Martin Kihn at Gartner continues a great series of posts describing Adelphic’s patented methodology for cross-device identity matching.

Cross-device identity matching is the way marketers try to map devices and browsers to the same consumer to improve personalization and measurement. in our super-popular Part 1, we got half-way across the bridge by describing one common way this is done. Our case study was a particular patent issued to the mobile demand side platform Adelphic.

 

The post inspired many thoughtful squibs and expansions, including some builds from our friends at Drawbridge, who are well-known for their cross-device identity matching solution. Drawbridge pointed out to me a distinction that might seem obvious but isn’t.

Namely, there are two problems the marketer must solve to perform a successful cross-device identity match:

• (1) Identify a singular device
• (2) Match that device to a person

Of course, you probably can’t do (2) without successfully doing (1). And either or both can be attacked using deterministic or probabilistic methods. It’s possible to use a deterministic (one-to-one) method for (1) and a probabilistic method for (2) or vice versa. Which opens up a wide gate through which parties can march their dogs and ponies.

I have heard vendors say “we use only deterministic methods,” when they were referring only to step (1). A device-matching vendor who doesn’t do any probabilistic modeling at all does not have a model, it has a lookup table — one it likely acquired from someone else who does probabilistic modeling.

There is nothing wrong with probabilities, friends; they are probably inevitable.

All of the major stand-alone third-party matching vendors — Tapad, Drawbridge, Oracle’s Crosswise, Adbrain — use both deterministic and probabilistic methods. Which combined or hybrid approach is exactly what Adelphic describes in its patent.

PROBLEM #1: WHAT DEVICE IS THIS?

Problem (1) may seem simple to solve but is not. Imagine you are a publisher or marketer and a device communicates with your site. You know this device exists — after all, it’s talking to you. But what is it? Would you recognize it if it appeared again? Does it have an identity?

Apple used to communicate a unique device ID with eace server call, but it stopped doing this three years ago (citing privacy concerns). In its place, both Apple and Android created a unique, consumer-controlled ID available to apps selling ads. This is called IDFA and AdID, respectively. So apps can choose to share this ID if they want, but only if (1) the consumer downloaded their app, (2) is using it, (3) has not manually opted out of ad tracking (which they can do with both IDFA and AdID), and (4) the app really sells ads.

So it’s not available to mobile browsers, people the app doesn’t want to share with, opted-out consumers, and non-ad-sellers. In other words, it’s not a cure-all. And it is tied to a device, not a person. The IDFA is different on my iPad Mini and on my iPhone 6.

IDFA and AdId are often called “deterministic” IDs, because if you know them, you know the device. What is the long-suffering marketer to do if she doesn’t have either IDFA or AdID? Give up? Well, if the consumer is in a Chrome browser it can be cookied, of course, but what if he isn’t, or it can’t?

IDENTIFYING A DEVICE WITHOUT AN ID

Here we venture into the territory that used to be called “fingerprinting” and was associated with firms like BlueCava. As we’ve said, this is a form of probabilistic device identification, meaning it ID’s unique devices within a range of probability. Setting aside the right name, let’s describe how it works, again using Adelphic’s patent as our guide.

The patent refers to something it calls a “signature.” This is a combination of attributes that collectively may be used to identify a unique device. These attributes are pieces of information that are shared in the course of routine communications with the app publisher or mobile website owner.

We described some of these attributes last time. They include:

• “system-type” data such as OS version, local time, phone model
• “usage-type” data such as headers, user query parameters, referrer, plug-in data, location, URLs viewed

How is this “signature” created? The patent describes it as a kind of list that contains some or all of the above-mentioned system and usage attributes that have been encoded in such a way that they can quickly compared to similar signatures sitting in the master ID database. The attributes can be encoded as numbers, categories, or even distributions.

Of course, it is not simply a list of all attributes. Much of any data set is noise. It is a selection of those attributes most likely to mean something to the system (the selection process is described below).

