X
NEW WEBINAR
Beyond Silos: Data Collaboration Strategies for Marketing Success WATCH NOW

Fact Check Series: Contextual Targeting, Audience Targeting, and “Match Rates”

Despite slick marketing from DMP start-ups, publishers need to understand the differences between contextual and audience targeting solutions and how to effectively match consumers to marketers at scale.

by Adam Solomon, Chief Growth Officer, Lotame

This blog post is the second in a three-part Fact Check Series to help publishers and marketing ecosystem participants cut through the noise, smoke, and spin coming from a select group of DMP start-ups, and facilitate a more direct and transparent conversation on finding innovative and effective ways to connect, enrich, and activate consumer data in a manner that respects consumer privacy and empowers choice.

In Part 1, I detailed the facts on profile identifiers, browser storage mechanisms and privacy regulation. I demonstrated that Lotame does not “depend upon 3rd party cookies,” but rather uses 3rd party cookies, 1st party cookies, as well as browser local storage,  to give our publisher clients the most robust and flexible techniques, in order to store profile identifiers associated with their website visitors in a privacy compliant manner.

In this Part 2,  I focus on the evolution of media targeting techniques and clarify what the term “match rate” really means. In particular, I describe match rates in the context of Contextual Targeting and Audience Targeting.  I demonstrate that when transforming Contextual Targeting activations into Audience Targeting activations, the Lotame platform returns the expected ~100% match rate to Google Ad Manager (DFP).

Finally, in Part 3 I will propose a re-framing of what effective next-generation data management technologies should provide to publishers, with an emphasis on scaled and privacy-enabled connectivity between data sources and platforms. I will demonstrate that as an industry, we have seen at least 3 generations of DMP technologies, and the functionality currently being offered to publishers by start-up DMPs such as Permutive are more akin to DMP 1.0 from 10 years ago. While “DMP 1.0” is a cleaner story to tell when regulation is complex and increasing, regressive technologies won’t bring the industry forward or meet the needs of any but the most basic requirements.

True forward-thinking innovation for publishers will be based upon connecting publisher data to marketer data in meaningful ways across channels, platforms, and devices and providing the means for all parties to analyze, enrich, and activate for commercial benefit — while always respecting consumer privacy and empowering choice.

PART 2

Putting “Match Rates” into context (pun intended)

If you’re a publisher that’s been approached by some of the new DMP start-ups such as Permutive, you’ve probably been presented with a playbook to perform a “match test” that will purportedly show how Permutive is superior to your current DMP (e.g., Lotame, Salesforce) the assertion being that their platform delivers higher “user match rates.”  

These DMP start-ups present publishers with a prompt that asks “How many of your users are invisible for targeting in your DMP?” and sets out to answer that question with the following steps:

  • Create an “Everyone” segment in your current DMP (all users with 1+ pageviews)
  • Run an adserver (i.e., Google Ad Manager/DFP) ad impression query against this “Everyone” segment
  • Run a second ad impressions query without the “Everyone” segment (e.g., untargeted)
  • Divide the “Everyone” segment by the untargeted segment — and according to them — that’s your “Match Rate” of targetable users.

Where to begin? I’ve heard the term “match rate” used in many ways and in many contexts through my 20+ year career in marketing and advertising — including almost 10 years in digital ad products and operations at Viacom, Time Inc. and Hearst. But I must say that this is the first time I’ve seen a company purposely confuse and conflate ad impressions with users, adservers with DMPs, and devise a testing methodology that confuses Contextual Targeting with Audience Targeting and distracts from the unique benefits that DMPs can provide to publishers. 

To bring clarity and truth to the dynamics at play here, we will need to break this down piece by piece. I must state at the outset that Contextual Targeting and Audience Targeting are two distinct approaches for targeting and delivering digital advertising to consumers. They require different technologies and platform configurations, they are valued and activated in different ways, and capabilities in one approach do not necessarily apply to the other. I will demonstrate that DMP start-ups like Permutive are presenting Contextual Targeting solutions as if they are Audience Targeting solutions, and publishers need to better understand the differences in order to make informed decisions for their businesses.

