IAB Australia has recently issued an urgent call to action for brands, agencies, ad verification firms, and other companies in the digital advertising supply chain to better manage their blocking of media buying against topics including keywords such as ‘crisis’, ‘COVID-19’ and ‘coronavirus’. Australian media owners are seeing huge jumps in online traffic, but many are said to be suffering as brand spends are being blocked from appearing around any content mentioning these specific keywords. Terms such as ‘coronavirus’ are now so ubiquitous this is having an impact, both in terms of overall spend on premium content sites but also for advertisers be unable to effectively access these regularly returning users.
We have been keen to build awareness around this topic but have also made some simple recommendations in relation to buyers utilising semantic contextual targeting to fully understand the editorial context. We are also currently working with the MFA and AANA to update the Australian Digital Advertising Practices which will soon be released. However, we were also keen to provide further perspective and guidance on this topic through some of the key verification vendors that are our members and sit on our various councils.
With that in mind, we have since prepared a few questions for our verification vendor members to which we’ve received the responses below.
We hope that you find this useful…
June Cheung, Regional Director AU & NZ, Oracle Data Cloud
How does contextual targeting differ from the use of simple blocklists of keywords?
Blacklisting is a simplistic brand safety tactic that blocks ads based on keywords appearing in URLs or nearby content. Because traditional blacklisting focuses on individual words, rather than the context in which they appear, it can be overly broad, and it can block a wide range of safe and brand-appropriate content. Blacklisting can also be difficult to customise for the unique needs of each advertiser.
Context-based advertising, on the other hand, allows advertisers to identify brand-suitable environments for their advertising based on the context of each word, not just individual terms, so advertisers can utilize more inventory and reach more consumers in safe and brand appropriate environments. Oracle’s Contextual Intelligence evaluates and categorizes billions of web pages and other digital content, so advertisers can not only avoid inappropriate or unsafe content but also find content that reaches their customers and achieves their business objectives.
Isn’t this also an opportunity for buyers to be more creative with their targeting attributes and utilize them in conjunction with publisher’s audience data to take advantage of the enormous levels of regularly returning highly engaged audiences on premium content sites? Can this help buyers to build-out models that are much more precise, meaningful and robust?
We believe there’s always opportunity for all parts of the industry to work more closely together to customise solutions or apply more creative strategies to advertisers’ needs. Precise solutions already exists via contextual targeting, next step is further education to ensure the right questions and considerations are made when implementing these strategies to maximise our ability to capture highly engaged audiences across premium content sites.
Imran Masood, Country Manager Australia & New Zealand, DoubleVerify
How are you currently supporting your clients with these challenges and opportunities, both in their Postbid and Prebid programmatic buying?
DoubleVerify are focused on being empathetic listeners who deliver solutions and guidance that enable our clients to adapt and grow. This means providing unparalleled customer service and communication. As the crisis ramped, we swiftly offered guidance on how to use our tools and technology to ensure that brands adapt their suitability strategies during the coronavirus pandemic, while still supporting trusted news outlets.
We believe that every advertiser sets its own definition for brand suitability, as each brand has its own unique positioning, its own target audience and its own set of brand values and sensitivities. DV’s solution offers the ability to entirely customize settings – down to the media partner, site, app and, in the case of exceptions, at the URL level.
Specific actions brands can take to achieve a balance between protection and scale include:
- Content Classification – Brands should consider exempting trusted news publishers from our “Natural Disaster” content classification category.
- Keyword Blocklists – Advertisers should also consider exempting trusted news providers from coronavirus-related keyword blocklists.
- Exception Lists – Brands should strongly consider adding trusted news site homepages and section pages (National, Health, etc.) to their DV page exception lists. Page exception lists allow a brand’s ads to run irrespective of the brand suitability settings applied to the rest of their media plan. This is especially useful for programmatic buys and on high-volume entry pages where the consumer associates the brand with the news publication more than the dynamic and aggregated context of the content on the page.
These recommendations apply to both pre-and post-bid buying. All of the above settings are configurable for post-bid monitoring and blocking, and for pre-bid, DV has a unique offering called Authentic Brand Safety targeting. Authentic Brand Safety targeting allows brands to perfectly align their pre- and post-bid brand suitability criteria, creating a single segment that can be deployed across multiple buying platforms. Authentic Brand Safety targeting is the most advanced brand safety solution in the market today, offering heightened levels of advertiser protection and driving substantial gains in both operating efficiency and campaign performance. Not only does Authentic Brand Safety targeting ensure all the avoidance criteria for brand suitability are activated, it also honours the exceptions policies put in place by advertisers to extend reach to publications or pages they’re comfortable running on. For example, a buyer can avoid content in the Natural Disaster category generally while creating exceptions for major newspaper sites, apps or home pages to ensure they continue to bid on impressions from those trustworthy publications.