So we can think of the “signature” as a streamlined version of the attributes that includes the good ones and not the noise. The patent puts it this way:

“The entity identity is generated by applying to the feature data [i.e. attributes] one or more rules … identifying which of the feature data to use to generate the entity identity …”

LINKING A DEVICE TO A PERSON

All this talk of “entities” brings us to a rather subtle point about cross-device matching. In the Adelphic patent, explicitly, an entity can be a device, a person or a household. So in the same way a probabilistic signature for a device can be created from a weighted subset of attributes, a “signature” can be created for a device — or for people.

Why? It’s simple, really. The attributes we get are the only ones we’ve got. We can use them in a step (1) way to identify an unknown device without a deterministic ID in a sea of devices … OR we can use them in a step (2) way to try to link devices we have already identified as belonging to the same person (or household).

All of these IDs and attribute signatures are going into the master ID database. Even if a deterministic ID is available (like AdID), the database will include a signature based on probabilistic attributes. Why? Think about the app world. What if the person of interest pings you from the same device but a different app, one that doesn’t sell ads or otherwise lacks access to the AdID?

What if they use their browser to hit up your Bernese mountain dog tea cozy shop rather than your amazing BMD app? You’re going to be glad you maintain both deterministic and probabilistic device identifiers.

Now, you have been very patient. Some of you have abandoned the ship and are well within site of the cabana. To the rest, I say: It is time to discuss how to match a device to a person.

First, the system will do (1). It has the device. Next, step (2). Who is it? The system will try to find out if there is any personal determinsitic data available. Data that can be linked to a person include phone number, email, customer ID. Usually, personal deterministic data is known only if the person has a relationship with the app or site owner or has provided it in the session.

The matcher takes the device ID and the deterministic person ID and sees if there is a match within its master ID database. If so, it will look up to see what it knows — e.g., that this person has been flagged by Target as a super-shopper to get massive deals now! or whatever.

if not, the matcher will try to see if it can match the device ID to someone it has in its master ID database some other way. and you all see this one a-coming … yes, it’s …

RECORD LINKAGE!

Say what?!

One of the more enjoyable sections of the adelphic patent is its almost rapturous encomium to a concept called Record Linkage. This is not a term encountered often in the digital marketing literature. I mention it here because it turns out to be a rather well-developed method to do exactly what we are trying to do: take two different sets of attributes and figure out if they actually belong to the same person.

The patent points to “A Theory for Record Linkage,” a seminal paper published in 1969. It started a line of development that’s cropping up here. Record Linkage (RL) encompasses both what we’ve called deterministic matching and probabilistic matching.

RL is described like this:

“[It] takes into account a wider range of potential identifiers, computing weights for each identifier based on its estimated ability to correctly identify a match or a non-match, and using these weights to calculate the probability that two given records refer to the same entity.”

In other words, probabilistic matching as described here has two steps:

• (1) Take all the available attributes and figure out which ones deserve more weight, depending on how well they identify people; and
• (2) Go through the master ID database full of signatures and figure out whether the particular device matches any of them

That’s a lot of “figuring out.” We can be more explicit. Step (1) here is a classic machine learning problem and can be done either on labeled or unlabeled data (i.e., records that we have already matched to people or ones that we haven’t). The preferred method mentioned is to take the master ID database and look at devices that have already been matched using deterministic methods (e.g., by email or phone no.).

The system can then look at all the various attribute data also captured with those devices and run machine learning algorithms to estimate the weights for different attibutes. (If you’re interested, the specific algorithm mentioned is EM, or Expectation Maximization.)

The output of step (1) is a “rank score function,” or a formula that can take the atributes on the unmatched device and bump them up against the device attributes in the master ID database (that is, for already-known devices) and calculate a score. This score is a number from 0-1, with 0 meaning definitely-no-match and 1 meaning oh-yes-match-baby. A higher number means more probably a match. This is step (2).

The process is described:

“… the system computes the distance of each feature against a subset of candidate matching records in the database. A matching rule takes the distances as input and makes a decision if the features are mapped to an existing entity identity in the database.”

