The Rise of Curation

Kraftisms #4 - Curators are using sell-side technologies to enrich publisher inventory and provide a real service to buyers.

Before we get to this week’s issue…

I’ve been sitting here for three weeks staring at this draft… and not finishing it. I’ve had all sorts of warring emotions and thoughts to feed my writer’s block:

• There’s so much going on in market; should I talk about that instead? But I’m not a journalist.

• How many “explainer” newsletters do people in this industry really need?

• Weekly? How do Ari Paparo and Rishad Tobaccowala come up with such brilliant topics each week? How did Brian Wieser post 12 great updates last week?

That said, I’ve now held over a hundred Listening Tour meetings — and have a lot of thoughts on this industry, and business in general, informed by that process. I want to share those thoughts with you.

So I’m getting back on track — but I’m going to write about my own passions: more than just adtech. In addition to industry perspectives, I’ll continue to talk about the lessons I’ve learned and the stories that made me who I am. Let’s jump back in:

Quote of the Week:

“Everyone is a genius. But if you judge a fish by its ability to climb a tree, it will live its whole life believing that it is stupid.” —Albert Einstein

Kraftism of the Week:
The Dumbest Person in the Room

I wanted to be an astrophysicist.

At the age of four (after seeing Star Wars at the Ziegfeld Theatre in New York), I was captivated by the concept of outer space. At seven, Carl Sagan’s Cosmos series on PBS cemented my path. I knew exactly what I wanted to be and how to get there. My dream was to study astrophysics at MIT, learning from legends like Alan Guth (creator of the inflationary big bang theory).

So I studied. I worked hard. And I made it. I got into MIT, where I double majored in Physics and Theatre, taking a class with Professor Guth. I was ready to learn the secrets of the universe — and the theatre and film skills to create the next Cosmos someday. As I had always planned. I was ready. I was excited. I was… a straight-D student.

Someone has to be the dumbest of the people smart enough to get into MIT.

That would be me.

I stood no chance of competing with the 15-year-old freshmen, the triple majors, the 160+ IQ geniuses. I was surrounded by Sheldons, Leonards, Howards, Rajs, Bernadettes, and Amys. But I brought other skills to the table, especially one that was super-rare in the MIT environment: I could communicate with other human beings.

Classmates wanted me on their lab teams despite my GPA. Why? I could understand the science, and, more importantly, I could communicate that science. They did the lab. I aced the presentation. So when I dropped out of college to lead the nascent web development consultancy efforts at Sapient at the birth of “the Web,” I took with me one of the most important lessons in my life:

I actively seek rooms in which I am the dumbest one. Because if everyone there is smarter than me… it’s a good room! I’ll bring other skills to the table.

Applying the Kraftism:
The Evolution of Curation

Just as I had that experience of being “other,” while bringing my own talents to the table, curators are “other,” as well — not a publisher nor an advertiser — finding their own way to add value. Curators understand publishers, bringing unique access to data and inventory together to communicate that value to buyers in ways publishers find particularly daunting when attempting to do so on their own.

Curators (a term made popular by Audigent’s Greg Williams) are the companies that use data (be that audience data, contextual data, modeled audiences, performance analysis, or all of the above) to add value to inventory across multiple publishers. They are the confluence of the evolution of ad networks, DMPs, and data providers who have figured out how to efficiently use existing pipes rather than (like the ad networks of old) adding new intermediary hops.

The Rationale Behind Curation

Programmatic buyers want to make sure they are bidding only on the specific inventory and users that meet their criteria. Curation exists primarily due to three factors that have limited the ability of these buyers to meet that need solely through DSPs: limitations in information passed in the OpenRTB bid stream, analysis granularity, and how most DSPs handle the “OR” clause.

• OpenRTB bidstream limitations. Publishers understand their inventory at a very detailed level. Factors such as where on the page an ad sits, how the reader has travelled between pages, what other ads are being delivered into the environment, refresh rates, and more. These attributes can be passed to SSPs or curators in a variety of ways — but are not further-passed via OpenRTB to the buy side. This is also why the curation platforms are primarily created by SSPs or other platforms that have code on the publishers’ pages.

• Analysis granularity. Most early curation attempts, be they by ad networks, SSPs, or even DSPs (e.g. TheTradeDesk’s “Sellers and Publishers 500”) is a list of domains: a set of websites on which the ads might run. Today, buying is still primarily done by the DSPs on a domain basis — with a handful of enrichment companies such as Peer39 or IRIS.tv (on the video side) providing contextual signals. But curation can happen not only at the individual URL level, but in some cases (e.g. Scope3’s GMP+) at the placement level. Companies like Sincera are also providing ways to curate with more and more granular signals.

• DSPs and the “OR” clause. Data companies have either become curators or arm them with data… in large part because of the failure of most DSPs to effectively give the data companies credit. Many campaigns are trafficked with OR clauses — “find users who are in data segment A OR in data segment B”. Due to how their logging systems work, DSPs tend to record that the campaign delivered, but not which of segment A or B triggered the campaign. Which means… they have to either guess (random) which one triggered, pick the first one (A gets all the credit), double count (A and B both get full credit) or split the credit (A and B each get half credit). Data companies hate this, as they know they aren’t getting paid for all the inventory being identified… but they don’t have this problem on the sell-side, where the triggering data set can be logged for monetization purposes.

Enter the Curators

These factors have led to the rise not only of curators (gathering multi-publisher curated inventory into PMPs), but also the evolution of that partnership into SSP-powered curation platforms. Why SSPs? SSPs have code on the publishers’ pages (usually via Prebid adapters). By having that direct access, they are not limited to signals that fit into OpenRTB parameters, so can get a much richer level of information to use in decision-making. This is also why Audigent, as an example, seeks to put its Hadron ID code on as many publishers as possible — it gives a real advantage to have data at a granular level through direct access to the publisher’s page.

