply come up with a mix and try to find a winning angle or image.

3. How to optimze banners?
Even though every decision base on Statistical Significance, But it's very tricky in optimize banner, due to situation such as:
a.The best banner today, tomorrow will be the worest one, and several day later, it will be the best one again.

True, there will always be fluctuations for sure. The stronger the ads are, the more stable the performance will be. Same with volume, low budget campaigns that only run during certain parts of the day, will be less stable than high budget 24/7 campaigns.

How should we rotate the banners? base on CTR or CR, or Evenly? Diffirent method will impact the data directlly.

I pretty much always rotate evenly, unless a specific banner performs extremely well. And when that happens, I just pause the others and push the best ones heavily. And the metric I go by is almost always the ROI, that's the one that matters the most.

In the most case, how many days can a banner live before be kill by blindness?

Depends on how strong the banner is. Sometimes it's just a few days, but I had banners that gave me pocitive ROI for a few months.

the banners works for this site, but maybe will perform suck for the next one.

This happens too yes, different sites have different audience so the banner performance can differ too. But usually, a strong banner will perform decently on any site.

11-11-2019 12:12 PM #37 j80montes (AMC Alumnus)

Originally Posted by matuloo

Thanks, but I still cant' understand how many banners are we looking for? For example if I have banner A is green, banner b and banner c is breakeven, the others are red, should we only keep the green one for profit?

11-11-2019 12:54 PM #38 matuloo (Legendary Moderator)

Originally Posted by j80montes

Ah sorry, forgot to address that part.

Again, there are several ways of handling this, but let me write what I would do.

When I start a new campaign in a new market, GEO or whatever, I like to have at least 10 banners, so I can test faster and get some results asap. When you start with just 2 or 3 banners, chances are you hit all bad ones and the results won't help you. SO starting with more banners is better. Then again, if you put too many in rotation, the test budget needs to be higher. So to make a conservative bet here, I'd say you should start with 5 or so.

Once you narrow down the "better" ones and you end up with a situation like you described... I would keep the 3 banners (profitable + the breakeven ones) and ditch the red ones. The reason to keep running the breakeven banners too, is that due to the nature of traffic rotation, the breakeven banners might have received some traffic that wouldn't convert with any banner (bots, missclicks, just curious people who are not that interested etc...). So by keeping them in rotation, you give them another chance to prove themselves with a better, more balanced sample of traffic.

In a case where the profitable banner totally slayed the other ones, you should keep just that one and make variations of it. There are 3 possible scenarios here, the image was the winner, the copy was the winner or both the copy and image simply clicked well together. To find out what the case was, you should make let's say 2-3 variations with the same copy but different images, 2-3 more with the same image but different copy and rotate them evenly along with the original winning banner. Run this for a while and you should get an idea about why the banner worked well, then you can replicate this further.

11-11-2019 01:28 PM #39 j80montes (AMC Alumnus)

Originally Posted by matuloo

Thanks Matej,

Finally I know what I am looking for now, but one more little question, if start with 10 banners, how much budget should we set for each banners? 1x payout or 2x payout, or 3x payout? And after spending what kind of budget we can point out the profitable banner definitely will be final king?

11-11-2019 04:17 PM #40 matuloo (Legendary Moderator)

Originally Posted by j80montes

The very worst banners will show themselves quite soon, in some cases sooner than you spend 1x offer payout on them. Look for those with extremely low CTR and LP CTR, compared to the rest. It's simple math, in case a banner delivers too expensive clicks, it can't possibly become green.

For the better ones, spend 2x payout, then look at the performance. Those that converted at least ones should make it to another round of testing, unless you have much better ones in the mix. Remember that the key with this method is to drop the bad apples, one by one, so always look at the performance of all the ads/lps/offers ... if something clearly falls behind, cut it.

As for when to pick the final winner... well, the goal is to find funnels that are profitable. Once you reach profit, you kinda have your winners. Sometimes you will be lucky and hit the jackpot straight away, sometimes you will have to ditch multiple sets of ads and keep on testing for days. It's really hard when to say for sure that a banner is the winner or you need to cut the whole thing. I certainly wouldn't make any strong/final judgements sooner than after 10x payout, you simply need to give it time and enough clicks.

Just a note: keep in mind that miracles do not happen in AM, so in case you see very poor performance for all the ads or LPs, it's pointless to keep them running and hoping for some sudden turn. If all the ads or LPs perform poor and nothing stands out, it's time to tweak the funnel again or find a different offer.

11-13-2019 02:08 PM #41 j80montes (AMC Alumnus)

Originally Posted by matuloo

Wow, I have an clear idea now, thank you so much.

By the way, do you have ever found that some banners works for a while, and drop after a few days, then I paused them for a few days. And open them again, these banners works again, do you have any idea to handle this?

11-13-2019 08:39 PM #42 matuloo (Legendary Moderator)

Originally Posted by j80montes

This happens all the time too. When I have a good banner and it starts to lose steam, I pause it and return to it in a few months, in many cases it performs well again.