Some of you may be wondering what this “distance” is, exactly. It is a calculation that varies depending on the data type. Numerical data can simply be subtracted and normalized. Strings can be compared to see how many characters match (e.g., OS versions). Other types of data like location require special sub-functions to handle. Obviously, features that match perfectly have no “distance” at all.

A SIMPLE EXAMPLE

I’ll leave you today with an example. Let’s say the RL process has been run in the past and the output of the model was a scoring function. And let’s say it determined that the attributes that are the best predictors of two devices belonging to the same person are:

• Location (lat/long)
• Time of Day
• I.P. address

So a device shows up. It has an AdID but there is no match in the system. It passes its attributes and the system turns them into a “signature” and pings them up against the master ID database signatures, calculating a distance and determining a score against each. If there is one that achieves a high enough score to be considered a match, the AdID and new feature information is added to the existing match record.

And there you have it: a probabilistic match.

Then the fun begins.

To view this article in its entirety, visit Gartner.

The post How Cross-Device Identity Matching Works (part 2) appeared first on Adelphic.

This edition of Real-Time Banter features a conversation about Programmatic TV with Joshua Summers, Founder and CEO of clypd. Joshua was interviewed by Jennifer Lum, Chief Strategy Officer for Adelphic.

Jennifer: What is clypd?  

Joshua: clypd is a platform aimed at helping TV media owners get the most value they can out of their television inventory. We are focused on delivering software solutions that enable their sales teams to automate the workflow in delivering advertising solutions, but more importantly, to drive the use of advanced data and science into that process. We are operationalizing data sets that enable them to improve the yield of their media and deliver better value back to the brands that are advertising through them.

Can you share the founding story of clypd with us?

My co-founder Doug Hurd and I have been working together for about nine years now. Our last gig was at a company called WHERE. WHERE was a platform in location-based advertising and services. We built up a pretty decent business and ultimately sold that business to PayPal back in 2011. It was a fun ride and it really taught us about the need to sometimes pivot to find the right market fit and ultimately, to build a business that was able to produce sizable revenue and find an exit. We went onto PayPal and stuck around there for about 18 months while getting our master’s degree in online and mobile payments.

Then one day, we decided that we wanted to do it all over again. So over a couple beers up in Ogunquit, we decided there was an opportunity to leverage some of the advancements that had taken place in the digital advertising world and move them into a market that was ripe for innovation and that had scale beyond anything we’d ever seen (TV), but that also needed technology and data to come in and take it to the next level. That was our birth story: Where could we take some of the understanding we had from our time at WHERE and apply it to the TV marketing opportunity? It took a long time to get from that nugget of a concept to the right product market fit and the right value, but that’s been the fun ride of clypd over the last four years.

What types of companies are leveraging clypd’s platform today?

clypd is pure-play sell-side technology, so we work with the big media owners in television. We focus on linear television, which is the traditional living room experience where you lean back watching live TV or potentially time shifted TV on your television set.

The big media owners really fall under three key categories.

• MVPDs, which are the multichannel video programming distributors made up of the cable operators like Comcast and Cox, satellite companies like Dish and DirecTV, and telcos like AT&T and Verizon Fios.
• The local broadcast and station groups. That’s your local broadcaster in-market, and the station groups are the aggregate of a bunch of local markets under a single holding company.
• The cable networks and national broadcasters. That category includes ESPN, Discovery, Fox Network Group and others. This is where clypd has focused, where we’ve found our beachhead. We focus on national inventory and ultimately delivering value for these types of media owners.

What are the standards that are developing around programmatic TV and are they at all similar to OpenRTB?

When we first started the company, we were very familiar with OpenRTB, coming out of a mobile advertising company. We were hopeful and thought that we could adapt those standards to work in television, which would give us a strong, significant leap forward from where the TV market was. Unfortunately, you can’t take that digital standard and apply it to the linear TV space because everything about linear television is different – the time for delivery, reporting, round-trip delivery of data, the insertion of advertising and the tracking capabilities. Because of all these differences, we needed to start from scratch with redefining how that could work.

