The debate over transparency in digital advertising is far from over
YouTube is the latest player drawn into the mêlée, following an investigation by The Times that revealed in mid-March that the video sharing platform was allowing hateful and violent advertisements to slip through its filters unnoticed. The resulting outcry led to rather radical reactions in the UK. First, advertisers: McDonald’s, HSBC, and L’Oréal pulled their ads from the platform. Agencies were next, with the British Havas office going so far as to pull its ads from the entire Google network for all its client portfolios.
“Taxpayers are funding extermism”: headline from The Times in reaction to an ad from The Guardian – funded by the British government – broadcast on an Islamist channel that received compensation.
Ultimately, the controversy was the opportunity for the digital media industry to reopen the debate – already lively enough – about transparency and quality in digital advertising in terms of brand safetybrand safetyThe concept of brand safety refers to the necessity of not damaging an advertiser's brand image, by making sure that its ads are not displayed in an environment that could harm it or not be aligned with its values (for instance, on a pornographic website or on a website calling for violence). This issue is all the more important since the buying and selling of advertising space is now mainly automated (programmatic buying).Learn more. The difficulty, underlined by this scandal, is guaranteeing advertising broadcasting in an environment that is fully in line with the advertiser’s values, in the age of programmaticprogrammaticProgrammatic advertising refers to the automated buying and selling of ad spaces; most of the time, it takes place in real time through platforms such as ad exchanges, DSPs and SSPs.Learn more advertising and audience planning.
Though Google has been singled out as the troublemaker, the responsibility of protecting brand identity is not limited to advertising networks. Rather, it includes a much longer chain of players, each responsible at their own level for the quality of ad delivery and brand safety. This article will try to help you understand what is hiding behind this term, and what you can do to minimise the risk.
Not to be confused: brand safety, visibility, anti-fraud measures
To avoid any confusion, let’s take a quick look at the challenges facing ad quality:
- Visibility: “Are my bannerbannerA banner is a commonly used ad format, that takes the form rectangular-shaped image (still or animated), and is displayed on publisher’s advertising spaces.Learn mores truly visible – have they really been seen?”
- Reached audience: “Is there a real audience seeing my ads, rather than fake robot traffic?”
- Exposure context: “Is my work being shown in an environment that is coherent with my brand and its values?”
Brand safety only refers to this last question. Typically, brand safety means preventing banners from being displayed on extremist, violent, or pornographic videos – but it can also mean preventing brands like Nutella from being advertised on weight loss sites, or ads for a site like Ashley Madison from appearing on conservative media. Brand safety does not only deal with the editorial angle, however. It can also mean making decisions based on the current news context: for example, preventing banner ads for British Airways from appearing on the Times website on the day of an airplane crash. In conclusion, brand safety is simply ensuring an unhindered and foolproof perspective of the total reach of the ad delivery network, and being able to anticipate or detect the slightest lapse.
Programmatic ecosystems don’t make things any easier
There was a time where advertisers would buy ad space through media agencies on websites that seemed to have similar target audiences. They would receive their ad trackingtrackingTracking refers to the tools and methodologies that measure the activity and behaviour of visitors on a website (or a mobile app), including their journey, the source of their visit, or exposure to ads... Learn more report, and could proudly surf around the web pages in question to admire their banners. Then, programmatic buying arrived, destroying the old model, multiplying paths and middlemen between the advertiser and its ad, to ultimately become the market norm: in the US, it now accounts for 73% of display ad spending in 2016, with a rate of growth of 45% between 2015 and 2016 (representing $25 billion invested in the US alone).
Gone are the days where banner placements were chosen ahead of time: the decision to display a given ad is made in milliseconds, using a bidding system that takes several complex variables into account. Programmatic advertising no longer targets ad placements, and instead targets moving audiences. The analysis of advertising delivery and performance is getting more complex too, with audiences, segments, scenarisation, and data purchasing to consider. Possibilities are endless, as are KPIKPIKey Performance Indicators (KPIs) refer to the main indicators assessing the success of a campaign: they are determined based on their accordance with the chosen strategy. The generated turnover, or the number of pageviews, are examples of KPIs. Learn mores to analyse performances.
Faced with this two-pronged dynamic of growth and complexification, abuses and errors have multiplied, and the digital media market has not necessarily gotten good press. Today, YouTube is in the line of fire over brand safety. In 2016, Facebook was rapped on the knuckles for being careless after having made errors in its video KPI measurement for years. The consequence? Advertisers are increasingly cautious and demand better tracking and information sharing.
Scandals aside, is the brand safety situation really so serious?
