Programmatic: How we can stop fraud before the crime

As a buy/sell industry, we are busy following two fundamentally flawed rules:

1. We are looking at viewability and fraud as separate, stand alone metrics

2. We’re judging success against the average benchmark as the ‘we’re doing good enough’ barometer of success

There are agency trading desks relying on a post (after the) bid score, to measure their client’s campaign efficiency. What’s worse is that this ‘score’ is regularly counting botnet fraud and non-human traffic as viewable, premium inventory. There are publisher networks being told their viewability has dropped below the acceptable range, but not being told ‘why’. The ‘good enough benchmark’ FYI, is currently 40 to 45 per cent viewability. Adloox’s research has shown up to 21 per cent of this is actually fraud. Benchmarks are skewed and inaccurate. Numbers aren’t fully trusted, so what do agencies do? They employ a second viewability vendor. But this is also flawed. The viewability-only vendor is not capable of detecting the hidden, fake traffic and just reports it as human, viewable traffic. The real irony is that vendor two’s stats are almost identical to vendor one. Eureka! We’re paying twice for the same data, and it’s working, right? Wrong…as they are both missing the hidden fraud.

The fraudsters have logged off. The crime committed. The brand’s money? Already wasted. The Publisher? Taken off plan. It’s much like the film The Minority Report. We need to spot the murderer before the crime is even committed. If we are to stop online fraud, we need to be quicker. We are talking milliseconds here. Putting it into context, it takes about 300 milliseconds for an ad to deliver on what’s called the declared domain. That’s the website everyone reports on, scores and optimises against. Yet there are symptomatic signs of fraudulent activity much earlier in the delivery process – between zero and 300 milliseconds, which allow us to pre-determine whether a user/sub-domain is human or not. It’s this almost invisible layer of inefficiency, which is affecting overall campaign performance, and everyone’s left scratching their heads.

So what’s the solution? If pre-bid is used effectively, there will be successes both sides of the programmatic fence. Pre-bid decisioning can be made within just five milliseconds, preventing the fraud doing its damage. At Adloox, we’ve created a pre-bid algorithm allowing buy and sell side partners to blacklist beneath the surface (fake users not websites). This allows publishers to work more transparently to sell real, viewable traffic, and allows agencies to bid with more highly-trusted suppliers.

Full transparency. Stopping the crime from ever being committed.

The industry ‘accepted’ average of fraudulent traffic is five to 10 per cent. Personally, I don’t think that’s good enough. By catching the fraud before rather than after the event, we are seeing increases in viewability to over 70 to 80 per cent, and fraud/non-brand safe content reduced to under 0.2 per cent. Hitch Digital, a leading global RTB network, recently made the switch from multiple viewability and fraud vendors to a full suite provider. “We are now able to catch the fraudsters on the way in,” says CEO Leigh Hitch. “ Our investigations into fraud stretch much deeper, and allow us to bake Adloox’s data into our algorithms, driving performance. Pre-bid fraud exclusion means we no longer have to reconcile after the event with our clients. It’s all clean before they buy.”

It doesn’t need to be a science….but it does need to work.

Marco Ricci, CEO, Adloox

Tel: +44 7855 836750

Email: marketing@adloox.com

Web: www.adloox.com

Twitter: @adloox_live