We did some early work to develop a spec that we thought might work called “TVonTap.” We released that, working with a number of partners to get to a V1, and we found some early interest. The digital players wanted to participate and be able to help drive it, but it still didn’t have enough industry backing to become a standard. Recently, a group called GABBCON released their own standard called “ABCD,” which, fortunately for us, draws on the TVonTap standard and uses it as a basis for developing a new standard. They’ve got a nice group of participants in the standards body ranging from digital platforms and data platforms, all the way through service providers and into media owners. We are seeing some nice progress there, with the GABBCON ABCD standards, and we’re pretty excited about that.

How has the programmatic TV industry evolved over the past few years? If you think back to when you and Doug were first starting clypd, has your vision remained true? Has the market moved the way you thought it would?

Our vision has remained true in that we believed we could bring value to the media owner, the sell-side of the opportunity, and that we would be pure-play and that we would not rep or arbitrage any inventory. It has also remained true in that those media owners are interested in opportunities to optimize yield and to make more data driven decisions.

Ultimately, I would say the world is vastly different today than it was three or four years ago. When we first started, we found a lot of push back on just the terminology itself. “Programmatic has no place in television.” There were lots of words that we would avoid using in meetings because it would cause different opinions to be raised about challenges.

About 1.5 to 2 years ago, we started to see a pretty impressive shift. People in the industry  started having more conversations, they were more open around the need for automation and driving data into the decision process. Over the last year, we’ve seen the big media owners start to reorganize this and drive innovation directly in the space, bringing on Heads of Data-Driven Advertising, Audience-Based Selling or Programmatic TV.  They’re really starting to think through the concept of “How do you drive value between linear and digital and back, and how you build linear-driven TV campaigns around more advanced audience-based sales?” This has been a very big shift in the last year, and now it’s a very open discussion. Frankly, we are impressed and excited by how much it’s been a part of the Upfront conversation this year, for the first time. I think it’s been a dramatic movement for the industry over the last three to four years.

Five years from now, will the TV industry still be focused on the Upfronts?

Yes, I think the Upfronts are not a phenomenon that is related to the lack of technology, but a phenomenon related to the way the market is structured. In the digital world, supply is a vast, open space. There is almost an infinite level of supply, and demand is the constrained asset. But in the linear TV world, it’s the opposite. The inventory is constrained in its availability. There is only a certain number of TV networks and a certain number of spots available and it is not growing enough to keep up with the demand. Because of that, the demand almost creates a market in itself.

That’s what the Upfronts are – an opportunity to pre-negotiate and buy in-bulk portions of your media spend over the yearto lock it in and feel comfortable that you’re going to be able to use those dollars. The scatter market complements that with buying throughout the remainder of the year, but the Upfronts are, in my opinion, a product of that scarcity, not because of the lack of technology. I think in five years, the Upfronts will not only still be here, but will have all kinds of new technologies and data strategies that support them.

It’s interesting to think about how inventory being sold at the Upfronts may not be bucketed by show, but may be more focused on audience segments.

It’s certainly a direction we see happening. Now I wouldn’t say we believe all dollars will be sold that way, either. Like any good advertising or marketing plan, you want to spread the way that you’re purchasing your media. That means you could be going after different goals: content, breadth of distribution, more reach, specific audiences or specific targets. A well-rounded media plan will still be a well-rounded media plan five years from now.

There’s been a lot of progress with the industry adopting automation, data, and programmatic technologies. What else is needed in order for programmatic TV to reach meaningful scale?

We are at that point where we are seeing the tip of the iceberg. But there are always things that need to happen for adoption to continue to grow. Moving to more industry standards will absolutely help. Industry standards around how data comes into and leaves the system – both from a segmenting and targeting standpoint and from a reporting and verification standpoint. Being able to reach scale is important on both the buy-side and the sell-side. In the national market, we are seeing great progress. I think we’ll continue to see that progress also in the local broadcast, local station group and MVPD markets. Ultimately standards will help ensure that we don’t have a lot of point solutions or one-offs that create fragmentation of experience.