Do buyers need to better understand the verification vendor methodologies?
It is critical not only to understand the methodology your verification vendor employs for things like keyword blocking and category avoidance, but also to understand the tools at your disposal to customize your settings. Methodology will enable you to understand how content is being evaluated for the purposes of avoidance, blocking and monitoring. For example, DV’s Sematic Science Engine, working in conjunction with human linguists and classification analysts, comprehensively evaluates content in over 40 languages and assigns it to over 75 standard brand suitability categories. These standard categories are available to all brands and provide controls across broad suitability concerns, like Adult Content and Hate Speech, as well as more narrow topics specific to certain brands, such as Aviation Disaster. DV also offers a unique Inflammatory News and Politics category that protects brands against fake news and highly-polarized political content.
Brand suitability tools give you control over how and when avoidance, blocking and monitoring occurs. DV allows for different configurations of brand suitability settings to be activated on a per campaign basis or, within a campaign, by individual media sellers on the media plan. This gives a brand flexibility to build and apply different configurations of brand suitability settings per seller.
Poppy Hill, Managing Director APAC, OpenSlate
How are your Social Platform clients managing any coronavirus-related content during the current crisis?
The COVID-19 pandemic is likely to be the predominant news topic for some time. That being said, Coronavirus content is by no means limited to news. Social platforms reflect the world at large; just as nearly everyone is grappling with the implications of Coronavirus, creators across multiple content genres (travel, entertainment, family) are addressing it as well. OpenSlate is running daily queries of our database and has found the volume of Coronavirus-related content is increasing anywhere between 5-10% per day.
The social platforms are taking action regarding Coronavirus content. YouTube at first demonetized all related content, but then reversed their decision. At present, Facebook is focused on combating misinformation among other efforts.
OpenSlate has already augmented our data science models to flag Coronavirus misinformation (e.g. “bat soup”) so that clients aren’t funding creators spreading false information. We recommend that all advertisers take proactive action to avoid monetizing inaccurate, misleading and harmful content.
How can brands use technology to better support news content media in these critical times?
With the right approach, brands can support premium and credible news environments that attract high audience engagement across general and breaking news content. OpenSlate consults with brands to reflect and take control of their news suitability guidelines and standards first to establish a pre-campaign solution that determines what content they should be aligned too.
OpenSlate’s Content Categories provides brand’s with precision targeting to specific channels with content topics and themes that they deem most relevant for their campaign. Our proprietary taxonomy and analytics identify the most prevalent content type, ensuring brands can deliver within contextually relevant, quality environments, allowing advertisers to effectively reach engaged audiences at scale.
James Diamond, Managing Director, Australia & New Zealand, Integral Ad Science
Some industry commentators have said that most advertisers are blocking on covid-related keywords and that the only real winners in this are the ad-blocking technologies?
Firstly, regardless of whether an ad passes or is blocked we get paid the same. There is no incentive for us to drive the rates of blocking up and any suggestion that we win when more ads are blocked is wrong.
Secondly, while some Australian advertisers did add covid-related keywords to their blacklist, the vast majority did not. In Australia, we supported 355 advertisers in March across many campaigns. Only 25 Australian advertisers had added ‘coronavirus’ related keywords on their blacklist. That’s 7% of advertisers that are still in market deciding to actively block on covid-related keywords.
25 advertisers may not sound like many but if they are the largest advertisers in the market, they could be blocking hundreds of millions of impressions and costing publishers millions of dollars in lost revenue.
Firstly, the volume of ads that advertisers asked us to block for covid related keywords in Australia for the month of March is less than 15m impressions.
Secondly, when an advertiser asks us to block an ad it does not always result in the publisher not getting paid for the blocked impression. In fact, for the Australian market the publisher is generally paid for the blocked impression. There are two scenarios that result in a blocked impression being paid for.
The first is if the impression is purchased programmatically it will be paid for regardless of a block/no block decision. If an advertiser places a bid and wins, they pay. If they subsequently decide to block the creative from displaying, they will not get a refund.
The second scenario is where the publisher ad server is used as the bill of record. When an ad is blocked the ad request never reaches the 3rd party ad server/agency adserver, but the publisher ad server still records the impression. So for buyers that are accepting the publisher ad server count for billing purposes, they will pay for blocked ads, and many Australian media buyers still use this approach because they believe that publishers should not be punished for a brands decision to block.
Of course, there are exceptions, everyone’s tech works differently, buyers and sellers have different commercial terms to deal with blocked ads that may include making good on blocked impressions and so on, but my experience has been that most of the ads blocked are still paid for in Australia.
Some are asking the question – shouldn’t advertisers be brave enough to have their messaging appear beside the critical content that Australians are currently consuming? Is this really the time for an endless debate about brand safety?
