Home > Paid Traffic Sources > Mobile

---

Not succeeding on mobile? Maybe it's not (totally) your fault (30)

---

I just wanted to share my last experiences with mobile advertising.

I was tired to see hundreds of visits with absolutely no click through the offers on my landing pages.

So I've coded a PHP script in order to check out where the traffic really came from.
At the beginning it was designed to be simple, but it ended up being more sophisticated: I wanted to get some info like referrer, IP address, country, user agent and things like that.
It's a bit more interesting that the bot detection script you can find on STM, because it gives you more data to help understand your traffic.

So I launched 2 campaigns this month. One on Tapit, the other one on Decisive, and the results were sick!

Tapit:
Around 50% of the traffic = bot traffic
(or maybe some sort of scraper)
Testing volume: ~500 clicks, ~65 USD

I'm not exagerating, some of the placements there were litteraly bringing 100% of traffic from the same IP. (hundreds of visits)
It wasn't even from different IP into the same range of IP, it wasn't 2 or 3 different IP... No, the traffic was literally coming from 1 IP, with no interaction on the landing page.
Also, this traffic came from countries I wasn't targeting.

I mean, I can hear that when you don't spend billions, with 3 figures CPM, you can't access the premium spots, but come on... Some spots are nothing but visitors with 1, and only 1 IP. And Tapit charge you for this traffic.

The problem is, if you're just using Prosper202 as a tracking tool, it will filter out the majority of these clicks.
(The discussion about click discrepancy sounds familiar...?)
Even when you want to show 'all clicks' on Prosper tables, you can't see this traffic, for some reason, it doesn't appear in your tracker.
But in reality, you're charged for this traffic and it really screws your stats, your budget, and your maketing efforts.

Decisive:
Pretty much the same problem here.
Around 50% of the traffic = redundant IPs, or traffic from untargeted countries, that come dozens of times on your pages
Testing volume: ~4000 clicks, ~200 USD


Well, I don't like to reject my failure on others, and I'm not trying to launch a debate or defaming, or things like that... but if I could see this problem with a php script that took me a few hours of coding, as professionals, they should be aware of this problem and could prevent their customers to be charged for this traffic.

I'm not trying to denigrate these 2 traffic sources, I know a lot of people are doing very well there, but maybe my experience could help reassure a little those who can't make anything work on these places, like me.

Please note that I'm not hating, or doing bad publicity because I suck at affiliate marketing, and it's true that I'm not a seasoned affiliate marketer. But unless there is a valid reason to get hundreds of visits from untargeted countries with one IP only, I think there is something wrong here.
Tapit and Decisive are still very great interface with awesome user experience, but it's not the first time this problem shows up on forums.

The lesson here, and from now, it's a full part of my process: Never trust the stat from your traffic source and from a tracker you haven't coded yourself... Always do a complementary check up on your end!

---

Had pretty much the same experience with all DSPs (Decisive, tapit). That's why I decided to work with Redirect traffic & popads.
I had an LP with 500 visitors and 11 clicks 2.2% CTR while the same LP had a 30% CTR on Redirect traffic sources.
Leadbolt had a good CTR as well since they have their own inventory.

---

Yup.. this is about the size of it.

People will argue that the same IPs are blocks used by certain mobile ISPs.. but smells of bullshit to me.

Basically there's next to no click fraud detection in place on mobile apps. Well at least compared to actual websites on the main exchanges.

---

The worst culprit is Airpush..

---

@Julien

We added in variable frequency capping so you can control number of visits per day (this is under audience) for each ad on Decisive which should be helpful for you.

The ip blocks do move around so we do use the most updated version of Maxmind.

Hopefully these two components help make our mobile advertisers more successful. Please do let us know if you see other tools that would help you be more successful on mobile.

---

How does it work? Do you drop a cookie on each mobile user? How do you check returning visits?

---

Hey Andrea, my ads were having a frequency capping set to 3 per 24h.
It didn't prevent the redundant visits

---

I think another problem is if you're just using P202 for mobile campaigns, period.

Not having the ability to perform post-click redirects, as well as the ability to handle lots of volume well without slowing down (among other things) is a serious problem, especially in mobile.

Most traffic sources (if not all) will suffer from bot traffic. No way around it. Just gotta find a way to get rid of as much of it as you can (blocking the worst pubs) and get into the green. It's 100% possible.

---

Best advice you can get. It is like it is... and you are right, lots of bot traffic around!
Why did you code a php script if trackers like Voluum have such features built in?

---

Wanted to correct a few of your comments.