When we get to a point where everything is truly automated – and I would say we have just started to really see the benefits of automation, but we’ve got a long way to go – it opens up the floodgate to really reach that scale. Automation is more than just connecting to buyers and being able to source demand effectively, it’s connecting all the way through to the  delivery systems, traffic and billing systems, into video content delivery systems and ultimately reporting and data systems. There are a lot of disparate technologies using different integration capabilities and different ways of talking to each other and that just takes time for the industry to really connect.

What are some interesting new data sets and measurement solutions that are being applied to programmatic TV?

The TV world is quite different than the digital world. The ability to take a cookie and attach data to that cookie and dress up impressions is not the same. First, we have to talk about two types of data around television. There is currency data and then there is the advanced audience data you can attach to that. The vast majority of everything we see in terms of U.S. currency is Neilsen, which is based off of age and gender.

Once you have that currency set, you have to look at where to go beyond that, and that’s into advanced audiences. There are lots of data sets that have rich, advanced targeting capabilities and are in some ways becoming potentially interesting currencies. Some of the obvious ones that are getting a lot of talk are folks like Rentrak, now part of comScore, and TiVO TRA. There are other data sets that have been fused into primary currencies like a Neilsen, such as MRI Fusion, or NBI, which is credit card data, or Catalina, which is shopper data. clypd takes an agnostic view to the data, on both the currency and on the advanced audience data. We work with whatever our media owner customers and their buying partners want to use. Our job is to apply a gold standard to the way that it is used. clypd’s role is to ensure quality of use in data and how to operationalize it, but not to decide on the currency or the data set itself.

Now the final piece that’s starting to become pretty interesting is where we see digital data sets start to work their way into television. We are at the early days of this, but where it gets exciting is when customers can bring first-party data that they’ve used to deploy campaigns in non-linear digital video and bring those targets into television. I wish that were a straightforward process, but it’s certainly not a process outside our ability to deliver a solution on.

When you think about platforms that are interesting there, you have to think about things like the Oracle data initiatives – the Oracle cloud – and the Neilsen marketing cloud and their eXelate solution, and how they’ve brought data in and ultimately their ability to influence targeting and segmentation through digital data sets into linear TV, which is a very interesting and an emerging opportunity.

Absolutely. They are allowing marketers to get closer to the true cross-device dream.

That’s right.

A final, fun question for you. What was your first mobile phone?

In 1997, I had a Sony Ericsson. I think it was called the D1000. It had a pull up antenna, was about three inches thick by about three inches wide, and probably eight inches long. I would stick it in my pocket and it would stick out so far – it looked absolutely ridiculous. It had a leather wrapper around it. And my first mobile plan was thirty minutes of talk time but I had free incoming first minute so I would have everyone call me and I wouldn’t talk for more than a minute.

You have a great memory.

It was a meaningful phone for me – it was very exciting.
Thank you for speaking with us today Josh!

 

The post Real-Time Banter: Adelphic and clypd Discuss Programmatic TV appeared first on Adelphic.

Adelphic’s VP of Product, Yael Avidan, provides an overview of Adelphic’s new cross-channel solution. 

Understanding consumers’ device usage is beneficial, but understanding consumers’ behaviors across multiple channels, in addition to the devices used, provides much richer data to power campaigns. This is what comprises our proprietary data set, which we refer to as our Behavior Graph™–which powers real-time optimization based on a complete, cross-channel user profile.

Having the Behavior Graph™ integrated into Adelphic’s decisioning layer enables marketers to transition to a real user-centric approach to campaign planning, management, optimization and reporting–and drive superior performance and efficiencies for campaigns targeting consumers on the move.

Cross-channel sits at the heart of Adelphic’s platform.