But I guess you're talking about banners that work for a day, then stop, then work again... right? This is because they are not that good match with a wide-enough audience. So on some day, they happen to be shown to the "right" people and they convert... on the other day, the demographics changes and the banners lose performance. The only way to solve this is to find out what kinda audience they work with and target just those people. It could be a specific niche interest, time of the day or a specific place in the bidding chain. It's kinda hard to use these successfully for a long period of time.

It can also be related to what kind of traffic you're buying in general on the specific days... so the bidding position, specific placements... if it's the case, it's not a question of the banner and any ad would show this erratic behavior.

11-14-2019 12:30 AM #43 j80montes (AMC Alumnus)

I see where is the point now, I need to find the stronger banner as possible to match wide audience.

11-14-2019 12:40 AM #44 j80montes (AMC Alumnus)

Originally Posted by matuloo

[B]
1. The first thing we need to do is to make sure we are not buying fake/BOT traffic or traffic that is simply too expensive to possibly turn profit. To judge this, you need to have some experience with the possible offer performance so you can make a ROUGH estimate of EPV (earning per visit).

Example: If I know that an offer is highly unlikely to convert better than 1:50 and the payout is $1, I cannot buy clicks that are more expensive than $1/50=2 cents. Otherwise I can't be profitable. I also know that I can increase the CVR somewhat by optimization or find a better offer, so I will not use those 2 cents as the cutting line, but I will increase it to 3 cents. I also know that some sources might perform way better than others, so let's increase it to 4 cents per click max.

The first thing I will do is run a BOT test and I cut everything that has 80% or more BOTS – only cut the WORST, this is the key, so based on the traffic profile you can increase the limit to 90% even. This is something you need to judge on a case by case basis – check what the numbers look like and get rid of the very worst.

The next thing is to look at the banner CTR and remove every banner that is giving me clicks more expensive than those 4 cents I calculated.

NOTE: this phase is not about finding a winning lander or offer, it's just about identifying BOTs and banners with lousy CTR. So don't worry about slower loading times due to the bot test script. You'll look for winners later on.

In the phase 1, should I also cut the placement with lousy CTR? If it is true, after the placement spend how much we can make the correct decision to cut it?

11-14-2019 12:18 PM #45 matuloo (Legendary Moderator)

Originally Posted by j80montes

In the phase 1, should I also cut the placement with lousy CTR? If it is true, after the placement spend how much we can make the correct decision to cut it?

When following this method, always compare the performance of specific placements to the rest. In case you see that the average CTR seems to be, I don't know, 0.1% and there are placements that have CTR of 0.01% then yes, you can cut these.

These very poor ones will be easy to spot very soon, it only takes a couple 1000s impressions really. If you want a ballpark figure, let's give them 3-5000 impressions for example, if nobody clicks then cut.

11-19-2019 03:11 PM #46 j80montes (AMC Alumnus)

Originally Posted by matuloo

When I start a new campaign in a new market, GEO or whatever, I like to have at least 10 banners, so I can test faster and get some results asap. When you start with just 2 or 3 banners, chances are you hit all bad ones and the results won't help you. SO starting with more banners is better. Then again, if you put too many in rotation, the test budget needs to be higher. So to make a conservative bet here, I'd say you should start with 5 or so.

Hi Matej,

When we are testing banners, lets say 15 banners need to be test, start with 10 banners, keep 5 banners after cutting, should we wait for the final winner and then test the rest 5 banners, or just add 5 more banners directly.

11-20-2019 09:22 AM #47 matuloo (Legendary Moderator)

Originally Posted by j80montes

You can add new ones as you remove the bad ones, then compare the performance of the new ones with the data you have for those that made it to round two. So the best ones from round one, become something like a benchmark that you want to beat with the new banners.

A small tip, sometimes it's better to wait some time (a day maybe) before adding new banners into the mix, in order to give the better ones full traffic rotation and test how that influences their performance, chances are some of them will perform worse.

11-20-2019 12:21 PM #48 Admolly (Member)

Thank you very much!! everything is super helpful

11-21-2019 09:26 AM #49 matuloo (Legendary Moderator)

Originally Posted by adthority

Thank you very much!! everything is super helpful

You're welcome

11-25-2019 02:43 PM #50 j80montes (AMC Alumnus)

Originally Posted by matuloo

In a case where the profitable banner totally slayed the other ones, you should keep just that one and make variations of it. There are 3 possible scenarios here, the image was the winner, the copy was the winner or both the copy and image simply clicked well together. To find out what the case was, you should make let's say 2-3 variations with the same copy but different images, 2-3 more with the same image but different copy and rotate them evenly along with the original winning banner. Run this for a while and you should get an idea about why the banner worked well, then you can replicate this further.

Hi Matej,
I finally find the banner that slay all the others, also the result shows the angle plays here.
What should I do next? Keeping create more variations with only this angle, and always cut the worst forever? Or should I settle down for a while?
By the way, I read guys always say like this, when the campaign get profit, just keep running and refresh the banners sometimes, what "refresh the banners" suppose to be here?

