Home > Native >

---

Making Native Ads Work in 2017 - Man vs Machine (20)

---

Hey guys, hope all is good .

It's clear that automation is becoming more and more important to affiliates these days.
A select few that have been taking advantage of it for years and getting massively rewarded.
For the rest, it's a fight against time. It's obvious, if you are not using it, you are starting the race with a huge handicap.

We can clearly see the trend here. In a few years, most of the optimization process done by affiliates will be totally replaced by machines and artificial intelligence.


So I decided to start this case study about natives where I'll go through the whole process from scratch. From finding the offers, landers, creatives, picking up 2-3 different traffic sources and a couple of geos to split test, down to the optimization.


The goal here is to compare campaign performance and time spent when doing things the old fashion way (aka manual) and using platforms like Theoptimzer.io - Native

For this purpose I'm going to create a few campaigns on different traffic sources that I'm going to optimize manually and then the same campaigns will be totally automated and wont be touched after the creation.

I will report here daily the evolving of the campaigns, the approx amount of time I spent manually optimizing the campaigns, what were the decisions I made manually and those that TheOptimizer made.

This is the basic idea.

Tomorrow I'll go through all the details on how I pick the offers, creatives, landers, geos and the traffic sources to test.

Meanwhile, if you have any suggestions, feel free to comment below.

Peace!

---

This is awesome, following closely!

---

Fantastic idea! Can't wait to see the data

---

Hey guys!

So today is about deciding what is the vertical (offers, creatives, landing pages), geos and traffic sources that best fits the requirements of this case study, and uploading the campaigns.

Given the the goal here is to compare manual vs automatic optimization, I'm looking for low payout (so I can hopefully have some quick data to work on) , and also creatives/landing pages that can get easily approved on different networks.

So what better place than AdPlexity to create a quick idea about what can do for my case?

Fired up AdPlexity and just after a few scrolls I immediately noticed some games that seems to be very popular atm.

Bingo, that's perfect!

After opening 2-3 ads noticed that they all redirect to plarium.com, so decided to do an advertiser specific search and see which geos and sources are more popular. The screenshot above ilustrates all the process:



Offer: found the plarium offers on different networks after a short search. Decided to go with Vikings.
Payout for US $5.5, for TR $1.

Geos: there were no surprises, US is on top as always.
Had a little hesitation whether to include US in the geos to test (because of the expensive traffic and the high competition) but then said what the hell, it will be more interesting .
So I opted for US and a cheaper country, TR.

Angles: I noticed that basically there were 2 different ones: "if you own a computer you should try this game" and the one that goes more in the direction of a "sexy game". Going to use 10 creatives for each campaign, 5 for each different angle.
Not sure if the "sexy" angle will get approved everywhere but at least will have a fallback with the normal angle.

Landers: When it comes to landing pages used, I noticed that mostly these games are ran direct link (their own landing page looks pretty good).
There are also some affiliate landers around but I'm pretty sure will not be possible to get some of them approved on all the traffic soruces.
Anyway trying doesn't cost anything, so I'll give'em a go .

Direct Link:


Sexy LP:


Adicted/Action LP:




Traffic Sources: as you can see from the screenshot above, the game is present on Outbrain, Mgid, Taboola, Revcontent.
But given that we do not support Outbrain yet on TheOptimizer, will start with the other 3.


So, the final picture will be:


Taboola:

- 2 US campaigns (1 manual & 1 automatic),
- 2 TR campaigns (1 manual & 1 automatic)
- daily budget $30 to start with

Revcontent:

- 2 US campaigns (1 manual & 1 automatic)
- 2 TR campaigns (1 manual & 1 automatic)
- daily budget $100 (the minimum for revcontent).
Note: Will try to not spend all of it (by pausing campaigns manually after a certain spent or using the rules in case of TheOptimizer ) as it is a high testing budget for these payouts.

Mgid:

- 2 US campaigns (1 manual & 1 automatic),
- 2 TR campaigns (1 manual & 1 automatic)
- daily budget $30 to start with


Uploading the campaigns now.

Once approved, the next step will be to define a set of rules that we're going to use during the manual and automatic optimization process.

As always, keep the suggestions coming

Cheers

---

Im liking this, following along as well.

---

Very interesting follow-along, big fan of automation so will definitely be keeping tabs on this!

---

Hey guys, whatsup?

Still waiting for campaigns to get approved.

Meanwhile just wanted to update you quickly about how the final Flow looks.

I made some small changes to the original idea and decided to also split-test Direct Link.

So to achieve this I created a flow on Voluum where I'm going to test 3 paths:






Here are also the Ads for each angle:


Sexy angle








Adicted Angle







Direct Link






Honestly I'm not very hopeful of seeing the "Sexy" ads approved, but anyway lets see what happens .

---

Hey guys, sup!

Finally all the campaigns got reviewed and we are ready to go.

Here is the final approval status for each source:

Now lets set quickly go through the strategy we will be follow during this case study, means bids, budgets, and optimization process:


Budgets:

We will start with a daily budget of $30 for each campaign (it is suitable for the testing period, given the payouts we have, $5 US and $1 TR).
Will only increase the daily budget for the campaigns that are showing some potential, means they are either profitable or a small negative ROI.


