The digital advertising world is dominated by a few large players. Facebook, Google, etc. You could probably name them offhand even if you barely knew the industry.
These huge, established players are known as walled gardens. They’re called this because they jealously guard all their user data behind a (metaphorical) wall. So only Google and Facebook get to frolic in their Eden-like gardens. The rest have had to learn to live outside the garden.
In order to thrive outside these digital oases, the exiles have had to be inventive and nimble on their feet. Problems are constantly popping up, and solutions aren’t always forthcoming.
One of the primary tools that those outside the walled garden have been using to stay competitive is known as cookie syncing.
Cookie syncing itself is a relatively complicated and inefficient system. In fact, it’s so complicated and inefficient that several companies are trying to replace entirely.
The replacement systems that they are proposing are called universal IDs.
But first, let’s look at why cookie syncing is necessary and why a universal ID would be better.
The current ecosystem
The current programmatic buying ecosystem is a complicated beast. It’s an enormous web of connections that link many disparate and ever-changing parts.
Before you can understand the value of a universal ID, you need to understand the system in which it will exist.
Right now the system works something like this:
First, a user arrives on a page. The minute he arrives a header script contacts the publisher’s Ad Server to load any ads it might have. Any unsold inventory (i.e., ad space that is still available) is identified.
Then, to sell the remaining inventory, the header sends a request to one or several Supply-Side Platforms (SSPs). To keep this simple, we’ll use one.
Via an Ad Exchange, the SSP will send out a bid request for the available inventory.
Demand-Side Platforms (DSPs), Ad Networks, and even other Ad Exchanges will place bids on this inventory.
The winning bidder then forwards the ad information back down the chain to the publisher.
This outline is just a straightforward, high-level overview of the programmatic advertising ecosystem. But it should give you an idea of the large number of moving parts involved.
The entire system needs to function in a fraction of a second for each impression, with millions of impressions occurring per minute.
The importance of data
Pretty much every single platform at every single step of the programmatic buy process is going to be collecting as much data as possible. This is an industry that runs on data, after all.
The main point of interest is the user that is behind the impression. The better the user can be understood, and the more that advertisers know about him, the better he can be targeted.
Many different stages of the buying process, from the publisher himself to DSPs, make use of Data Management Platforms (DMPs). These are platforms that help other platforms and publishers manage their different kinds of data from a variety of providers.
DMPs play a vital role in the identification of users. Using the different information gathered or bought, they help DSP identify users and use this information to make a bid.
Bids depend on user identification
User information is the foundation of the modern programmatic system. In fact, a DSP generally will not even place a bid on inventory if it cannot identify the user behind the impression.
This is because without identifying the user, the DSP cannot determine the value that this particular user brings to its campaigns. Without this knowledge, it cannot place an informed bid.
This is because user tracking and user identification are at the very core of programmatic advertising. Without understanding who the user behind an impression is, then the whole system ceases to work.
The problem is that part of the process is keeping track of user data on their own. Cookies are generally the tool utilized to do this tracking, and the data gathered is shared up and down the chain.
The primary issue here is that all of these different moving parts are keeping track of the same people on their own. A result of this is that they end up calling the same users by different names.
But why do all of these different platforms and firms use different cookies? The simple answer is that only the issuing domain can read a given cookie, and adtech firms are jealous of their data.
The result here is that each platform has a different cookie – a different name – for the same user. But to effectively target and serve ads, they need to be able to recognize that user.
Since they all have different cookies and different names for each user, this is no small hurdle.
Eventually, the industry as a whole found a solution to the communication problem. It’s a solution
It’s called cookie syncing or cookie matching.
In essence, it works by having each part of the programmatic buying process build a large data table. This table is called a match table.
This way, information can be exchanged between the different parts of the system.
To keep these tables updated, the different platforms continuously sync with one another, updating their tables over and over. There could, in theory, be over one hundred names for a single user in a DSP’s match table. And with this information, the DSP will be able to communicate with an equal number of platforms effectively.
It’s far from perfect
However, there are plenty of ways that this system could break down and lead to a platform not recognizing a user. They might not sync with a particular platform; there might be errors in their table, etc. In short, there is a lot of data degradation that occurs when so many different organizations have to keep enormous lists of names for a single user.
Vocab: Match Rate – The percentage of cookies that two different platforms manage to sync with one another. Anything over 60% is considered good.
The current system is far from perfect. It’s been a good stopgap measure. However, if smaller players are to compete with giants like Facebook and Google, a new solution is in order.
Right now, breakdowns in this system cause a myriad of problems for the various programmatic platforms. DSPs, for example, often can’t identify up to 40% of users in bid requests. This lowers their auction participation rate and, therefore, their potential revenue (not to mention lost potential conversions for their customers).
The system can be broken down by things as simple as clearing cookies or banning 3rd-party cookies in browser settings.
And if they can’t identify the user, then a bid won’t be placed.
Further, Frequency Capping is hugely complicated when users can’t be identified with certainty. If a user can’t be identified, how can you be sure how many times he has been served a given ad?
The big guys don’t have to do this
The industry behemoths like Google and Facebook do not have to deal with this problem. Due to their size and scale, they have both been able to build enormous troves of data on their users. Much of this data is freely given to them from the start when users register. The rest they glean through tracking their actions within the walled garden.
