Bot test

[Martiko]

Hi to all,

I am running a bot test according to this Matuloo's guide:
http:// www .matuloo.com/bot-traffic-how-to-detect-fake-hits-and-what-to-do-about-it/

Unfortunately, I do not see bot clicks. I will explain it below. Firstly what I use:
Bemob as a tracker
PropellerAds is a traffic source
Mobidea is an affiliate offer.

I used the bot trap, which waits for 350ms and then redirects to the real campaign. Here is the screenshot:

The first line is bot trap and the second line is the real campaign. As I understood it, the first line should have about 6.557 clicks. And this is how it should help me to detect bots. But I do not see any clicks. Did I understand something wrong, or why I do not see clicks in the first line?

When I made a little check, I did not see anything suspicious. Either in IPs, or placements. So it looks for me, that some places are making human traffic and also bot traffic. If you know about a more accurate guide on how to setup bot, I will appreciate.

 

[Graybeard]

1. Use HTTP/2.0 that kills many old style bots
2. Ban datacenter IPs (allow those you want to have access)
3. Ban VPNs, tor as required -- these should be banned at the PCI-DSS (transaction level anyway)

Google headless is the new bot and is often hard to detect.

There is no guide that is not a bunch of bullshit (IMHO). Use an advertising network that scrubs their traffic well and pay the higher price. The problem with bot traffic is that it distorts your true metrics and makes optimization misleading.

 

[Martiko]

I am not very techie. I will be more specific with questions:
1. Http/2.0 - I googled it and found something. But truly frankly, I have no idea how to implement it, where to implement it. Does it rely on LP, or to the tracker?
2. How do I recognize datacenter IP? Why it's not good?
3. How can I ban VPNs, tor? - Do I understand properly, that this recommendation is for offers that convert by payments?

I understand what you typed - the problem with bot traffic distorts my metrics.
And yes, I am trying to use traffic networks, which are legit. The problem here also is, I am a newbie and have limited access to networks. No one wants a person with almost no skills.

 

[Martiko]

OK then, I will manage this differently and take your advice as a hint to consider in the future. But the #1 looks promising and doable immediately. Thanks.

 

[Graybeard]

Well, it took me years to learn how to do this and where the pitfalls and shortcomings are.
You can pay for a ad fraud mitigation service but they are expensive unless you are a network or a person with thousands a day in ad volume.


HTTP/2.0 is used by real browsers for years now. There are exceptions where this is problematic and you may deny some users that are using some form of scanning of your URLs.
I admin my own servers and manually write the virtual host configuration files and the other server level bot blocking files. I don't know of any way to do this properly with some *tool* The bot blocking configurations that I use are of my own making and require constant updating.

1. If you use push ads there will supposedly be fewer bots. pay CPC
2. If you can use SEM (search engine test ads) the links are protected for many bots
3. You will probably pay for 10 to 40% bot traffic depending on the network's filtering.
4. If you can find an ad network that uses Forensiq Forensiq Ad Fraud Detection - Impact - Impact that might be a good start

EPC (earnings per click) will vary by network on the same offer -- that is how you can segment and optimize. The bottom line is what matters when you haven't or can't control the funnel variables (of the referred customer/lead). At the end of the day; performance becomes the arbiter of fact.