The insider's take on ad fraud: how ad platforms fight it and what users can do to protect their campaigns.

Ad fraud: An insider’s perspective

We’ve all encountered ad fraud in one way or another. Ad fraud has many faces, and so do the methods used to tackle it. This time, we sat down with Peter, RollerAds’ Head of RTB, to get the inside story: what’s happening under the hood of ad platforms, how they fight ad fraud, and how users can protect themselves if they suspect fraudulent activity.

You can either watch the video or read the summary below.

Fraud is a business

“It all comes down to money,” says Peter. In performance advertising someone gets paid per impression, per click, per conversion, so people will always fake those rather than actually earn them.

It can be a publisher inflating their own numbers. It can be some organized group generating fake traffic at scale and then selling it to the highest bidder.

Everyone else pays for it. Advertisers lose out on their budget and end up with campaign data they can’t trust. Publishers can be hit even harder with domain spoofing, where the fraudsters are pretending to be their site and the publisher loses both revenue and the trust of their audience. For the ad platform itself, the cost is trust. If the advertisers don’t believe in the traffic then they’re not going to stay.

It’s not just bots

When people think of ad fraud, they typically think about bots. According to Peter, two other types are click farms (where real people are actually paid to click on the ad) and malware-infected devices clicking behind the scenes. That’s why fraud is so hard to spot. Some of it looks just like real user activity.

How RollerAds checks traffic

We have a team that’s solely focused on traffic quality, and the checks go on all the time. Our anti-fraud system looks at over 50 technical and performance metrics to determine whether the traffic is coming from real users. It’s looking for emulation patterns: mismatched time zones, signs of device hijacking, and all the other oddities at the device and browser level.

Over 20 of the metrics are performance related. A source can pass every technical check and still be cut on the grounds of how its traffic behaves. We’re currently filtering out about 25% of the traffic before it even gets to the advertisers. 

Domain traffic gets another layer: every partner with parked, expired, or typo domains submits a moderation request with the details of their portfolio and traffic sources, and a manager will then manually go through that before they’re given access. If you want the complete picture, we’ve also written about how ad networks go about combating ad fraud.

Warning signs to watch out for

Even with careful filtering, some fraud can still get through. It’s always evolving and a new pattern can slip past all the checks until there’s enough data to demonstrate that it’s behaving differently. That’s why advertisers need to be looking at their own campaigns as well.

Peter splits the warning signs into two groups.

  • Direct indicators: sudden spikes in clicks, repeated actions from the same IP, device, or user agent, unusual location data, or if one subID is acting very differently from the rest.
  • Indirect indicators: very short sessions, unusually high bounce rates, numerous clicks but only a few completed actions, or a large number of registrations but almost no email confirmations or payments.

Indirect signs don’t necessarily prove fraud (they could just as easily be the result of a problem with the funnel, the offer, or the tracking). But when you’ve got a number of them all coming in at the same time, it’s time to go and have a look.

What to do if you suspect fraud

Contact your account manager or support team immediately. You should also thoroughly verify the funnel by ensuring that your domains are working, checking links and redirects, and ultimately tracing the user’s journey from clicking to converting.

A tracker is extremely useful in this scenario. If you’re reporting an issue to any traffic source you’ll want to be able to provide your tracker stats, your click logs (with the IPs and user agents), and your landing page CTR. As Peter points out, that’s generally the bit that decides the outcome of the case. A lot of advertisers get stuck at that point as they don’t track at all and so have nothing to go on. If you haven’t got a tracker in place yet, go ahead and set one up using our postback tracking guide.

We look at the data and compare it to our anti-fraud metrics and internal logs to find out where it came from. If it’s found to be fraud, we resolve it by blocking the source and settling fairly with the advertiser. The findings then go into our detection rules to easily identify if this occurs again.

Can anyone ever be fully protected from ad fraud?

Not really. No one in the industry can guarantee that fraud will never happen—there are whole companies that exist to combat ad fraud and if there was a truly effective solution, this would have been resolved years ago.

The best approach is to filter on our end, have advertisers keep an eye on their campaigns, and be able to quickly dive in if something looks suspicious. You just have to be able to detect fraud quicker and intervene before it does more damage. Check out our guide to protecting your campaigns from invalid traffic for a step-by-step checklist.

If you’re looking for an ad platform that takes traffic quality seriously and keeps working to detect and prevent ad fraud, we’d be happy to have you on board. Sign up and explore the traffic opportunities on RollerAds.

Boost your campaign revenue with RollerAds

Recent Posts