In the beginning there was Contextual Targeting…

It is said that modern advertising began to take shape with the advent of newspapers and magazines in the 16th and 17th centuries. Whether it was newspapers and magazines — or later radio and TV — most advertising was similar in that it relied on the associated content to aggregate an audience with certain attributes (demographics, interests, etc) and then marketers could engage those consumers in the context of those experiences. This is Contextual Targeting — aggregate and target consumers in the context of a content experience. The early days of digital advertising were remarkably similar. Whether it was advertising on dial-up services such as AOL, or the emergence of the World Wide Web, advertising came in the form of “banners” that were contextually aligned with digital content. 

Fast-forward to today, Contextual Targeting is still an incredibly effective and compelling technique for marketers to access relevant groups of consumers in known — and many times premium — digital environments. Technologies have evolved and improved to provide everything from semantic content analysis to transmitting content categories and keywords into programmatic bidding exchanges. Some marketing technology companies have chosen to exclusively focus on Contextual Targeting technologies, and others offer Contextual Targeting technologies among many media solutions to publishers and their marketing partners. 

One of the key benefits to Contextual Targeting is that it does not have direct dependencies on cookies or other identifiers in order to target and deliver advertising. In a market environment with increasing challenges persisting cookies/web identifiers, and expanding privacy regulations, Contextual Targeting continues to be a compelling and important tool in a publishers media selling toolkit. 

There are very important roles that DMPs can play to support publishers and marketers in their Contextual Targeting initiatives. DMPs with strong inter-platform connectivity — such as Lotame — are excellent tools for providing analytics and insights on consumers that visit certain web pages or consume certain publisher content. So for example, if a publisher (and their marketing partners) would like a 360 degree view of the interests and attributes of consumers that read certain content on a publisher’s website, then a connected DMP such as Lotame can provide valuable insights by overlaying 1st, 2nd and/or 3rd party data segments onto such content reader segments. This can help publishers and their marketer partners better understand whether certain contextual media buys will provide the marketer with a material volume of consumers that fit the marketer’s user criteria. It’s important to emphasize that in order for DMPs to effectively layer-on such data for this purpose, there needs to be robust platform connectivity between the DMP and the sources of the data (e.g., other DMPs, CRM platforms, 3rd party data providers).

However, there are also important shortcomings to Contextual Targeting as a stand-alone approach. Today’s modern marketer is data-driven, and utilizes an array of data types and sources to drive their media buying across partners and channels. In particular, marketers are increasingly using their own 1st party data and select 3rd party data sources to help drive their media planning, activation, and analytics initiatives. This is typically accomplished via programmatic tactics that include RTB on open exchanges, private marketplace deals, and programmatic guaranteed. So while pure Contextual Targeting is an important tool in a marketer’s toolkit to access consumers in certain channels and in premium context at scale, there is still a significant need for marketers to more precisely access data-driven consumer audience segments to bring better efficiency and efficacy to their marketing initiatives. That’s where Audience Targeting comes into play.

Audience Targeting and consumer addressability

About 10 years ago, the first generation of DMPs (e.g., Lotame, Krux, Demdex) came on the scene to provide publishers with the ability to create 1st party audience segments (e.g., people who read particular articles), and then activate those audience segments through the publishers’ adserver — most commonly DoubleClick for Publishers (DFP). This provided publishers and marketers with a set of Audience Targeting tools to augment their existing Contextual Targeting tools. One way to think about the differences between Contextual Targeting and Audience Targeting is that Contextual Targeting reaches all consumers in the context of particular content, with no ability to distinguish one consumer from the next, but Audience Targeting can more granularly reach individual consumers irrespective of context. Simply put, Audience Targeting can be thought of as “addressable” consumer targeting.

It’s important to note that in those early Audience Targeting days, the DMP-to-adserver integration technique of choice was to pass audience segment IDs from the DMP to the adserver as “key-values” via in-page Javascript. Those of us involved in ad operations back in the day will recall that there were limitations in the number of characters that you could pass into a DFP tag back then, and as a result, DMP admins and ad ops teams had to constantly rotate which audience segments were “live” at any time so that we didn’t break the ad tags. Ah, memories!

