In an advertising landscape transformed by data privacy and tech giants, contextual intelligence will become the dominant form of targeting and is already making both targeting and brand safety more effective.
30-second summary:
Last week saw a milestone marking the renaissance of contextual targeting when Integral Ad Science bought contextual provider Admantx. Behavioral targeting is giving way to contextual targeting, rebooted for the 2020s.
In an advertising landscape transformed by data privacy and tech giants, contextual intelligence will become the dominant form of targeting and is already making both targeting and brand safety more effective. If you’re an online advertiser, it’s essential to understand what contextual targeting is, why it’s become so important and how you should be using it in your campaigns.
To understand the significance of this acquisition, we need to understand the company. Integral Ad Science was founded in 2009, their main competitor DoubleVerify (DV) was founded in 2008. They both provide an ad verification service allowing advertisers to ensure that the money they paid for advertising, has actually been spent on advertising.
They also measure ‘viewability’: how much of an ad has been seen by a human and for how long. It may seem like a simple task, but in the murky world of programmatic trading, with increasing ad fraud, this service has become essential. Both IAS and DV are now integrated into the main DSPs and advertisers are increasingly demanding their media agencies use them for their campaigns.
Since these two players already have such a wide penetration among DSPs, it makes sense for them to provide related services. For example, both now have their own brand safety capabilities, allowing buyers to ensure that the inventory they buy is ‘safe’ and that their ads won’t, or shouldn’t, appear on pages with inappropriate content.
Back in January, DoubleVerify bought Leiki, a Finnish contextual targeting platform giving them the ability to offer their customers contextual targeting. IAS had been working with Admantx for some time, so buying them outright was the natural next step and put IAS on a level playing field with DV. The other big recent acquisition in contextual was that of Grapeshot for $325m by Oracle in 2018 to become part of their comprehensive marketing cloud.
So why have these companies invested so much in acquiring contextual capabilities?
There have been some fundamental and structural changes that now limit how online advertising can operate. The first is powerful, global data privacy legislation. Over the last 18 months, since GDPR has been introduced, the marketing industry has had to reevaluate its relationship with data. A much higher standard has been set for protecting personal data which threatens behavioral targeting, hitherto the dominant method of targeting online ads.
The GDPR which has global reach, protecting each and every EU citizen wherever they are, began the change. It set the standard and other states are leveling up their data protection. The winds of change are reaching around the globe.
In January next year California will introduce the California Consumer Privacy Act (CCPA), and countries such as Canada, Argentina, South Korea and Japan are introducing their own equivalents. A further game changer could arrive in the guise of the ePrivacy regulation. This new EU law was originally planned to be launched at the same time as the GDPR, but has been beset by lobbying and internal disagreements.
The original text threatened to force browsers to block third party cookies by default. This would effectively kill third party cookies. The specific wording (article 10) has been struck from the draft legislation, but the current Finnish presidency has hinted that this is ‘still open for discussion’, alongside a more purist interpretation of the original purpose of the regulation.
We’ve still no firm date for when the ePrivacy regulation will finally be published but many expect it in 2021. There are other concerns about the legality of cookies, the provenance of cookie data obtained via obscure supply chains, leakage and sensitive content. Because of this many pundits have been speculating about the end of the cookie for at least the last year, and finally it seems that the big corporations, advertisers, publishers and tech companies are taking notice.
At the end of 2018 Apple released Intelligent Tracking Prevention (ITP 2.0) in its Safari browser that by default would block third party cookies. At the beginning of this year Google were rumored to be considering a similar change to block cookies. Actually, in August they announced that they would be creating an ecosystem to support the use of third party cookies in a privacy safe way.
Google’s Privacy Sandbox opens up Chrome for developers to build privacy compliant plug-ins to manage and control cookies within Chrome. Google cited their commitment to ‘not destroying the publishing industry’, in their choice of supporting third party cookies and trying to make them better and more privacy compliant. At the same time announcing that they would be blocking any type of finger-printing, another technology that attempts to identify individuals on the web without cookies.