Bids:

Regarding the bids, will start with an average bid for each source that will possibly allow us to spend the entire daily budget.
Will not touch the bids for at least 3 days unless one of the campaigns is not able to reach the daily budget.


Optimization Process:

As I mentioned on the opening post, we will be using TheOptimizer.io for the automatic optimization process.

I have already setup our Voluum tracker and all the traffic sources so the environment is ready to go.

For the optimization process I have created 3 kind of rules that will form the basics of the optimization:

1. Bot widgets blocking rule (bot widget = very high LP Ctr or very low LP Ctr):
   
   The most standard way to identify "bot widgets" is to look at the LP Ctr they generate. Usually they either have 0 or very small LP Ctr (<5%), or really high LP Ctr (>80%). So we have 2 rules for this:
   
   
   - Low LP Ctr rule: IF clicks>50 AND LP Ctr < 5% AND Revenue =0 ==> block this widget
   - High LP Ctr rule: IF clicks>50 AND LP Ctr > 80% AND Revenue =0 ==> block this widget
   
   
   
2. Bad performing widgets blocking rule
   
   A bad performer is a widget that has a normal LP Ctr, and either has no conversions or it is converting but losing money anyway.
   To deal with these kind of widgets we need to take in consideration the payout, and give them the right amount of testing budget before blocking. So in this case will have different rules for US and for TR given that the payouts are different.
   
   - US not converting (spent $10 and no conversions): IF cost>$10 AND Revenue=0 ===>> block widget
   
   - US bad performing (has conversions but losing more than $10): IF NET<-$10 AND Revenue>0 ===>> block widget
   
   - TR not converting (spent $5 and no conversions): IF cost>$5 AND Revenue=0 ===>> block widget
   
   - US bad performing (has conversions but losing more than $5): IF NET<-$5 AND Revenue>0 ===>> block widget
   
   
   
3. Bad performing campaigns pausing rules
   
   We need these rules to control the spending at campaign level so that we don't have campaigns that burn all the budget with no conversions at all.
   This rule will pause campaigns once they reach a certain amount of loss.
   
   - Stop Campaign losing $20: IF NET<-$20 ==>> pause campaign

All the above rules are set to run every 15 minutes.

Here is how they look on TheOptimizer:






For the manual optimization we're going to use the same strategy, with the only difference that we are going to do it manually few times a day.

I will try to also keep track of the time spent during the manual optimizatin.


Stay tuned.

Peace

---

Interesting thread. Following just to see how the automation bot does vs manual

---

ahh damn cant utilize a custom tracker? seems like it need to connect to one of the 3 listed only?

---

Interesting topic. Keep posting! We are so excited to see the result. I am a newbie and do things manually. Your sharing is eye-opening for me. Love learn more.

Keep up great work!

---

very interesting

---

Hey.

This is a really nice experiment.

Got any updates?

---

Hey guys,

sorry for the slow update here, and thanks for the interest on this.

Unfortunately had a bad timing and ran out of budget on Taboola and Mgid , so had to wait for the funds to be available so I can continue in parallel with all the sources.

So we finally got the $$$ added yesterday evening and started getting some traffic.

The only issue is that the initial bids we started with ( Mgid US $0.15, Mgid TR $0.07, Taboola US $0.17, Taboola TR $0.09) were too low to spend the $30 daily budget (campaigns spent $5-$6 so there was nothing to optimize), so I decided to increase:

Mgid US -> $0.18
Mgid TR -> 0.09
Taboola US -> $0.21
Taboola TR -> $0.11

Hope to get some data to analyze at the end of the day with the new bids.

Will be back with updates soon

---

Ok, finally got some numbers to work on.

As you will notice from the screenshots below, same campaigns on the same traffic source, with same bids and same ads are not getting the same amount of traffic.

Anyway this is a known behavior for those who have a little experience with natives. For some reasons, when their internal algorithms see the same ads in different campaigns they tend to give priority to just one of the ads (I think randomly, or the first that got the first clicks).


Here is the spending overview for each source for first 2 days :

Screenshot from theoptimizer.io (Native)


So the overall situation on Revcontent is not very good.
The main reason for this was one of the manual campaigns, as you can see from the screenshot above, that went out of control and spent already over $100 by the time I was able to go and block the widgets manually.



Screenshot from theoptimizer.io (Native)


The other campaigns were paused automatically by the rules once the Net was < -$20, so we did not lose any money.



* time spent to optimize manually: 15mins

* money spent over the assigned daily budget: $100






Screenshot from theoptimizer.io (Native)


Not much to say here, still the spent is not picking up as we need it, so will try to increase CPC again. Now we are paying 0.19 will try to increase to 0.21 and see if we can get those $30 daily spent.

One thing to mention is that on the auto campaign that spent $26 the rules did their job and blocked a site that spent $7 with no conversions


* Time spent to optimize manually: 15mins (needed to go over the sites and contents and check their spent/revenue anyway)








Screenshot from theoptimizer.io (Native)



A bit more activity on Mgid as you can see. All the campaigns managed to spend and have some conversions. Mainly I think because we got the "sexy angle" approved, so the CTR is higher.