Today, the number of “risky”* digital impressionimpressionIn the context of online advertising, an impression is the display of an advertisement on the Internet. It is also used as an indicator in the buying and selling of ad space: some publishers charge advertisers for each set of a thousand impressions (especially for CPM-based pricing models).Learn mores hovers between 3% and 10% depending on the country. France has remained fairly stable with a rate of 3.4%, while the US exceeds 8% according to a study carried out by Integral Ad Science in the second semester of 2016 (for all types of media buying).
Though the programmatic approach shows higher rates on average, online purchasing also suffers from inconvenience, with a risk rate far from zero: 4.6% in the United States for online buying versus 9.5% for programmatic buying. This means that though programmatic buying is definitely not solely responsible for risky advertising, it is still worth remedying the problem, and initiating a wider reflection on the regulation of the digital media buying industry.
So who is responsible for brand safety and how can risks be brought closer to 0?
From the provision of an advertisement to the moment it is displayed, brand safety is far from perfect, but merits special and regular vigilance for all players in the value chain of media buying. Here are some operational best practices to minimise risky advertising.
Beforehand: everything should start with a real push from the advertiser, making sure to define and communicate content requirements. These requirements will guide the media agency as it defines a buying strategy. For example, private deals can be made (between an advertiser and a publisher, by outlining the broadcast inventory), and private ad exchangead exchangeAn ad exchange is a platform that automatically runs the buying and selling of media advertising impressions; the price of these impressions is determined through automated bidding, between buyers of advertising space (advertisers, media agencies...) via DSPs, and inventory sellers (publishers) via SSPs.Learn mores can be set up. Remaining in closed networks should theoretically help to limit risk.
The teams should then manually control part of the delivery network settings by establishing lists of sites to be blacklisted in the DSPDSPA DSP (Demand Side Platform) is a platform that allows advertisers and agencies to automatically manage their media-buy on multiple ad exchanges through a unique platform; it is mainly used to buy display advertising space through RTB (Real Time Bidding).
Learn more. Each DSP has its own native content verification tool, and acts as a first filter. However, the methodology is often opaque and generic, and is thus limited.
In real time, it is possible to use “pre-bid” segments provided by third-party ad verificationad verificationAd verification tools allow to insure that ads are visible on web pages and that they are displayed in an appropriate way, that is to say seen by the chosen target and broadcast in an environment that does not harm the brand's image. These tools fight against ad fraud and protect brands from the potential drifts of automated ad buying. Learn more players. Information about the delivery framework is analysed with each bid request, and cross-checked with more detailed data on the broadcast site and its content. The advantage of this solution is its automated reactivity in real-time, while blacklisting remains fixed and can be optimised only manually. Third-party players on the market such as Adloox and Integral Ad Science offer these pre-bid segments, generally natively integrated in DSPs and resulting in an extra cost added to the CPMCPMCost Per Mille (CPM), or Cost Per Thousand (in Latin, "mille" means thousand), is an online advertising pricing model that refers to the cost an advertiser pays for one thousand impressions of an ad. Unlike other revenue models, it is well fit for a mass strategy. Some actors of the advertising ecosystem buy CPM ad space and resell it to advertisers on a CPC basis (Criteo, for instance).Learn more. The effectiveness of these solutions depends on their ability to collect granular enough information from pages to be analysed and cross-checked (keywords, video content…). For example, today on YouTube only titles and comments – but not the video content – can be detected by some players.
Lastly, the work should be completed with continuous verification afterwards, by using delivery reports available in media tracking tools, or via third-party verification tools such as MOAT, Integral Ad Science, or Adloox. At 55, in addition to automatic alerts, a team of media experts is in charge of building and analysing these delivery reports to detect and exclude sites that are deemed risky.
Brand safety thus relies on operational precision, and good intelligence shared by various players
However, zero risk remains evidently impossible to attain – but much progress can still be made to minimise risks and this is an important stake for the advertising industry on the whole. Everyone must get involved: advertisers, agencies, publishers, tools, and tech platforms.
The “bad idea” that some are promoting would be to ask platforms to auto-regulate content themselves and to decide if specific ad delivery networks are acceptable or not. On the one hand, the subject is too specific to each brand, culture, or tendency – it seems impossible to leave the decision to one sole decider (Facebook’s censoring of a Courbet masterpiece is proof!). On the other hand, it’s not up to tech giants to issue these kinds of injunctions – it’s up to third-party judges.
The good news is that the industry is taking steps in this direction: YouTube recently announced that it would work with Integral Ad Science to unlock its walled garden in order to guarantee brand safety and win back advertisers’ trust.