The free P202 1.8.x has been able to do post-click redirects for a little over a year now.

and

"The ability to handle lots of volume well without slowing down", is a thing of the past in P202Pro

http://prosper.tracking202.com/blog/...000000x-faster

---

They must be bad placements. Just keep filtering. People have been dealing with this on SireScout for years, for example. Everyone deals with it so consider it a cost of doing business, it keeps out a number of competitors who don't want to deal with it.

What I don't get though is how you would be charged for clicks from one IP beyond your set frequency cap. Can you confirm the clicks were all charged?

Does anybody know if it's possible to get a refund for such traffic from a DSP like Decisive? It's probably easier from the individual exchange that owns the relationship with the site or app. I mean, it's part of the game, but if the frequency cap is violated, I'd be pissed.

On the other hand, not annoying your traffic contact with refund requests is probably a good idea if it's in the lower range $.

One cool thing to remember is, if you are running clean, good placements (no bot traffic) are made available (or stay available). So just keep hustling bro, you'll get there!

---

My apologies, I stand corrected.

It's been well over a year since I've tested P202, and during that time there were no post-click redirects. Looks like you guys have added quite a few new features to P202 since then.

I've been extensively testing Thrive this past week for performance, and now that you mention this I'm a little curious to put them up against each other to see how well they compare in that regard.

---

Zeno had a great post about this - http://stmforum.com/forum/showthread...Your-Campaigns

---

No worries man

Yeah, frankly p202 1.72 was getting outdated (It was over 3 years old, before all the new work on 1.8.x from last year, unfortunately there are still a ton of on 1.7.2 today)

Don't want to hijack this thread too much. So will just end with, we are under very active dev work, fully self-hosted and still 100% open source.

OP: To get back on track a bit, in p202 a filtered click still gets sent to the offer etc. I'm interested in see if you were doing direct linking or using a lp? I know because of the use of javascript on the lp, if the bot doesn't eval js it will never get logged in p202.

---

@Dario We use device ids (all different kinds), user ids, useragents, ips, and placements to determine the user. And then check against those for each incoming bid request.

---

Unfortunately, I must say that I'm seeing the same results on Decisive. I followed Caurmen's guide ang got an astonishing 50% of what appears to be Cylons, R2D2s and Roombas traffic

---

one way to look at it is this:

Once the ad networks figure out how to kill this bot traffic and if you have a profitable campaign. Then most likely you will have 3x 5x 10x profits

---

Bot traffic is part of the game. I don't care what traffic source you use your always going to get some bot traffic. Some sources do a better job of filtering it out then others the best you can do is optimize and see what's converting for you and what's not. In my honest opinion I don't think there will ever be a way to completely filter out bot traffic. According to a specialist on Bloomberg its estimated that advertisers will spend 6.5 billion on bot traffic in 2015.

---

It is maybe "part of the game", but if we are being completely honest, many traffic sources have the ability to do a much better job blocking fake and bot traffic than they do now. It really is instead much more a question of will.

---

Tapit, Decisive and most all other DSPs get their traffic from the same places, so it isnt the DSP, it is the source.

I would say if you dont succeed in mobile it IS totally your fault. Why? Because all of us deal with the same placements and DSPs, yet some of us are highly successful on DSPs.

You have two choices:

#1 - Build scripts to prove to yourself that you arent a loser and it is the DSPs fault and fail.

#2 - Test the placement and learn how to blacklist it so that you dont spend any more money there in the future and succeed. This is also known as "testing."

With option two, you dont need scripts. Once you have got enough data from a placement (and it doesnt take a lot) to know it doesnt convert or maybe sending bot traffic, you simply dont run traffic there anymore.

---

The ideal solution would be DSPs opting out of auctions where the user is suspected to be a bot, without actually opting out entirely from that placement.

The exchanges/SSPs can do a better job of removing and excluding bots. DSPs can probably do a better job of excluding bots, though this may be difficult depending on the back-end mechanics.

In the end, adapt or die.

If bots were removed we would all do a lot better but competition would rapidly increase as people were able to spend more on impressions > bids all go up.

You could argue that, if you assume all bots provide zero value on the back-end, then if you removed all bots your revenues would be unchanged but costs would go down and the traffic source costs would then normalise back to give a similar cost/revenue status.

The main thing that would change is that you would no longer pay for tracking costs associated with those bot clicks.

But would removing bots revolutionise the industry? I don't think so.

---

cmdeal

I completely agree with you a lot of these sources can do a WAY better job then they are currently doing now. I can personally do a better job then some of these sources. Do I think any of them can put a complete end to it. No. We will always have to deal with bot traffic unless a new game changing technology comes out.

