Whilst 'context' has been sidelined in the past, it is experiencing a renaissance and now finds itself right at the centre of brand safety and agency concerns, as John Snyder, CEO and co-founder, Grapeshot, explains. Speaking exclusively with ExchangeWire, Snyder highlights why context in audience targeting needs to get back in the driver's seat.

ExchangeWire: What has the evolution of the contextual targeting market been, and what does Grapeshot do for brands and agencies?

John Snyder: Context has always been the mother of all advertising strategies. Think of the first print ads you ever saw, and the realisation that Vogue magazine had very different types of ads compared to USA Today. Advertisers choose the context of the content consumers are reading as one proxy for the profiles they wanted to reach.

When looking at today’s digital advertising, it becomes clear that the early contextual providers have been pushed aside by a behavioural targeting frenzy. “Let's chase cookied users around the internet and find them anywhere and everywhere – whether on the New York Times or in some side alley of the internet." This seemed to be an efficient use of media spend making it easy to cyberstalk users as they traversed the internet, often with a product they had already purchased days before. Ad blockers were allowed to start their insidious invasion of the advertising industry because context had been seemingly thrown out the window.

Now that big brands are mainstream users of programmatic, the need for safety, control, and transparency is paramount. The audience still needs to be targeted in the right context.

Grapeshot allows agency trading desks to target specific pages based on the combinations of words they positively want to target (e.g. Volkswagen, hybrid car, environment, battery power) as well as specific contexts they need to avoid (e.g. diesel emission scandal, recall) in an easy, flexible, and transparent way.

What measures should agencies be taking to ensure ads are placed in safe context on behalf of the brands they represent?

The reputational risk of misplacing inside unpalatable, toxic page content goes beyond budget waste; it can erode the very brand value built up diligently over many campaigns and years, in an instant. Google lost ad spend this year when they misplaced Verizon and AT&T brands on ISIS terrorist content on their flagship YouTube inventory – even when the company has AdSense keyword capabilities to target both positive and negative contexts! Independent verification from technology vendors like Grapeshot extricates buyers from the need to rely on the media vendor to block ad impressions.

No one can afford to drop the ball with a brand’s reputation – and whilst it is impossible to guarantee 100% protection across the wide vagaries of the global internet, a brand safety SLA really must execute to 99.999% reliability.

This means ensuring your brand safety tools can:

1. Reduce budget wastage by deploying both pre- and post-bid filtering instead of only post-buy ad blocking.
2. Use simple, easy tools that enable the ad ops teams to implement brand safety in an efficient way.
3. Consider third-party verification tools, and do not only rely on large media owners to do their own safety housekeeping.
4. Ask questions on the actual brand safety filtering methodologies deployed, keeping a focus on the ease of use, self-serve capabilities, and reporting/sharing tools.
5. Deploy transparent systems that can explain the logic of targeting and block specific contexts to the brand owners, with end-to-end auditability.
6. Consider the order of magnitude of financial loss from the reputational risk of misplaced brands.

How can agencies better understand the different environments that would suit their brands?

The biggest mistake we made at Grapeshot was, perhaps, to think that programmatic advertising was all about how one targets an ad. If you are chasing users across the internet, the cookie ID becomes a dominant factor for buying an ad. Yet the URL can also unlock huge value, especially when collected as part of the digital trace of what the media budget has spent its money on. Linking clicks, conversions, dwell time, and other downstream campaign metrics to the URL impressions just purchased unlocks untold value to any brand. Which page contexts are driving my highest engagement ratios? Why are conversions performing best for my auto brand in entertainment and food segments; yet not politics or celebrity contexts? How shall I modify my targeting strategy, based on the contextual reporting of campaign analytics?

Importantly, deploying custom segments using a choice of keywords specific to each individual brand underpins the curation of new custom audiences – dividing the URLs and cookie IDs from successful programmatic bids into new audience segments, using the context of every URL impression purchased to colour and characterise the audiences who saw your campaign.

Overlaying campaign metrics of spend, CTR, CPA, hover rate, streaming time, etc, can improve your understanding of those custom contexts and custom audiences that perform best, and provide upper-funnel intelligence for finding new audiences – as well as adding new persona insights against the consumers you already know.

But there is a catch. You need to isolate the impressions that are not human, and not viewable. If you are genuinely going to leverage your digital trace of the cookies and URLs where your ads were actually seen. There is no point bringing 'noise' into your optimisation and custom audience modelling routines: endeavour to work with the subset of purer 'real consumer' impressions.

What have you seen as the main triggers for growth for contextual targeting or context advertising?

The simple answer would be the fuss over Google and their misplacement of brands on the very inventory they sell impressions for media dollars. How safe is safe when the walled gardens taking 80% of media spend cannot show simple, effective, transparent methods of protection that an agency can control or customise?

Yet, I believe the issue is more fundamental. The cookie universe has a glass ceiling where growth has two diverging growth trends. The actual human population is not growing as fast as the widening-out of a multi-device consumer who uses mobile(s), tablet, laptop, and other internet terminals. Secondly, the cookie itself is in retreat, with privacy GDPR regulations and the pivot to broader identity social-graphs, or even Facebook/Google log-ins. On both trajectories, the fundamental case of pricing bids in the programmatic auction process becomes indexed to the bid density for each impression – how many buyers are chasing the same cookie or user ID?

Yet, the URL diversity of the internet has no glass ceiling like the human population. Supply is everywhere, but has many dangerous crevasses of reputation risk and danger if buyers step 'off-piste' from premium content providers.

Media budget efficiency will put pressure on the price of the bid at which media is bought, coupled with the requirement for brand safety and relevance in delivering proper 'duty of care' for each specific brand.

Indeed, in a more cookie-less world of multiple mobile devices and GDPR opt-in regulation, the opportunity to buy inventory based on the single pageview or app context will only get more and more important. I predict regulation and the need to get price-efficient ways to buy contextually relevant impressions across the diversity of the internet will be two economic drivers that make context important again.

'Audience in context', the mantra of traditional print, will shape the post-cookie world of competitive 'identity-graph-enabled'. Context will matter more than ever before, as the choice across 140 billion impressions outstrip the 7.5 billion people on the planet today. Content diversity will only increase faster and faster in all languages of the world.

What opportunities do you see coming from the movement of ad tech into the martech industry? What does the future context marketing look like?

Humans matter to all types of insight. A brand’s website logs, outbound email campaign clicks, and e-commerce site traffic will all contain a degree of 'noise', which needs cleaning out to create the subset of 'pure consumer impressions'. The foundation of all optimisation routines, the creation of custom audience modelling in Data Management Platforms (DMPs), and the planners’ insight isolate those important real consumer impressions. Martech needs that clean signal as much as any other part of our digital economy.

However, there is an even bigger opportunity for ad tech to impact martech. We in ad tech have engineered systems that run at three million QPS (queries per second) – listening to the user IDs and URLs of real-time bid streams. Extracting the trending keywords and creating clusters of consumer personas, based on the content that people read, starts to unlock a real-time consumer listening engine. Harnessing the live contextual engines of the bid stream to website logs, email clicks and e-commerce navigation and suddenly the CMO team has the widest reach of real consumer trends and habits.

Our programmatic engineering is producing the new scaffold for activating real-time insights and activation across the whole consumer experience set. Not just targeted ads, but a wider array of content experiences.

Getting to the live contexts of insight and real-time triggers that modify the activation stacks is the when martech gets real-time too – at the scale of three million personalised decisions per second. Content and context will matter even more, when 'audience in context' combined makes the ultimate differentiator for an all-consuming consumer experience.