Anyway not much to optimizer yet as the cost was distributed among many widgets.




Screenshot from theoptimizer.io (Native)


The rules did not block anything for the auto campaigns, just had to block a couple of widgets manually for the manual campaigns.


* Time spent to optimize manually: 15mins (needed to go over the sites and contents and check their spent/revenue anyway)




So, my overall opinion is that chances to get something profitable are real only on Mgid and Taboola. Revcontent seems far, but anyway will give it a couple of more days.

So far automatic optimization has not helped on making things profitable, but for sure has helped keep the loss under control, which is really important during this first period.

Next step of optimization will be to start cutting off bad performing contents. So Im going to create also some rules for that.


Will see how it goes, stay tuned

---

Awesome post man, following along closely!

---

This is a really important point on automation: it's great for minimising your spend, which gives you more rolls of the dice.

The difference between pausing an ad as soon as it hits statistical significance and pausing it six hours later when you wake up can add up to real money pretty fast.

---

Hey everyone, how's it going?

So, here I am again with some final numbers about this follow along.

It came out that the offer I picked for this follow along was not as "easy" as it looked at first.

Went through a few issues during these days that, along with the fact that the offer did not perform very well without a pre-lander, made the follow along more difficult.
It first had some technical problems for a couple of days so we had to pause, then they told us that the quality was not good, so we were capped at 50 conversions daily plus saw the performance go down.

So at the end decided to end this follow along here and post what we got so far.


But of course it doesn't mean it will stop us from comming with new more interesting stuff .


So far the optimization consisted only in blocking the bad performing placements, but we are missing another important part of optimization:

"Scaling the good performing campaigns/placements".


To achieve this TheOptimizer is getting equipped with some new powerful tools:

- Automatic bidding: create rules that will change the bid of a campaign/placement based on some performance conditions:

example: "If ROI for last 3 days > 30% raise CPC by 20%"

This way as long as the ROI will stay stable we keep increasing the bid. Final result of course will be higher profit


- Automatic budget changing: same as bidding, the user will have the possibility to change daily budgets based on performance

- Change weights of landing pages and offers: we have been working for months now on a proprietary algorithm that will suggest the right weights for your landing pages and offers.
This is not based simply on the ROI or EPC (what would a normal user do) but takes into consideration a lot more variables (LP CTR, CVR, CPC, payouts, daily budgets, etc).


With these 3 tools in hand:

1. Blocking bad performing placements
2. Scaling good performing campaigns/widgets
3. Finding the best performing Landing Page/Offer combination

the optimization process can be said to be 90% complete (user still need to work on good CTR creatives, and good LP CTR).


Now back to this follow along.


It was clear from the begginging that from all the traffic sources only Mgid had some potential to get something profitable, and that happend.

As you can see from the screenshots below, one of the automatic campaigns in Mgid was profitable without ever touching it. And I think this alone is amazing

The fact that this happend on Mgid is simply because it was the only source that allowed us to run the landing pages.

But TheOptimizer did its job on the other sources as well (beside the fact that they were far from getting profitable), it kept the loss into reasonable values, while the manual campaigns got out of control a few times.



Here is what happened for each source:

To be noted that the loss was higher on almost every manual campaign compared to the ones that had rules associated (automatic campaigns).


MGID:









Revcontent:









Taboola:





Here is a recap of the rules we setup on TheOptimizer for the optimization:

Would love to hear any kind of feedback


Thanks for following and stay tuned for the next follow alongs.

---

How did you come by your rules to blocking certain widgets?
Do you keep logs of what is blocked day to day?
The reason I ask the last question is what if your landing page goes down which means there are clicks but no click through which may end up banning all your widgets.

In automatic budget changing, wouldn't you be leaving money on the table if it converts well at that bid and you keep upping your cost? At some point, you'll be making minimum profit. You'd rather want an optimum bid that keeps your conversions rather than upping the bid every 3 days.

I definitely see how good a tool this would be. Caurmen has it right. It gives you more roll of the dice. The campaigns still need that TLC

---

Hi sharkweaver,
thanks for your good points.

This mainly comes from our experience with natives . But of course depends a lot on your personal optimization habits

This is a feature that is ready. We are doing the last tests locally and will release it in the next 2-3 days.
It will log every single action that happens on the platform, manual or automatic.

This can happen if your are optimizing based only on today data, which is not the best possible practice.
But if you consider the stats for a longer interval, for example last 7 days, even if the LP goes down for a few hours or for a day it will not happen, because you will have LP clicks for the other 6 days.

Not sure I totally understood your point here. But anyway upping the budget doesn't mean that the profit goes down.
The idea here is to have a rule that, as long as you are having a stable ROI with the existing bid, it keeps increasing the daily budget of the campaign, so it can spend more.

Spending more with the same ROI means higher NET profit.

This simply avoids situations when you have campaigns with very good ROI, but they stop getting traffic because they reached their daily budget.

---

Home > Native >