---

If you spend some time on Android Dv. forums you will be amazed what lengths some guys go to to inflate their impressions, force geo and cheat ad networks. Its pretty much a losing battle for ad networks - many doing a lot to try and prevent it. On DSP's there are much more room to cheat for these guys

---

I had the same experience on Decisive spent couple of hundred to get 1 fing lead using the mobile cookbook, allways being stuck at the "cut bad placements which dont convert" part as i was not getting any conv. After which i was like a dog chasing its own tail, making new ads with different angles etc hoping to get at least one conv, rince and repeat... annnnd finally i had that one magic conv but too little too late I already moved on to greaner pastures (traffic sources)
Tryed another traffic source with the same method and after only 10$ i had 3 leads.
Still have aprox 100$ on decisive for a couple of weeks untouched as i cant convince myself to "donate it"
I really doubt these traffic sellers care at all if the traffic is bot or not as they still get their money, noobs beeing suckered in with "you need atleast 1k$ for testing pourpuses" which is bs as i dont need to spend 1k to see the traffic is coming from bots. The Mobile Cookbook should contain "sign in to 5 traffic sources, spend 100$ on each and use the one which has the least ammount of bot traffic"
There should be on all IM, AM forums a subforum for calling traffic sellers out with their shady practices maybe and just maybe sellers will think twice . They are selling bot traffic because they can get away with it and you are funding them.
Ever wondered why high profile users guides/follow alongs are so succesful yet most people in the follow along fail? Thay are probably put in a high profile traffic pool which has no bot traffic, making a follow along with their traffic source is a goldmine cuz noobs will be all over it buying lots and lots of bot traffic.
I would even go as far as dear the high profile users to create a new account on any of these traffic sources , deposit 1-200$ and see if thay can replicate the the same results as thay have on their high profile account on the same traffic seller. Or is the "big secret" the "golden gem" which traffic sellers do actually sell traffic?
Why is that most people subscribe to stm for 1 month and never come back? If you are thinking "thay gubble up as much content as thay can and leave" which is a false assumption.
Having an account at stm isnt 100$ its either by the hundreds or thousands, people who come here have a substantial budget to invest, following guides/tips and after 30 days most of them are in the hole and give up entirely. Give some value to our subscription , protect us from shit like this, call them out and stop endorsing them, help us to stay active and pay our subscription teach us where not to spend! Make guides/follow alongs tru a noob or tru a new account and not with ur avg 50k mo spending account which has traffic sources eating from ur hand where u could literaly make money from shit.
Sorry for the long rant, this is not just about Decisive but most traffic sources.
Just my 2 cents , take it with a pinch of salt.

---

RTB Fraud is not something that can be solved easily, people really need a better understanding on how the whole RTB ecosystem works.

At the bare minimum DSP's receive bid requests from exchanges, those bid requests contain relatively limited data, IP, Useragent, that data alone makes it hard to pinpoint fraud.
Bid requests need to be answered within 100ms most of the time, from whereon you either won or lost that impression, if the DSP bids on it for you (if you're running a campaign).
A bid request isnt made by the user who initiated the impression, but by the SSP/Exchange -> DSP, so the DSP pretty much needs to blindly bid based on the limited data available.

Another problem is that while the DSP's client will ask for a refund for fraud traffic, perhaps the DSP will actually honor such a request, its not likely that the DSP will be able to
get a refund from the exchange.

There are a couple things that can be done to avoid as much fraud as possible, but none actually will be a definite solution, for example
A DSP could keep track of historical performance of any given ip over all their clients
A DSP could keep track of complaints actively fed by the DSP's users, and ignore bidrequests containing similar data from there on
A DSP could integrate with 3rd party data providers, as much as possible, trying to get more data from the specific bid request

However each one of them, especially the last one will substantially raise the CPM. The first 2 will be reactive solutions instead of proactive, the third option perhaps a more proactive solution but will most likely at least double your cpm if not much more in most cases.

The traffic quality thats often obtained by affiliates/performance marketers is a lot of the time less due to the nature of what we want to advertise. There's performance in play so conversions/performance are crucial. We do anything to get the most possible from the impressions we buy, which in turn creates quality issues for most of the more "premium" placements. Take a look at for example SiteScout and have a look at the top placements site rules, it gives you an impression of what they allow or not, a lot of which is being run wont ever qualify to be placed on such "premium" placements.
Also, in general the better the exchange filters out shit traffic, or the better quality publishers they have, the more strict they are (think Google, FB, Rubicon etc)

Of course there is still plenty of inventory available, but then think about how it ended up being available for you. Does the publisher not care at all what kind of ads are being shown on their properties? Are they new? Do they just want to fraud their way making some money, generating bogus impressions? What I'm trying to say is that we will always deal with the fact that where we buy traffic, for the models we are used to, we will only be able to buy on placements that have a similar mindset as us. If you have a website or an app thats making you money, would you want to risk your userbase by flooding it with bullshit AV banners and landers, combined with a pin submit pretty much ripping off users?