11-26-2019 08:46 AM #51 matuloo (Legendary Moderator)

Originally Posted by j80montes

Thats great!

So now, I would try to make variations of the banner, keep the angle, use different images. The goal is to have several banners that work well and rotate them in the campaign, to slow down the burnout process and keep some variety in your campaigns. As long as they perform decently, do not cut the lower performing ones, try to keep the variety. Once some start to fall behind significantly, you will have to pause them again.

You should also try to scale this "winners" to different spots or networks.

By the way, I read guys always say like this, when the campaign get profit, just keep running and refresh the banners sometimes, what "refresh the banners" suppose to be here?

Refreshing banners means to add new banners, basically. If the angle is strong, just refresh/change the images. And it's also a good idea to dedicate some part of the profits to test entirely new banners, so you have something to run, when the winners start losing steam.

11-26-2019 08:59 AM #52 j80montes (AMC Alumnus)

Originally Posted by matuloo

So now, I would try to make variations of the banner, keep the angle, use different images. The goal is to have several banners that work well and rotate them in the campaign, to slow down the burnout process and keep some variety in your campaigns. As long as they perform decently, do not cut the lower performing ones, try to keep the variety. Once some start to fall behind significantly, you will have to pause them again.

That's what I am doing now, I 3 solid variations now, is it enough to keep some variety for my campaign, should I try to figure out 5 more, or even 10 more banners?

11-26-2019 09:04 AM #53 matuloo (Legendary Moderator)

Originally Posted by j80montes

That's what I am doing now, I 3 solid variations now, is it enough to keep some variety for my campaign, should I try to figure out 5 more, or even 10 more banners?

3 should be enough, but if you can do more, that worn hurt. When you have enough solid ads, you can occupy several spots on the same site or even run two campaigns in the exact same spot, as it looks basically as a competitor... it's also easier to scale to other sources when you have more solid ads at your disposal.

But for a single campaign, 3 solid ads is more than enough

11-26-2019 11:48 AM #54 eduaffiliates (Member)

Helpful post. Thanks!

11-26-2019 03:59 PM #55 j80montes (AMC Alumnus)

Originally Posted by matuloo

Thanks a lot with your help.
But the offers has been capped due to I sent over a few hundreds within 1 day, and the AM told me the advertiser may be deduct those leads over 100, hopes they don't do this.
Thanks again, I have full picture to optimize banners now.

11-26-2019 09:43 PM #56 matuloo (Legendary Moderator)

Originally Posted by j80montes

Yup, offers are usually capped initially, always ask your AMs about possible caps. They won't deduct the leads in case the quality is ok, at least usually they dont, in the end they all want as much leads as they can get. Holding fingers crossed for you.

01-06-2021 11:57 AM #57 albavnw (Member)

Hi @matuloo, I have a question regarding the "worst": Let's say I target "Browsers". There are multiple browsers like chrome, safari, firefox, edge... The "edge" is the worst so I will cut the "edge" targeting. After I cut this "worst", should I continue to run the campaign for the remaining 3 browsers chrome, safari, firefox and hunt for another "worst" among these 3 browsers and cut it, and then continue the campaign and keep hunting for the "worst" among the remaining browser targets. Or should I just cut the first "worst" (in this example it's "edge" browser), then move to find other "worst" targets.

For example in this case https://i.gyazo.com/2d741c0e98b81205...044bcfaeaa.png, if I cut the worst which is Android 2.2 I don't think it will make any impact on the overall campaign performance.

01-06-2021 05:54 PM #58 matuloo (Legendary Moderator)

Originally Posted by albavnw

In theory, you can continue for as long as you need to reach profitability. But, some targeting options cannot be optimized infinitely. For example OS... you cut android, then iOS, then windows and you have no traffic left

Same with browsers, you can cut edge, then chrome and bigger half of your traffic is gone.

Speaking about browsers specifically, and OS too, there are offers that simply won't work with android, but perform well with iOS... same with certain browser versions. In such cases, it makes sense to cut the non-performing ones, even when it means a large drop in volume. But at some point, it makes no sense to look for the worst variation anymore, as the volume would fell too low.

With placement optimization, you certainly have more "rounds" available, as new placements get added all the time, so you can definitely look for "worst" ones again and again.

But even here, simply cutting the "worst" won't magically turn a campaign green. If after several rounds of such optimizations, you still cannot reach profitability, it's time to look at the rest of your funnel as the problem might be somewhere else... offer, LP, ADs etc.

if I cut the worst which is Android 2.2 I don't think it will make any impact on the overall campaign performance.

Yup, this particular OS version is getting a minimal amount of visits, compared to the bigger ones and by cutting it, there wont be any impact on the campaign. You can pretty much ignore such low traffic targets completely and don't waste time by optimizing them.

05-27-2022 04:27 PM #59 bluecape (Junior Member)

Hey matuloo,

How long should you run the bot test for? I'm currently running one (using Caurmen's guide) and I realized that it's an infinite spawn game. New placements/zones keep popping up.

Home > General > Affiliate Marketing Forum