 

Adelphic’s cross-channel solution is comprised of:

• Adelphic’s Device Graph:  Adelphic’s proprietary device graph leverages non-PII behavioral signals in addition to contextual and location signals in order to identify and link devices.  The device graph includes intra-device as well as cross-device, cross-channel links and makes use of deterministic data when possible, but scales using a probabilistic algorithm.   Over trillion data points serve as starting point and then filtered.  Between 50-100 features are used to identify user pairs. The graph is based on roughly 40% in-app requests and 60% web requests and is continuously updated based on real-time data collection.   Adelphic holds a patent for audience recognition across multiple devices.

• Adelphic’s Behavior Graph™: Behavioral signals, captured in real-time, are mapped back to the device graph to form a richer dataset that joins identity and behavior.  Behavioral signals range from performance (e.g. clicks and conversions) to behavior patterns (e.g. time of day, location, browsing behavior).

• Adelphic’s  ^tag (“a-tag”):  A persistent identifier that overcomes the lack of a standard user identifier for mobile and desktop (and future channels).  Using Adelphic’s Behavior Graph™,  multiple devices and their associated behaviors can be assigned to single ^tag, creating user profiles that incorporate data from multiple channels

• Adelphic’s Predictive Performance Engine:  User behavior and ad performance history across devices and channels are used in real-time to predict the value of a user for a specific ad  (e.g. did this user see this ad on a different channel? Does this user convert more on mobile or desktop?).  The engine leverages cross-device user models as well as contextual models (combined via a combination model) to drive bidding decisions and superior performance.

Our solution enables flexible onboarding of audience segments (1st party, 3rd party, campaign data, pixel-based), forecasting available inventory pools and targeting them across channels.  Adelphic’s Predictive Performance Engine sits at the heart of the solution by leveraging users’ complete profile to drive performance and efficiency.   

Key strategies that are supported via Adelphic’s solution:

• Cross channel targeting across Display, Mobile,  Video and cTV.
• Extending the reach of advertiser’s retargeting audiences
• Extending the reach of lookalike segments
• Extending the reach of behavior-based segments
• Creative sequencing across multiple channels
• Frequency management across channels 

More information:

• How Cross-device identity matching works – Here
• Looking for a Cross-Device Solution? 3 Questions Ad Buyers Should Ask – Here
• Adelphic Launches First Behavior-Centric Cross-Channel Programmatic Ad Solution – Here

Central to our cross-channel solution is, of course, privacy.  Adelphic is fiercely focused on consumer privacy and takes specific steps to ensure that our user profiles are anonymous. Neither we nor the clients who partner with us can glean personally identifiable information from the consumers we engage, even by accident. Adelphic is a member of the Digital Advertising Alliance and complies fully with the DAA’s Self-Regulatory Principles for Online Behavioral Advertising and Self-Regulatory Principles to the Mobile Environment. Adelphic also allows opt out from behavioral tracking through Ghostery.

The post Adelphic Is Proud to Announce a New Approach to Cross-Channel Advertising appeared first on Adelphic.

Adelphic’s Director of Product, Dr. Justin Pniower, lends his perspective on 3 questions media buyers should ask about cross-device advertising.  

Ad buyers have grown accustomed to siloed campaign strategies – separate budgets for mobile, desktop and television. But an audience-based approach executed across the devices of a consumer can be a powerful tool, allowing you to interact with and follow users as they move between devices and channels. But it can be hard, even for experts, to evaluate the quality and scale of these solutions, and industry standards do not yet exist. To determine whether the promise of cross-device solutions can deliver, here are some questions you need to ask.

How is quality measured?

Precision, the percentage of links that are correct with respect to a test dataset, is typically the key metric for assessing the quality of a cross-device solution. But, as is the case with other forms of measurement, values can differ substantially based on the details, like what counts as a correct or incorrect link with respect to the test dataset, or how a link is counted if one device is outside the test dataset.

The test dataset itself also has an effect on the precision measurement. You should inquire about the source and quality of the test data, but you should also inquire about the size of the dataset. By shrinking the size of the test dataset, vendors can inflate precision measurements.

How many connections?