In the ensuing years, the DMP-to-adserver integration technique of choice shifted from in-page key-value exchanges to server-to-server integrations. In order to pass audience segment data server-to-server, DMPs and adservers had to perform a simple ID-sync (sometimes referred to as a “pixel sync”) on the page so that the DMP and adserver could each access their own Profile IDs (PIDs) for those users, and pass those PIDs to the other platforms to help translate one PID to the other. So for example, the DMP might access their PID from a cookie and see that the PID was DMP_123. An adserver such as DFP would access their PID from a cookie and see that their PID was DFP_456. By syncing the PIDs between platform partners, both parties would know that DMP_123 = DFP_456. So when syncing data server-to-server, either platform could transmit data to the other using the other platforms native PIDs. 

This pixel-sync integration technique has a material reliance on 3rd party cookies for both platforms. Without going too far into the technical weeds, if 3rd party cookies are blocked — and only 1st party cookies or local storage are available to store PIDs — then it makes it very challenging to sync PIDs for server-to-server audience segment transfers from DMPs to adservers. Therefore, as a result of Apple Safari ITP and Mozilla Firefox ETP blocking 3rd party cookies, publishers have been reverting to the classic in-page key-value passing of segment IDs from DMPs to adservers such as Google Ad Manager (DFP). DMP start-ups such as Permutive feature the in-page key-value integration method with adservers. As mentioned earlier, in-page key-value passing has been in Lotame’s toolkit for over 10 years and is our preferred technique to sync audience segments associated with PIDs stored in 1st party cookies or local storage. Lotame makes available both the in-page key-value method and the server-to-server pixel sync method depending on a client’s needs and requirements. 

What Permutive doesn’t tell publishers about their “match test” methodology is that if you are currently using a pixel-sync server-to-server method with your current DMP, then it will only return addressable consumers (and media avails) that are associated with persistent PIDs stored in 3rd party cookies or Mobile Ad IDs. Their supposed test is designed to convince publishers that they have a deficiency in their DMP’s capabilities, but instead it highlights that publishers should examine their DMP-to-adserver integration techniques and augment/replace server-to-server integrations with in-page key-value integrations. Established DMPs such as Lotame, Salesforce, and Oracle all feature in-page key-value integration options for publishers. 

It is troubling that DMP start-ups such as Permutive are purposely providing a “match test” methodology to publishers that purports to point out how a publisher’s current DMP is technically deficient, but is really taking advantage of a lack of knowledge around differences in adserver integration techniques. Permutive knows full well that regardless of which DMP you use, the in-page key-value method for an “Everyone” segment will always return more users (and media avails) than the pixel-sync server-to-server method due to the absence of PIDs stored in 1st party cookies and local storage in the latter technique. When using the same in-page key-value integration technique, a full-featured DMP platform such as Lotame will return the same number of “Everyone” users and media impressions as the new breed of contextual targeters. Permutive is taking advantage of this purposeful confusion as their primary lead-gen technique for pitching new clients. At Lotame, we believe that long-term successful relationships with our clients are based on truth, transparency, and collaboration — not parlor tricks.

But wait, there’s more!

Another of the greatest hits coming from DMP entrants is their claim that they offer superior targeting capabilities for “passer-by” traffic versus other DMPs such as Lotame.

Sticking with the theme of parlor tricks, this claim utilizes very clever sleight of hand that publishers need to understand. As described by Permutive, “passer-by” traffic are users that visit a publisher’s site for the first time, consume one pageview of content, and then are never seen again. Permutive claims that their DMP provides incremental revenue opportunities for passer-by traffic using 1st party (publisher) data that is not available to publishers with other DMPs. In fact, in one recent webinar, Permutive claimed that they increased available targetable ad inventory for a publisher by almost 8x (yes, 8 times the baseline!!!). Sounds too good to be true? This metric alone is so fantastical that it should give pause to any publisher as to the veracity of anything else they might hear from such a company.

