The ASA strengthens focus on regulating online ads with potential use of machine learning


The ASA’s strategy

In November 2018, the Advertising Standards Authority (“ASA”) released its latest five-year corporate strategy vision, outlining its plans, aims and focus over the next five years (the “Strategy”). We have picked out below some of the more interesting objectives and commitments made by the ASA, which are likely to be of interest to brand owners and advertisers across the industry, as well as looking a bit more closely at some ways in which the ASA may realise its aim of using new technologies to improve regulation and enforcement.

Online focus

The headline statement of the Strategy is a renewed and strengthened focus on the regulation of online advertising. No reader will be surprised to learn that an increasing number of brands will be focusing their advertising spend on the online market – and given the meteoric ongoing rise of programmatic advertising – this trend is undoubtedly set to continue. In terms of regulation, the ASA’s figures back this point up, with the vast majority (88%) of the 7,099 ads amended or withdrawn following ASA action in 2017 being online ads. Similarly, over two third of all the cases resolved by the ASA last year were regarding ads that appeared online.

Online advertising has long been seen as a bit of a “wild west” in terms of regulation and we will all have personal experience of seeing misleading, incorrect and potentially offensive or harmful advertising when browsing online. With the vast consumption of media taking place online across multiple devices and in new and immersive ways, it makes sense that the ASA should redouble its attention to protect us in the online sphere. But in a world where advertising or promotional content can be created and uploaded by anyone within seconds to a national or global audience, how can the ASA go about ensuring it is fair, responsible and not harmful? And equally importantly how can it ensure it is addressed, amended or removed in good time?


Another of the key strands for the Strategy will be getting the “buy-in” of both the public and the industry. The ASA will be seeking to work a lot more closely with the “large online platforms”, invite their engagement and collaboratively explore ways to ensure advertising compliance. It remains to be seen exactly how this will work, but the goal at least appears to make sense, both from a technological and practical context.

We have seen recently – particularly in the context of targeting – both the ASA and advertisers under investigation relying on or referring to tools, statistics and data from the likes of YouTube as part of the analysis of the ad in question – be it the ad’s viewership, exposure, tagging etc. If the ASA can work more closely with platforms such as YouTube, it is hoped the result is that – through new technology and tools – brands and ad agencies will be able to more effectively tailor and direct videos to their intended audiences. However, particularly with platforms such as YouTube where there is no requirement to “sign-in” to watch a video (and therefore YouTube has no data about, for example, the age of the viewer) – it appears that ads unsuitable for children may continue to be viewed by them in this context.

Platforms engaging with, and addressing, regulation is something which has already happened with other social media giants such as Instagram, whereby functionality now exists during the process of uploading content to allow brands and other users to identify such content as paid for advertising, using a check-box tool. This has helped to in some way address the regulatory issue of clearly labelling content as paid-for – where ensuring an ad is labelled prior to a user engaging with it is not always technologically straightforward. This is a classic example of technology adapting to incorporate regulation – something which the ASA will be keen to continue to promote.

As well as achieving the engagement of the giant social media platforms, the ASA also stresses the importance of getting small business and indeed social media influencers themselves on board. Much work has already been done in this area, including the ASA’s recent guidance – prepared to educate influencers and raise awareness of the potential issues and pitfalls to avoid when advertising both on behalf of themselves and when promoting brands by way of paid-for content. The ASA hopes this engagement will assist in educating the content creators and pro-actively reduce the number of misleading or irresponsible ads appearing online.


One of the key challenges facing the ASA in an online context will always be enforcement – or rather effective, prompt enforcement. The Strategy looks to explore the ASA’s “decision making processes and governance”, to, “simplify our regulation” and to ensure the ASA acts, “more nimbly”. Given the instant nature of the online world, the ASA’s procedures will need to be streamlined to ensure that an ad can be identified, investigated and if necessary removed or amended within a relatively short timeframe. There is reduced worth in an online advertiser being told to remove an ad many months after it has been promoted to the public, garnered attention and “gone viral” – potentially causing harm and offence along the way while generating significant publicity for the brand. To that end, as discussed below, the ASA will be looking to target the ads that matter most to the public and to ensure cases are dealt with and resolved more rapidly to maintain confidence and ensure the enforcement is effective.

The nature of the investigative process means that enforcement and removal are never going to be in real-time and there is always likely to be a delay. However, if the timeframes can be streamlined and reduced then it will undoubtedly assist in making the ASA more visibly effective.

Ad regulation and machine learning

Interestingly, the ASA has also set out in its Strategy a desire to explore how machine learning could improve its regulation. Although this is a broad and fairly vague commitment at this stage, it is no different to what corporates in all industries have been doing of late – i.e. assessing how efficiencies and improvements can be achieved through exploitation of new technologies such as “AI” or machine learning.

Elsewhere in the Strategy there are further hints as to how the ASA may utilise the latest technological developments to aid its cause. For example, there is an indication that some form of machine learning will be used to pore over and analyse social media “chatter” in order to identify and assess what irresponsible advertisements are causing a stir. This is a specific example, but more broadly the Strategy talks of “more technological tools (including machine learning)” that will assist in trawling the web to identify non-compliant advertising – thus enabling the ASA to, for example, root out scams earlier for referral to other statutory regulators.

The ASA also plans to use “data driven intelligence gathering” in order to assess which ads to target with enforcement action – i.e. which mean most to people and which are likely to have the biggest or most harmful impact. This will enable the ASA to better prioritise and focus its resources on those ads that matter most. What form this intelligence gathering takes remains to be seen, but it appears the targeted approach at the most impactful ads would be a deviation from the ASA’s previous remit of addressing each and every complaint with at least some form of action.

It is difficult to see how machine learning may impact on the regulation itself in terms of decision-making and enforcement. Given the inherently subjective nature of any decision and the scope for interpretation of the CAP Code – as well as the need to consider and interpret arguments and data provided by both the complainant and the advertiser – it is difficult to see what part machine learning may play. As a supplementary tool however, it is easy to see how it could assist in the data gathering process in order to identify inappropriate ads, make the most efficient use of ASA resources and to achieve the most prompt outcome.

Inappropriate targeting

Data, its value and importance are inextricably linked with advertising in the modern day. The very functioning of programmatic and other online advertising is built on a wealth of data about consumers. As such it makes sense that the ASA is also intending to increase its work with the ICO and to have more meaningful input into the use of data for “data-driven marketing”. With targeted advertising becoming ever more sophisticated and with the wealth of enriched data that advertisers now have access to – it is easy to see why inappropriate targeting will be firmly on the ASA’s agenda. The significant number of ASA adjudications in the past year surrounding ads inappropriately targeted at children (e.g. pre-rolls prior to a YouTube video for kids, or as banners or pop-ups as part of a child-focused website) are likely to continue as the ASA redoubles its efforts in this area – with a particular focus stated in the Strategy on protecting children and other vulnerable sectors of society.

What does this mean for brands and advertisers?

At Bristows we regularly advise client brands and ad agencies on the interpretation of the CAP Code and how best to ensure compliance while retaining the artistic aim or purpose of an ad. In terms of the effect of the Strategy for such clients – given the nature of the document, the plans involved are long term and therefore no immediate changes are expected. While we can expect to see greater focus and enforcement by the ASA in targeting online ads, it remains to be seen how its processes and tools will evolve to meet this significant challenge. We will follow further updates from the ASA in this area with interest.