When it comes to cross-device measurement, precision does not indicate anything about scale.  You can have perfect precision with just one link. Generally people consider the central metric for scale to be recall, the percentage of links in the test dataset that the solution was able to identify. But you can still have good recall without good scale, and vice versa. For that reason, you also want to know the total number of connections associated with a given level of precision in order to get a better sense of scale.

But the number of connections alone won’t provide all of the information you need. It’s important to understand the number of each type of connection – intra-device, cross-device and cross-channel – as well.

Not all connections are between different devices. While “device graph” is the common term for a map that links individual consumers to each of their own devices, it’s actually a bit of a misnomer. At the most granular level, the nodes on the device graph are unique device identifiers, not unique devices, and there can be multiple identifiers for a single device. When vendors characterize the size of their graphs, they may be sharing the number of links between device identifiers (intra-device), not the number of links between devices (cross-device). When it comes to your return on ad spend, the difference is important.

A user’s browsing and location history can be very different on a mobile device and a desktop making these links, across channels, among the hardest to establish. In order to identify a cross-channel connection, a robust dataset for each channel is required. As such, the majority of links within a graph are likely between devices in the same channel. Inquire about the percentage of connections that are truly cross-channel as well as their relative precision.

What is the effect on my data and KPIs?

While the counts are important indicators, they do not guarantee scale for your data. First-party data is the ad-buyer’s “holy grail,” so it’s important to know the match rate for identifiers you want to extend using the graph in order to truly optimize your ad spend – and your campaign – across specific consumers’ devices.

But if a cross-device solution cannot improve your KPIs, then other metrics do not matter. Performance KPIs like awareness, engagement and conversion rates work well to measure a cross-device campaign. Solutions that enable sequencing and frequency capping across devices can help bolster these KPIs and reduce a consumer’s urge to activate an ad blocker.

While cross-device technology has been around for a while, we’ve only just begun to activate it. As the number of screens in front of consumers continues to grow, the industry must work quickly to develop standards that empower media buyers to evaluate and leverage this technology to shift to a true audience-based approach. In the meantime , it’s important to be educated on the topic in order to make the best possible buying decisions.

The post Looking for a Cross-Device Solution? 3 Questions Ad Buyers Should Ask appeared first on Adelphic.

Martin Kihn at Gartner provides a great primer and explanation of Adelphic’s patented methodology for cross-device identity matching.

Cross-device identity is the thing that tells marketers and other nosy types that Device A and Device B and Browser C all belong to the same Person X. Devices (and browsers, which we’ll call devices for convenience) are matched to people. These people may be known to us by name or they may remain quite mysterious. But here we are concerned only with the process of matching devices together that all point to someone.

How is this done?

Talking to our friends at Adelphic recently, they clued us in to their U.S. Patent 8,438,184 B1. Adelphic is a mobile demand side platform, in the business of helping advertisers target people mainly on mobile devices, so they’re interested in identity. Why? If they’ve served an ad to Person X on Device A, they’d like to know that fact before they bid good money on yet another ad to the same person on Device B. Or if they happen to know something interesting about Person X (say, they like Bernese mountain dogs), then when that person shows up on a new Device C, Adelphic can put in a high bid for their Bernese mountain dog tea cozy-selling client.

Boom! But only – of course – if they know that Devices A, B and C all belong to the same Bernese mountain dog lover Marty, I mean, Person X.

How? That’s the topic of the patent, which I have read in some detail. (Patent reading is about as fun as going to Coney Island to count the sand, so you’re welcome.) I summarize highlights for the people.

ADELPHIC’S PATENT

It turns out that the “matcher” (e.g., Adelphic) is in the business of building up a large database of individuals. Each individual has a unique identifer (we can call it the Master ID) and a set of known Attributes. These Attributes are things that the database knows about them based on previous interactions. And – importantly – this Master ID and set of Attributes also tries to include other ID’s that are uniquely tied to particular devices.