By way of example, if a user visits a Hockey page on a website, and if this is the first time that this website has ever seen this user, Permutive says they will help the publisher monetize this visit in a novel and improved manner only using the fact that this user visited this one Hockey page. If this sounds like an odd claim to you, it’s because it is an odd claim! A user’s visit to a Hockey page should be passed contextually to the site’s adserver and/or programmatic tags as a Hockey content visitor. What Permutive is saying/doing is that they can take this single page visit, transform a contextual visit to a Hockey page into the creation of an audience segment representing “Hockey users”, and then pass that audience segment into the adserver and/or programmatic tags. To put a finer point on it, Permutive is touting their ability to transform a Contextual Targeting impression into an Audience Targeting impression, but somehow in the process of this targeting alchemy, they say that the publisher’s targetable inventory can increase something on the order of 8x. This is confounding. No new ad avails are created in this process. And no new info about the user has been developed (we knew it was a hockey page when we served it!).

A more likely explanation of what is happening is that by taking this single passer-by moment and transforming the associated Contextual Targeting opportunity into an Audience Targeting opportunity, the publisher is shifting the media buying avail from one platform to another, and perhaps from one marketplace to another (e.g., direct sales to programmatic sales).  Overall targetable inventory for the publisher does not increase due to this alchemy. The buying dynamics and perhaps revenue implications might change, but no new inventory is manufactured. I am not saying that there’s anything wrong with transforming Contextual Targeting inventory into Audience Targeting inventory. There might be very good reasons for publishers and marketers to want to transact media in this manner. However, what I do object to is when DMP start-ups impugn the capabilities of others by saying that they have magic technologies that can conjure up 8x targetable inventory from thin air. 

Also absent from this entire scenario is what happens if the passer-by user who visits the website has 3rd party cookies enabled? Since Chrome has 3rd party cookies enabled and commands a 64% global web browser market share, then what happens in that scenario? Any DMP platform with robust 3rd party cookie connectivity will be able to monetize that passer-by visit in a robust manner using 1st, 2nd or 3rd party data. Just because it’s the first time this user visited a publisher site, it doesn’t mean that it’s the first time that a connected DMP like Lotame, Salesforce, or Oracle has seen this user on the web. Since contextual DMPs such as Permutive tout the fact that they don’t feature 3rd party cookie connectivity, then publishers will be missing-out on the significant revenue opportunities for passer-by users on Chrome (once again, 64% of global web browser market share). You can imagine why these contextual DMPs conveniently leave out that part of the analysis for passer-by traffic.

More about Contextual Targeting to Audience Targeting alchemy

This is where we come full circle on the topic of “match rates.” The DMP start-ups focus their set-up and test methodologies on the specific use case of transforming Contextual Targeting avails into Audience Targeting avails, and one should therefore expect almost 100% coverage rate for an “Everyone” audience. When the Lotame platform is configured in the same technical manner and targeting configuration as Permutive, we will also return a 100% coverage rate in Google Ad Manager. The Lotame platform processes new URL visits in real-time, and passes back any new segmentation to adserver and programmatic ad tags within the same pageview. It does not take Lotame 24 hours or a week to “process” new URL visits — no matter how many times Permutive might purposely repeat that lie. Moreover, this “test” doesn’t even contemplate the more important benefits of a DMP platform, namely the definition and activation of audience data and segments that bring incremental value and capabilities to publishers beyond their contextual offerings (e.g., non-endemics).

A more important question in all of this is whether the Contextual Targeting to Audience Targeting match rate is the most important metric for publishers to focus on? We believe the answer is a resounding “No!” 

At Lotame, we believe that the most important metric for publishers to focus on with their DMP technology investments is ROI. An effective DMP technology provider will provide publishers with compelling ad products/capabilities, improved abilities to respond to marketer RFPs, and the tools and data required to deliver performance to marketers across platforms. In addition, a modern day connected DMP will provide publishers with solutions to assist with audience development, content development and personalization, consumer marketing and subscriptions, and open up new revenue streams such as data licensing.

I will dive deeper into this more expansive view of the role of connected DMP technologies in Part 3 of this Fact Check Series, with an emphasis on the true benefits of multi-platform and multi-channel Audience Targeting capabilities for the benefit of publishers and their marketing partners.

Want to read Part 1 of the Fact Check Series? Click here to read “Cookieless Meets Truthfulness.”

Beyond the Cookie Part 3

 

How are marketers and publishers adjusting their customer acquisition strategies in light of nonstop industry change? We surveyed over 1,400 decision-makers to better understand identity’s role in their cookieless future, what they’re adding and removing from the next-gen tech stack, and where they plan to invest today and in the future. Get the report here.