Bots: the secret problem marketers are missing
Bots: the secret problem marketers are missing
A view from Duncan Trigg

Bots: the secret problem marketers are missing

Most advertisers naively think the negative impact of ads being seen by bots is limited to wasted impressions and budget. However, comScore's advertising effectiveness VP...

You may already know that ad fraud and non-human traffic (NHT) cost you money because your budget is being wasted on impressions that are never seen by people – the IAB estimates 36% of traffic is bogus. Based on the IAB’s latest display ad revenue figures, that’s about £943 million a year wasted in the UK alone. And this is probably where you think the problem – albeit a big one – ends. You couldn’t be more wrong.

What you probably don’t realise is these issues contaminate pretty much every performance and effectiveness metric you use to assess how your campaign performed. They infect your results, giving you misleading findings, causing you to make potentially bad decisions going forward.

I’ll illustrate exactly how on two major areas of campaign analysis but, essentially, ad fraud/NHT feeds artificially inflated numbers into your actual calculations, pulling down ad campaign ROI calculations and making campaign performance appear lower. The great news is, once you fix it, you’ll see campaign impact increase dramatically.

How NHT affects viewability

Let’s start with the impact on viewability – the table below shows how dramatic excluding fraudulent/NHT traffic can be.

Let’s say a campaign serves 15 million impressions. If we exclude 2 million of them as bogus, using basic NHT detection, and assuming 85% of the rest are viewable, we get a reported viewability of 74% (11.05 million divided by 15 million). However, if we exclude 5 million as bogus (much more realistic through more sophisticated measurement), the reported viewability drops to 57%.

Which one seems most appealing? The 74% of course. But it is a dangerous illusion.

For the sake of argument, let’s say each campaign delivered £100,000 in sales. In example A, the 8.5 million viewable impressions are doing all the work – an ROI of £11.76 per 1,000 impressions. In example B, 11.05 million are doing it – an ROI of £9.05.

Thus, here comes the bad decision going forward. The mistaken fixation on maximising viewability means campaign B is likely to be seen as more successful – viewability is 30% higher, relatively – but in fact campaign A is delivering 30% higher ROI. Hence you need to be much more accurate in identifying and routing out bogus traffic – not maximising viewability – so you can correctly optimise the right campaign going forward.

The problem is, obviously, far worse if you don’t take out any NHT at all – in our scenario you’d be led to believe 85% were viewable (a relative over-estimation by 50%).

How NHT contaminates audiences

Let’s say your targeting females. The example below shows that out of 20 impressions delivered you’d initially think 65% hit your target audience (5+8 / 20), whereas in reality, only 5 impressions (25%) hit human females.

So, not only are you getting a completely inaccurate sense of what the campaign delivered, you’d also be totally misled on its effectiveness.

Again, let’s illustrate the fall-out using an example – in this case if the campaign delivered £100 in sales. The 13 impressions delivered to ‘females’ would lead you to believe they’re each returning £7.70 in revenue. In reality, it’s just 5 human impressions that have done the work – £20 each, nearly three times the ROI you mistakenly believe.

In other words, you’re dramatically underestimating how effective the campaign is in when put in front of human eyes. Consequently, you can’t correctly optimise what you do going forward.

So, how to disinfect?

Of course, the impact of bogus traffic is considerably different between premium and non-premium sites so it always helps to minimise any impact by focusing on the former (although even they’re not immune as bots do open legitimate sites behind a user’s screen to create attractive profiles which are then targeted by marketers).

A study we did in the US late last year revealed NHT accounted for at least 5% of traffic on 1 in 3 non-premium sites compared to just 1 in 10 premium ones. It accounted for at least 25% of traffic on 9% of non-premium compared to less than 0.2% on premium sites.

As fraudulent or NHT contaminates everything it touches, you need to make sure whoever’s buying your impressions or measuring the results is using the most advanced filtering techniques to remove bogus traffic. Ideally what you need is a single unduplicated measure of impressions that are viewable to a human within a target audience – a verified impression.  In other words, could the advert have been seen by human eyes and in an environment that isn’t detrimental to my brand?

Only if you have this, are you getting an accurate sense of how the campaign performed and how to maximise effectiveness going forward. What it is undoubtedly clear is the impressions that are doing the work for you now are delivering a hell of a lot better than you think.