Having talked with many advertisers and media agencies about appearing alongside covid-related content, I saw brands fall into one of four buckets.
Bucket 1: No work – These advertisers are not blocking specific covid-related content, rather they are out of the market all together mostly because their shops are closed and as such they can’t capture any demand they stimulate. It is also the airlines, the cruise ships, tourism boards and so on. Most do still want to be in the market with a message of sorts.
Bucket 2: Loads of work – These advertisers are overrun with demand and they don’t need to stimulate more demand if they are already out of stock. Many of these advertisers still want to have a message in market, but one that is less about ‘buy now’ and more about longer-term brand building. However, before they are back advertising again they need to jump through bucket 3.
Bucket 3: Rework – These advertisers want to be in market and are happy to run against covid-related content, but last months’ marketing message now just comes across as completely tone-deaf. Best case they need to rework their messaging and creative, which is difficult when you are doing it remotely for the first time. Creative teams that might be used to jumping in a room with a whiteboard and a bunch of coloured post-it notes now need to collaborate over VC…. It’s going to be slower.
For others in this bucket, they are not just needing to rework message and creative, but rather their go-to-market strategy altogether. Again, this is going to take time. It’s going to be frustrating for publishers as they wait for this reworking process to finish and campaigns come back online. But from what I am hearing brands plan to be back soon with a new, more appropriate, advertising message.
Bucket 4: Can’t work – These are brands that are staying in market but where advertising against covid-related content just makes little sense and so they will continue to block against covid-related keywords. It is not that they find the content bad, but it’s just an inappropriate brand fit. If your slogan is “oh what a feeling” it’s going to be difficult for those ads can sit alongside content about people feeling terrible, feeling sick and feeling sad. So far brands in this bucket seem to be continuing to advertisers but just choose to not run on any covid-related content.
I have never been responsible for a brand and so I am not in a position to judge the decisions that brand managers take during this crises, but I don’t think it’s a question of bravery. Some of these brands have taken decades to build, they sit as large intangible assets on the balance sheets and represent significant shareholder value. Given that, I think it’s less endless chatter about brand safety and more about those people who are tasked with guiding a brand through these difficult times are putting the right processes in place to mitigate downside risks.
Tim Beard, Managing Director APAC, Silverbullet
How does leveraging the capabilities of contextual compare to simply managing brand safety requirements via blocklists?
The main problem with blocklists stems from outdated approaches. Simply analysing random keywords without context gives you a very high probability to misinterpret, or misrepresent what an article is really about. That’s where blocklists fall down every time. They are a blunt tool from a bygone era.
Their original intention was to help advertisers avoid appearing next to content that could put their brand at risk. This can be helpful, but when it fails it usually fails spectacularly. Articles about boxing are blocked because of the word ‘fight’, or content about Meghan Markle is blocked because the word ’’sex’ appears in ‘Duchess of Sussex’.
Good contextual targeting, like what we’re focused on at 4d, uses natural language processing and other advanced techniques to truly understand the semantics and context of an entire article the same way a human would. So when we see the word ‘fight’ in an article that also has the keywords ‘Anthony Mundine’, and ‘super-middleweight champion’, we’re able to determine the content is about boxing and still suitable.
The industry needs to begin using smarter and more nuanced solutions to help brands find suitable content to advertise on, while also helping publishers monetise reputable content that needlessly ends up on the cutting-room floor.
Is there also an opportunity now for brands to be more creative and take advantage of the increased traffic on premium content sites?
Absolutely.
The biggest problem with the industries reliance on Blocklists is that we’re ignoring the value of important publishers and unintentionally turning the internet into a worse place. We’re penalising reputable news sites and content creators by taking away the ad dollars that keep them in business, and moving those budgets to safer but less impactful long-tail publishers. While these long-tail sites are great, I don’t think anyone wants an internet that’s exclusively cat memes and food recipes.
The better approach here is to marry valuable and unique publisher assets with a contextual solution that’s smart enough to understand contextual nuance, and flexible enough to support custom brand safety requirements. Every brand is different and has different safety thresholds and concerns, so you need to also build a safety strategy that accounts for that. With this combination you can block the content and sentiments that a brand wants to avoid, without unintentionally blocking everything.
A great example of this is the 4d partnership with Factmata. Using natural language processing we can not only understand the whole context of a web page and tell an advertiser exactly what it’s about, but we can also score it based on nuanced topics like fake news, political bias, or clickbait.
Using that approach, a publisher could enhance brand suitability in a much more intelligent and nuanced fashion. You can guarantee an advertiser that while their ad may appear next to premium content about Coronavirus, it will always be in a positive context, alongside content that is factually accurate and reputable, while still avoiding any unique/brand specific safety.