So at its core, the matcher is a system that is confronted with device identifiers that it bumps up against its Master ID database over and over in an attempt to see if that device belongs with an existing Master ID, i.e., a person they know. If not, a new Master ID is created and becomes part of the matching game.

Some of you may wonder: “Is it using deterministic or probabilistic methods?” The answer is: both.

First, it will try to determine if there is a literal match – that is, if the device identifier is already in the database. If so, there is a direct match that we call deterministic. If not, we need Plan B: probabilistic matching. Plan B uses some fancy math (which I will sketch out below) to figure out if the various Attributes it’s seeing are similar enough to Attributes in the Master ID database to make it PROBABLE we have a match.

If not, rinse and repeat: New Master ID, attached to these Attributes, and associated with the device ID.

This is all somewhat less mystical than it sounds. I will take a moment here to list the specific pieces of information that Adelphic’s patent mentions. It’s interesting because it shows how sparse info can get in ad tech.

Deterministic matches are commonly made by finding one-to-one identity with:

• Cookie – any unique browser cookie set by the advertiser or its agents, like a DSP
• Phone number
• Email address
• UserID – explicit identifier that is similar to a cookie, known only by advertiser or its agents
• Device ID – some of the ones listed in the patent don’t exist anymore (like Apple’s beloved UDID and various open standards that didn’t take); practically, this is limited to ADID and IDFA available only to app developers for Android and IoS

These identifiers are often sent with the HTTP request as a query parameter. This means something that looks like a URL is sent to the matcher’s server and it contains a string after the matcher’s address, like:

www.matcher.com/api/identityrequest?cookie=XYX2883838383aasdf;email=person@site.com

The matcher has set up an API that can be pinged with this request. It sends back a match (if it exists) and other relevant info (ditto). Of course, many times a marketer will not have an email, phone number or even known cookie. In practice, this kind of deterministic identity is useful for people who have explicitly given us information – usually because they are a customer or prospect – and this is stored either in their browser (e.g., cookie) or on the web page itself (e.g., email).

Most people will not have a deterministic identifier attached to their API request. So we go to probabilistic matching. Here the matcher will use any and all information it can find. Some of it is simply the kind of information that is always sent back and forth when machines on the internet communicate. This information is contained in the IAB specs and is routinely part of any exchange of data on the wires.

It comes in two flavors: device and system data; and so-called behavioral data.

Device and system data spelled out by the patent include:

• OS type and version
• Device brand and model
• Clock setting
• Time zone
• Speed
• Language default

These seem quite mundane but contain more information than we think. OS versions can get very specific. While default language = english doesn’t say much, I’ve heard that clock settings down to multiple decimal points can be quite revealing.

Then there is so-called behavioral data. (I say “so-called” because it isn’t always about what we humans call behaviors.)

Attribute data of this type are:

• HTTP header – this is hidden text sent with the HTTP request and includes things like the date and time, characters accepted, various settings
• User agent – the browser and version
• App launch time
• Page load time
• Referrer – previous page visited is usually stored in the browser
• Plug-ins used
• Geography (latititude / longitude)
• URL – specific page person is on; this can also be visited to determine the type of content being viewed (more on this below)
• Typing frequency
• Gesture – these last two apply to apps, if you can capture some patterns in the person’s interaction with the app; I have no idea how often this is used and can’t find any real information about it, other than the obvious fact that people can have different typing frequencies and gestures. (If you know, DM me at @martykihn or comment below.)

This does not seem like a whole lot of information to describe a unique person, and it isn’t. But combined, it can be helpful. For example, if I open a browser and visit a Bernese mountain dog site and my referrer is a Crossfit gym in Bedford Hills, NY – well, that’s pretty unique. If I did the same thing on my iPad last week – which, to be honest, I probably did – there exists a MasterID with those behavioral Attributes and (assuming there are not thousands of people who do exactly the same thing, which there are NOT) the matcher can call us a likely match. Based purely on a URL and a Referrer.

To view the article in its entirety, visit Gartner.

 

The post How Cross-Device Identity Matching Works (part 1) appeared first on Adelphic.