Click Fraud - How do You deal?

[clearmachine]

Click fraud is a serious problem. In simple terms, click fraud is when someone clicks on an ad with the intention of generating revenue for themselves or causing financial harm to the advertiser. This happens through automated scripts, bots, or even people being paid to click on ads.

It typically is rampant in display ad networks like Google display where publishers get paid for people clicking their ads.

It was estimated that the costs related to digital advertising fraud worldwide would grow exponentially within the four years between 2018 and 2023, from 35 billion to 100 billion U.S. dollars.(source: https://www.statista.com/statistics...stimated that the,billion U.S dollars in 2022. )

So, how does click fraud work? There are a few common ways:

1. Automated bots: These are programs designed to click on ads automatically, often in large numbers, to generate revenue for the person behind the bot or to deplete an advertiser's budget. Some people try to block ip addresses for this but it is a worthless task since these bots are using residential ips.
2. Manual clicks & click farms: Some individuals may be paid to click on ads, either through third-party click farms or by competitors.
3. Competitors: Competing companies may use click fraud to deplete your budget.

So how do you deal with click fraud?

For native and display ads, it can help to look into tools like theoptimizer.io or Brax.io to automatically monitor and optimize campaigns. These tools use automated rules to help you prevent and block fraudulent clicks.

What are your tricks for dealing with this massive problem? From what I have heard using tools like Clickcease can actually result in really poor results.

 

[joeybabbs]

I have seen many ads for Clickcease for people running Google ads but honestly, I have also seen just as many objections to using it.

ClickCease apparently flags legitimate clicks as fraudulent, leading to false positives. This can result in blocking genuine visitors from seeing your ads, which can negatively impact your campaign performance. This has been proven by a few larger agencies that posted their results.

Unfortunately, ip blocking is not great and won't help.

One thing I noticed in your case study using Clickflare is that they have bot detection. This might be a great feature @cpv lab could implement and test but I am sure it would be a beast to code?

 

[cpv lab]

Hi,

This is indeed a problem. As a tracker which fits between a Traffic Source and an Affiliate Network, we had a lot of requests on this area.

What I can say is that sometimes it is very hard to catch the fraudulent traffic and to accurately identify such traffic.

In CPV Lab we have added some rules to filter and detect the Bot (from a few versions ago) and Proxy/ VPN traffic (from the recent version 10.0).
And this is helping in a few ways:
- if your traffic source is sending repeated bot traffic, then you can go with the report from the tracker and get a refund on that traffic
- you can block the IPs coming from bots and so exclude that traffic from your reports in order to optimize campaigns based on data from real visitors
- regarding Proxies and VPNs, you can also filter and have different rules for the traffic coming from them.

Optimizing campaigns with such traffic is not easy, but there are a few ways to do it.

Tools like theOptimizer or Brax have some good ideas, but personally, I used just one for testing purposes and there are parts I don't trust there. But that is probably just me, a privacy freak!

What other fraud clicks detection methods would you like to see included in a tracker? What would be useful for you?

Radu

 

[joeybabbs]

I think adding bot detection at least in some form of tracker notification or an additional column (likely a bot, not a bot) would be helpful. That or at least have a good resource on your blog or resources section describing how people can use your features to detect fraudulent placements & access those reports that show which IP address filters happened (& why). This could probably also be enhanced somehow with click timing analysis.

 

[cpv lab]

Hi @joeybabbs ,

Thank you for your ideas on having a resource created for this. Will do that! It is a very good idea!
The click-timing analysis is also a good idea... We will have to check a bit on the best approach here.

Regarding the Bot traffic filtering and also Proxy filtering, CPV Lab has that and it appears on your campaign stats and other reports.

All you have to do is:
- select to track bots and proxies when you setup your campaign

- Use the columns on the campaign stats to filter your data

Do you guys think we should track them by default for a campaign? Like having them checked by default so that you don't forget to check the tokens for Bot and Proxy?

We are still wondering what is the best approach from the user's perspective.
Your feedback would help!

Thanks,
Julia

 

[stuart]

Hi @joeybabbs ,

Thank you for your ideas on having a resource created for this. Will do that! It is a very good idea!
The click-timing analysis is also a good idea... We will have to check a bit on the best approach here.

Regarding the Bot traffic filtering and also Proxy filtering, CPV Lab has that and it appears on your campaign stats and other reports.

All you have to do is:
- select to track bots and proxies when you setup your campaign

View attachment 14963

- Use the columns on the campaign stats to filter your data

View attachment 14964

Do you guys think we should track them by default for a campaign? Like having them checked by default so that you don't forget to check the tokens for Bot and Proxy?

We are still wondering what is the best approach from the user's perspective.
Your feedback would help!

Thanks,
Julia

I think be default would be good - I mean imagine (and I can only imagine LOL) having started a successful long running campaign where you forgot to turn this on in the set up, that would be frustrating

 

[zlodei]

What I've done is created landing page with multiple offers and have 1st offer assign to invisible button (style, so only bots see it as button). That way I know only bots clicking it.

 

[joeybabbs]

What I've done is created landing page with multiple offers and have 1st offer assign to invisible button (style, so only bots see it as button). That way I know only bots clicking it.

Thanks, I have done this in the past as well. How do you implement yours? Also do you consider rthat maybe it's a review bot from say Bing or Google?

 

[zlodei]

Thanks, I have done this in the past as well. How do you implement yours? Also do you consider rthat maybe it's a review bot from say Bing or Google?

I styled white button on white background and placed it on top of the page so in the code it would appear first.
I only tested this on revcontent and was getting 50/50% of real and bot traffic on almost all of the widgets. Even on a widgets that I were getting conversions - I stll see a lot of bots clicking on that button, so not sure how to deal with that.

 

[joeybabbs]

I styled white button on white background and placed it on top of the page so in the code it would appear first.
I only tested this on revcontent and was getting 50/50% of real and bot traffic on almost all of the widgets. Even on a widgets that I were getting conversions - I stll see a lot of bots clicking on that button, so not sure how to deal with that.

That seems extremely high % of bots. I wonder if you are getting some false positives from some "friendly" bots if thats possible. Like a revcontent bot that scrapes your landing page or something...have you asked RC about why it's so high?

 

[zlodei]

That seems extremely high % of bots. I wonder if you are getting some false positives from some "friendly" bots if thats possible. Like a revcontent bot that scrapes your landing page or something...have you asked RC about why it's so high?

You are right - it could be some false positives. I already asked that question but haven't get a reply yet. I'll post an update once I hear what they say.