Through the conversations that I have, the events I visit and the online conversations I monitor, I do notice broad topics develop. Some come and go, but others stick around.
One of these is growing as a concern for all players in the digital advertising ecosystem, and quite rightly too: fraud, in which non-human clicks and impressions drain advertising budgets rather than reaching the right audiences.
First, let’s be realistic: no one can entirely stop or prevent fraud. It’s an unfortunate fact but one we have to face up to. Banks have to combat it, governments have to deal with it, and increasingly we as individuals are becoming more aware of the importance of protecting our identities, as social media and cloud-based services grow.
However I do believe that programmatic advertising for mobile has several characteristics that render it less susceptible to fraud than its desktop cousin. They mainly revolve around security inherent in mobile technology, and the nature of programmatic data.
Firstly, desktop advertising is heavily dependent on third-party cookies. These are what enable one website to display ads according to your activities on another, so for example if you were looking for white shirts on the John Lewis website, you might see an ad for the self-same shirts on the Guardian website.
However, there is very limited support for third-party cookies in mobile. This is why mobile has had to develop smart ways around the cookie issue, involving taking data from several sources which together build a profile of a user: the user’s device – that is, what make and model; the user’s platform, such as Android or iOS; and exchange data about the publisher and advertiser. We can bring this data together to form a unique and anonymous user ID, which we then cross-reference whenever a user wants to see a page. If the ID has been exposed to an ad before, we can retarget, positively or negatively.
These methods are actually much more difficult to exploit for fraud. It’s easier to replicate a third-party cookie, that is stored locally on your machine, than it is to replicate a Device ID. It’s harder still to synthesize a user ID that depends on huge amounts of data, and that is not even present on your machine.
And mobile also has ‘software police’, which are the app stores. If you’ve ever downloaded software from a website which also slips some malware onto your PC, you’ll know how careful you need to be with desktop. One variant of this is the stealthware that means a web window will suddenly pop up and then disappear again: it’s actually registering a false click or a page view for an ad to artificially boost a publisher’s earnings. But if you ensure that you always download apps from the official app stores, you can be assured that it’s gone through stringent approval policies and on-going monitoring from the ‘police’. It’s a much healthier ecosystem generally than desktop because of this degree of centralisation.
So mobile’s tech infrastructure is inherently less open to fraud. But there is also good news from the programmatic front, in that the quality of first-party data can be used to minimise fraud even further.
In programmatic we use four main types of data. Contextual data comes from the publisher and describes an ad space in terms of what the app or site is and where it’s being displayed; first-party data comes from the advertiser and is the profile they use to characterise the audiences they want to reach; second-party data comes from ad tech companies such as Byyd, who draw inferences from billions of ads served daily, such as a user’s propensity to click certain ads, their age, their gender and so on; and third-party data is bolted on from external sources, for example weather or geolocation data.
However I do believe that programmatic advertising for mobile has several characteristics that render it less susceptible to fraud than its desktop cousin.
The focus here is on advertisers who bring their own qualified audiences through their own identifiers, and who can then overlay that data onto programmatic buying platforms to make sure they reach the right people. This is what Weve has been doing for example, powered by Byyd’s technology. The joint venture between EE, O2 and Vodafone brings first-party data about their opted-in customer base and, as a result of this combination of programmatic performance and quality data, has achieved uplift in purchase intent of up to 39%.
So if you know as an advertiser who you want to target, and your first-party data is good, then you know you’re going to reach the right people. Your data essentially defines who you want to reach, and while this is not as ‘open’ as finding the inventory that simply works most effectively, it is definitely a means of shutting out unwanted audiences, including those who may yield false clicks.
Finally, running on quality, vetted inventory can further help prevent fraudulent activity. At Byyd we take brand safety extremely seriously and our platform, called Madison, includes further preventative methods including providing our advertisers transparency and control over the inventory they run their advertising campaigns across.
I’m confident that the open, programmatic means of handling mobile advertising is the way forward for handling scale while ensuring quality targeting. I shall be watching this issue over the next few months with interest.