Google rightly called out that finger-printing collects and exposes personal data. Google has also just announced it will be restricting access to the contextual content of the inventory that buyers see in the Google ad exchange (AdX). All of these signals suggest behavioral targeting will become harder within Chrome, which accounts for 70% of all browsers.
What started off as a few headwinds for the industry is turning into a full blown cookie winter. Advertisers are recognizing that these are far reaching changes to legislation and browsers that billions of consumers use every day. As the full significance sinks in, large agency groups and ad tech players are searching for solutions to future-proof their revenue in a cookie-free world. That brings us back to contextual.
Let’s re-familiarize ourselves with the principles of contextual targeting. The basic idea is simple and goes back to print advertising. If a publisher has a page of editorial content about a new family SUV car, then they would try and sell that inventory to a car brand who makes family SUVs. But they might also sell ads which are contextually relevant to the editorial content.
For example, we might find an advertiser that wants to target high income families who enjoy outdoor sports. Maybe they’re advertising a holiday, a mortgage or even a school. Decisions about which ads to place on a page are made based upon the relevance of the ad to the content on the page and who might be reading the page.
As behavioral targeting enters its autumn, contextual targeting offers advertisers the opportunity to target accurately whilst still being privacy compliant. The first step is to understand what a web page is about. This may sound easy, but even for a human it would be a Herculean task to read and categorize the billions of web pages available. Grapeshot which sold to Oracle for $325m last year offers a simple page categorization service based on processing keywords. Ads can then be placed on pages that have the relevant category.
The new generation of contextual companies including Smesh, use cutting edge computer science to more accurately and semantically analyse page content. Natural language processing provides the ability to extract units of meaning (entities) from sentences and paragraphs. At Smesh we then use machine learning to understand how these entities are related to other entities, processing millions of pages at scale, building a powerful knowledge graph. So rather than just analyzing the superficial keywords, we can now get an accurate sense of the meaning of the page. The more accurate the understanding of page content, the more accurately this can be mapped to relevant ads.
The second step is to target ads to specific pages which have a high relevancy score. As individual pages are analysed, they can be scored into industry categories (e.g. IAB) or custom categories specific to audiences. Advertisers can now choose to place ads on pages that score over a certain threshold for these categories and those highly relevant pages can be added to lists of target pages. Media buyers can activate this data to identify look-a-like pages, pages with related entities, keyword lists or to bring abstract audience segments to life.
Contextual intelligence also has a role to play in brand safety. At the moment, publishers are frustrated to see large swathes of inventory being incorrectly classified as unsafe. This results from the superficial analysis of keywords. Using machine learning powered contextual analysis, publishers can get a more accurate picture of page content and are able to identify inaccurate classification, releasing inventory and increasing yield.
Not only that, but contextual intelligence allows a spectrum of brand safety classes, rather than the current binary safe/ unsafe classification. Advertisers can choose the level of safety and suitability that they are comfortable with, and then choose pages that score above that threshold.
So strategically, this explains why DV, IAS and Oracle wanted to buy contextual partners. They recognize that contextual is fundamental to effective audience targeting in a privacy safe world and they are investing now to ensure that that they have the future capability to continue to serve their customers. They are already deeply integrated into DSPs for ad verification and brand safety, so it’s easy to add contextual capability and enhanced brand safety and make it available to customers at the click of a mouse.
The acquisition of Admantx is the beginning not the end. In the coming year, we’ll see more online spend using contextual targeting. It’s worthwhile marketers taking the time to understand how this works so they can hold their agencies to account, making sure their ad budgets are being spent effectively, and ensuring compliance with the ever evolving data privacy legislation. This deal marks the start of the new contextual era, expect to hear a lot more about this technology in 2020.
Tim Flagg is a commercial lead at Smesh, Contextual intelligence powered by machine learning.