This isn’t easy… adding value on top of publisher inventory that you don’t own and getting advertisers interested in buying through an intermediary just “because it works” requires intensive skill. As Harsh Jiandani (who was a leader on the team that commercialized Xandr Curate), Chief Commercial Officer at Koddi likes to say, “Scaling a curated PMP is a dark art.”

One note: Curators often talk about one of the benefits they bring being the ability to optimize campaigns… but it’s a bit of a slight misnomer, even though true. Curators optimize the inventory pools with data or other signals so that the buyer knows that every piece of inventory matches their desired criteria. But the buyer is still using a DSP to optimize to its own success metric. The long and short of it: Curators (via the SSPs they use as platforms) optimize inventory (often based on feedback from the buyers) while DSPs optimize results from that inventory — both being valuable services.

The Rise of Curation Platforms

Some say the first curators were publishers selling portions of their inventory through insertion orders and (later) PMPs. I call this “packaging” instead of “curation” as, other than during a brief surge in the popularity of reach extension prior to programmatic’s heyday, publishers were just selling their own inventory.

The history of curation, however, is really a history of gathering inventory from across the ecosystem.

• Ad Networks: The first curators were the early ad networks (e.g. Collective Media. Ad.com. Specific Media), creating “curated site lists” for categories like Auto or Travel that they sold to buyers. The curation was at a domain-level of granularity — a list of sites on which buyers’ ads could run.

• SSP Packages: Some call it Auction Packages, others multi-publisher PMPs, but at some point, the SSPs began to sell cross-publisher media to buyers in order to bring differentiated demand (and thus differentiated value) to publishers. These PMPs had more granularity than just lists of domains, and was really the beginning of systematic intelligence determining which inventory to show to buyers for specific purposes.

• Adding Audience: The data companies begin to integrate with the sell side. In most cases this was done asynchronously, often onboarded through intermediaries such as LiveRamp.

• Real-time Data: Years before curation was a thing, the AppNexus SSP had a 10ms “hop” prior to sending out calls to bidders. Companies that co-located servers in the AppNexus data centers could enrich impressions in real time (if they could respond fast enough) — adding data that didn’t make sense asynchronously (e.g. Weather data based on where a user was at that particular time). Most curation platforms don’t yet support real-time data (in part due to the cost of the infrastructure), but are building it to support a handful of the more tech-savvy curators.

• Financial Management. What’s been missing until recently has been the ability to manage payments across the value chain. That is the core benefit of the curation platform (vs. the historic manually-created PMPs: they allow a curator to mark up (rather than keep the price the same and take a cut as ad networks often do) inventory for sale through a PMP based on data or other attributes, conduct the auction so that the net payment to the publisher (the amount paid by the buyer minus the curator’s markup minus the SSP’s fee) is what competes in the Prebid auction, and then afterwards pay out everyone directly.

Curation Platforms are Now Table Stakes for SSPs

Today, almost every SSP has (or is building) a curation platform to capitalize on the advantages that curation brings - and that the sell-side systems have in empowering curation.

Some provide access only to their publishers to do reach extension, some offer it as a managed service, but most are allowing self-serve by curators of all sorts. Each platform, as well, is looking for their differentiator… from the identity spine of OpenX to the DMP-powered audiences at TripleLift to outcome-based optimization at Media.net to real-time data at Index to the established marketplaces and feedback loops at Pubmatic and Xandr/AppNexus/Microsoft . Some platforms even let publishers “package” their inventory into PMPs to fine tune the set of inventory for curators to consider in their curation-PMPs (what has become known as “PMP Daisychaining”) to help further optimize the inventory.

There is some question still about fees charged on the curation markup itself. SSPs have seen curators as “demand generation” partners and are leery of doing anything to disincentivize unique demand to bring to publishers. That said, SSPs (with a few exceptions) also charge the curator a portion of the curator’s markup as a fee. While this makes sense to do, the mechanisms aren’t as transparent as has become industry norm.

So Should Publishers Participate?

I’ve spent nearly 22 years supporting publishers… and 5 years being one. I recognize that publishers’ biggest fear is losing control of the value proposition of their inventory, be that via data leakage or reselling by third parties. Although a handful of SSPs tried to get publishers to “opt in” before realizing that these fears were being triggered and moved rather quickly to an “opt out” mechanism (with none of the curation platforms seeing any meaningful opting out by publishers), some publishers are still concerned by this growing trend. They liken it to the rise of ad networks (because in effect, curators are the new ad network). But curation is different in three key ways:

• Because curators mark up the price of the inventory (in return for doing the curation for buyers), there is no devaluation of the market value of a publisher’s supply, something that has typically occurred with ad network partnerships.

• Because the curation platforms pass on to the auction the net payment to the publisher (minus any markup or SSP fees), the curator only wins the impression when it generates the highest net yield to the publisher.

• Because the publishers can further package what inventory the curators get to choose between, they can participate in ways that actively raise the value of their own inventory.

So the long and short… should publishers participate? Yes.

Curation has had a long journey to get to where it is today. The first steps were clear and straightforward, but just as my path to astrophysics wasn’t the right one, and I had to pivot and evolve… so did the curation market and the platforms that empower it. The curators have found the right way to add value -- and the curation platforms continue to evolve to empower that.

Andrew

(Note: The graphic above was made utilizing GPT Plus - One of the things I’m learning in this Tour is the value of when to use AI, when not to use AI, and the importance of calling out when I do.)