Always ask how your ads are classified, every submitted ad will be classified before approval, and landing page urls submitted to the exchanges for review as well, or at least used to classify the campaign better.
If the DSP's wrongly classify the ads running through their platform, there are consequences for them, so don't expect them to classify your ads "better" or different.

So what can you do?
You can buy RON, and use performance data, not care about fraud since you will filter it out for not performing anyway.
but a DSP such as decisive would have maybe at least like 2000 apps and another say 1000 websites available in their inventory.
Say you have an offer with a payout of $3, following the general rule cutting a placement if it hasn't converted at 2x the ad spend ($6), simple math tells us you're now down 3000x6 = $18.000 for a complete test.
Personally I wouldn't want to START off buying RON from any network, especially starting out, which in my opinion is why many fail or get demotivated by their first attempts.

You need to think about the problems you're facing, in this case is RTB fraud, but it's not just that, here's some more:
- Offer performance
- Landing page performance
- Targeting

I'd want to solve the following question
- How can I decrease my exposure to RTB fraud, and get more reliable data on the landers/offers im running and how can I try to get better quality traffic?

Maybe segmenting your buys is a better way?
- Find out what your target demo is of the offer/angle you're running (or a similar offer) -> either caurmen or zeno made some interesting posts that might help you here, dig through their posts
- Find out what placements match that demo on the available inventory as best as you can
- Split up your RON buys in smaller chunks, so your money being wasted at 5000 single impressions spread across 5000 placements isnt an issue anymore
- Inform what your ad is classified as, investigate the consequences, ask how you can improve that (if its possible at all)

*edit: These are just my thoughts by the way, if an ad network representative is here to clarify more, please do.
After all, if you're eager to do business, and you're dealing with a bunch of RTB fraud complaints as we all experience, why not help educate people here on how to deal with it, the right way.

---

@karimelm You did a more than superb job explaining the ecosystem and have an excellent handle on the controllable factors as an advertiser.

---

I felt the same way for a while, then realized that there's something very good about that, even for anyone wanting to start out. Imagine if all the traffic was sweet and everything converted, then every lazy ass could just rip out a landing page bid higher and make profit. Sounds good? Not really...the costs would rise soon and it would be very hard to profit (you'll need better offers and direct deals to differentiate, or come with other angles that would be ripped very soon).

I think of the bot traffic as an advantage for people who really want to do this. I focus on placements and profit, while others focus on detecting bad traffic and trying to switch from a place to another until they run out of money or quit, so they don't stick too long to increase the prices for me.

I'm not saying to ignore it (you should always understand what happens in the system) and it's a big problem that should be solved with some better solution, but until the system evolves you have to play the game and use everything to your advantage.

Learn to live and love the bots lol.

---

@Phlong can you explain why redirect traffic would avoid this issue? Are there no redirect traffic bots?

---

Because my data is king, and I have worked long enough in internet marketing agencies to be aware that in the best case scenario, employees are just "curious" of what their customers are doing... In worst case scenario, the data of their customers help them to be "inspired".
Even if the CEO is the most honest guy on earth, as long as someone can access the data from a way or another, I prefer to deal with lower speed performance (which is in my opinion pretty decent on my stuff anyway).



Yes, they were charged, but as you said, I didn't bother them for $200, not worth my time but I know that I have to keep an eye on that now.
And thank you for your kind words



I was using a LP with JS tracking, you're right, that's why this traffic wasn't recorded into the tracker...
By the way, do you know that the new version of P202 disconnect you suddenly when you are tweaking the filters in the group overview table?



Yeah, I can hear that, but just imagine the shitstorm if Google or Facebook were caught charging 50% of fraudulent clicks...
Why being so indulgent because we're doing affiliate marketing? We're still spending real money that goes into real pockets.
It's not a "game" to me, I'm very serious about the way I run my business, I don't know if big brands use these kind of advertising networks but I don't know if it's a game for them either

Very rude post but I totally get the point.



Very good point, your whole post is very clear and inspiring. Thank you

---

@pepeneo - That's a very good recommendation.

We wrote the initial Mobile Cookbook guide around a single DSP to prevent new affiliates from getting hideously confused, but as soon as you've got the basics down, I'd strongly recommend testing a few different traffic sources, just like you'd test a few different networks, offers or landing pages.

Also, bot detection techniques are really valuable. That's something which has advanced a lot since I wrote the first Mobile Cookbook parts, and I'll be including bot detection when I revise them later in the year.

---

Home > Paid Traffic Sources > Mobile