These troves of data allow Google and Facebook to target users with extreme accuracy. The two don’t dominate the online advertising world like they used to. However, between the two of them they’re capturing nearly 48% of the market in 2018, according to eMarketer.
These giants are naturally unwilling to share the goods that allow them to reap in such a large part of an enormous market. And, due to the position that they’ve managed to build up, smaller players have a hard time competing.
The fragmented cookie syncing system compounds this competitive disadvantage.
Cookie syncing leads to several problems
Beyond the low match rates that many different platforms experience, there are a few other problems with cookie syncing as an industry-standard medium of exchange. Of these issues, the most important by far is the latency.
Latency, technically speaking, is “the delay before a transfer of data begins following an instruction for its transfer.” (Oxford dictionary).
It’s a word that you often hear in gaming circles, especially regarding fast-paced online games. But for digital advertisers, there is a wholly different problem to tackle: and that’s page load latency.
In other words, latency is the time it takes to load a page.
When a user arrives on a page, and the programmatic system kicks off, a lot of things are happening at once (as we illustrated earlier with our short walkthrough).
One of these things is the entire cookie-syncing system. Because with every new user, the syncing system has to fire up and connect with everyone with whom they exchange data, latency becomes a problem fast.
This is because there isn’t a central service wherein everyone touches base and exchanges data via a single connection. Instead, it’s a lot of different connections between many different platforms — all of whom checking and writing down what all the others are calling this user.
This massive exchange of data, small though that data may be, can have a very negative effect on page load latency. In fact, according to ID5, over 35% of sites take longer than 10 seconds to complete all their cookie syncs.
Now, oftentimes these calls are occurring in the background after a page has loaded. This is done in an attempt to limit the damage, as it were, caused by the cookie syncing calls. Nevertheless, all of these calls (ID5 puts the average at over 60 per page) can have a very negative effect on load time. Or, in other words, they can massively increase page latency.
This problem is made worse still if the website is using Client-Side Header Bidding rather than the much quicker Server-Side Header Bidding system.
The solution to this is a Universal ID
Programmatic advertisers, however, absolutely need these cookie match tables if they’re to be able to communicate with one another effectively. Without being able to express who, exactly, the user in question is, then there is no way to place bids.
But the cookie syncing system simply isn’t scaling well and doesn’t appear to be a good structure on which to build the future of programmatic advertising.
Thankfully, the industry is coming to this conclusion, and new solutions to this problem are finally under development.
There are a few contenders that believe that the best route is to build a better cookie syncing system. However, most of the proposed solutions right now focus on building a single unified ID system.
The three contenders
So far, there are a few primary contenders who want to offer the universal ID solution. The two leading contenders who are trying to introduce a unified solution are the IAB and TheTradeDesk. These are by no means the only two, however.
IAB – The Interactive Advertising Bureau‘s TechLab division is working on an open identification system known as DigiTrust. The goal is to create an industry-standard identification system for users. A large number of businesses are already involved in DigiTrust.
TheTradeDesk – TheTradeDesk is one of the most prominent players right now in the online advertising space outside of Google/Facebook. As a DSP they have a vested interest in making sure that cookie match rates are as high as possible.
Their universal ID proposal is the Unified ID Solution. TheTradeDesk has already had some remarkable breakthroughs in terms of adoption. Perhaps the most notable being their recent alliance with Prebid.
The other option
Some think it’s better to improve what works rather than replace it entirely.
As stated earlier, the universal ID system is only one proposal for solving the cookie syncing problem. The other solution is to continue cookie matching but to do it in one centralized place.
This is the strategy the ID5 has been using. Right now, ID5 has the most advanced centralized cookie matching mechanism on the market, and it’s attracting big names (they have also integrated with Prebid).
Right now, however, it appears that these are the two most promising paths to a solution to the cookie syncing problem.
Who’s going to come out on top?
So what is the future going to look like? Will it be dominated by a centralized cookie matching or a unified central ID system? Will ID5 or TheTradeDesk come out on top?
Right now it’s hard to tell. But no matter who comes out on top, the future is looking significantly more streamlined, efficient, and profitable.
Programmatic advertising works on the principle of bidding on impressions in real-time. But to place bids on these impressions, the different components of the programmatic ecosystem need to know which user is behind the impression. Otherwise, it is difficult to come up with a bid price.
Today, cookies are the primary means of user tracking. And this is precisely where the problem arises.
Cookies can only be read by the domain that placed them on the users’ browser. This limitation means that all of the different programmatic companies that make up the ecosystem refer to the same user by different names. For the system to work, the different parts of the programmatic ecosystem must continuously exchange information and “sync” their cookies with other platforms. In this way, they create giant “match tables” that contain the different names all the different platforms have for the same user.
This process is massively inefficient and causes both latency and data loss problems that have adverse effects on conversion. Several different consortiums and businesses are proposing a unified user ID solution that will eliminate the need for cookie syncing entirely.
Mobinner is a High-Performance Demand-Side Platform. Since 2017 we’ve been helping clients acquire users, build brands, and drive conversions. See how Mobinner can jumpstart your advertising today.