What is a First-Party DMP? The Definitive Guide (2020) | Adzerk

What is a First-Party DMP? The Definitive Guide (2020)

Chris Shuptrine
cardimage

With the rise of privacy laws (like GDPR, CCPA) and the current/upcoming deprecation of third-party cookies (by Safari, Chrome), publishers are realizing that their monetization futures cannot rely on programmatic ads and third-party data.

In order to remedy this, many have turned to their first-party data, through which they can create valuable user segments that advertisers can target at premium prices.

By making their first-party data targetable, publishers can own their revenue destinies and move away from a reliance on programmatic ad exchanges, which are providing lower eCPMs than they once did, due to the aforementioned industry changes.

However, to accomplish this goal of first-party data activation, publishers need a sell-side, first-party Data Management Platform (DMP).

Such first-party DMPs for publishers share the following characteristics:

  1. They store first-party data the publisher has collected on their users. Examples include demographic data (age, gender), behavioral (“Active Users”, “Shoe Lovers”), and other unique targeting information - like where they work, what their current weather is, and so on
  2. They tie this data to a Persistent ID, like a hashed log-in username or randomized cookie ID
  3. They allow publishers to build segments based on this data and assign users to those groupings
  4. These segments are connected to the publisher’s ad server and can be referenced at time of ad request - enabling those audiences to be specifically targeted by advertisers, when the user is back on the publisher's site/app

This article looks into what a sell-side, first-party DMP is and how it’s similar to, but apart from, third-party DMPs, buy-side DMPs, and Customer Data Platforms (CDPs). We'll also dive into the top 5 first-party DMPs.

Table of Contents:

1. What is a first-party DMP?
2. Buy-side vs sell-side DMPs
3. First-party vs third-party DMPs
4. Why first-party data is valuable
5. The 5 best first-party DMPs

What is a first-party DMP for publishers?

A first-party Data Management Platform for publishers is a database used for storing, segmenting, and targeting specific user segments when they are back on a publisher’s digital properties - including websites, apps, and digital-out-of-home devices.

Their purpose is for publishers to drive new revenue by:

  1. Offering high-value targeting options that advertisers will pay a premium for
  2. Enticing new advertisers who are drawn to the unique targeting capabilities
  3. Improving the relevance of the content the user sees - thereby helping with user experience and ad click-through-rates

First-party data refers to information collected about users based on how they’ve interacted with the publisher - such as website browsing behavior, registration forms, what interests they’ve indicated, point-of-sale data, CRM info, and the like.

lotame dmp
lotame dmp

(Source: Lotame)

This data would be stored in a first-party DMP tied to an anonymized ID - such as a hash of the user’s log-in name:

User 8743b5206: { 
"Age": 35,
"Gender": "Female"
"Interests": "Music,Running"
"Lifetime Purchase Amount": 500 }

The publisher, via the first-party DMP, would then create segments off of this data, compiled of all users who fit those filters.

They’d work with advertisers directly (or DSPs) to set up direct-sold ad campaigns whereby buyers could target those specific audiences when those users are back on the publisher's site/app.

salesforce dmp
salesforce dmp

(Source: Salesforce)

For instance, let's imagine an e-retailer called Shoes Emporium. They work directly with advertisers to display native ads within their site.

One of their customers is Nike, who is given the below advertising options:

  • Site-Wide Targeting - $1.00 cost per thousand impressions (CPM)
  • “Active Runners” - $5.00 CPMs
  • “Active Runners Who Have Purchased Before” - $10.00 CPMs

To Nike, site-wide targeting may be interesting but not effective at driving new sales. Meanwhile, even though it's 10x the price, targeting the third segment may drive better return on investment.

This is a win-win-win. Shoes Emporium increases their revenue with premium prices; Nike garners better acquisition costs; and the user sees more relevant ads.

To make this data actionable, the first-party DMP has to be connected to the publisher’s ad server in real-time, in order to incorporate the user-level data into the ad decision engine.

This just sounds like a Customer Data Platform (CDP)?

A first-party DMP for publishers can indeed overlap with a CDP, and many CDPs do have first-party DMP capabilities.

customer data platform
customer data platform

(Source: Email Vendor Selection)

A Customer Data Platform is more holistic, though. Similar to a first-party DMP, a CDP ingests user-level information and makes it actionable, but differences include:

  1. A CDP will store more user-level data than a standalone first-party DMP. There’s no reason for a DMP to house credit card information, for instance, since an advertiser isn’t going to target specific CC numbers. Instead, more segmentable, aggregated data like gender, interest groups, etc will be stored in the DMP.
  2. A CDP will process data from disparate data sources and make sense of it through unification. The purpose of a DMP is not to unify and deduplicate; that simplified data is instead pushed to a DMP after such work has been done.
  3. Historically, CDPs have been used by enterprise marketers, who are looking to have a 360 view of their customers’ journeys, gain insights about their users, segment audiences for advertising purposes, and optimize their direct marketing through emails, content recommendations, and mailings. A sell-side, first-party DMP is focused solely on segmenting audiences for monetization purposes.

We expect more CDPs to adopt first-party DMP capabilities by integrating with ad servers, but few CDPs currently have such functionality.

In that world, most CDPs will have DMP arms, but DMPs will not be synonymous with CDPs.

I thought DMPs were just for media-buying?

You’re right that DMPs have predominantly been used by advertisers/DSPs to buy targeted inventory across thousands of publishers. A publisher could work with these DMPs to buy/target third-party segments like “High Household Income” and show those users tailored creatives.

The main goal of using a media-buying DMP is to drive down acquisition costs for new users.

Without any targeting, a brand may see $100 cost-per-first-purchase from their ad spend, but by targeting, say, “High Household Income” users, perhaps it drops to $50.

This is also referred to as a “buy-side DMP” - which is why we want to separate out “sell-side DMP” / “DMP for publishers” from that use case, as it’s important publishers understand that they can use DMPs too.

I thought DMPs were just for third-party data?

Third-party data is information that a brand doesn’t collect themselves - but instead is compiled by a third-party vendor, who is aggregating it from a variety of sources.

While the veracity of such data is often questioned, it’s nonetheless used by advertisers to augment their marketing efforts.

While technically a third-party-data DMP is the technology used to store and make this third-party data actionable (while the buyers/sellers of that data are called “data providers”), the two overlap to the point where most data providers also have DMP offerings and vice versa.

salesforce dmp
salesforce dmp

(Source: DigitalKites)

This data is collected via a combination of sources - usually through buying it from apps/websites directly or paying for their tracking pixel to be dropped by a third-party. For instance, an iOS weather app may be collecting information tied to the Apple IDFA and selling that to a data provider.

This data is often tied to IP Address, mobile IDs, or cookie IDs, and advertisers generally don’t have access to the raw data. Since advertisers have limited ability to audit the data, they can’t independently validate its accuracy.

As mentioned above, third-party-data DMPs are mainly used by advertisers/DSPs to drive down acquisition costs and reach the right audience. They aren’t really used by publishers for monetization.

It’s important to note that “DMP” is not synonymous with third-party data, even though many people do equate the two. This is why I’m using the term “first-party DMP” to separate the two use cases.

These third-party DMPs will increasingly be made irrelevant with privacy laws and third-party cookie removals - as it’ll be hard to compile and sell this third-party data.

Can first-party DMPs for publishers be used for programmatic revenue?

As a whole, first-party DMPs provide little value to publishers focused on OpenRTB programmatic revenue. This is because DSPs are just not equipped to buy custom, single-publisher-based segments at scale via OpenRTB. Such intricate buys involve a 1-to-1 relationship in order to understand pricing and what goes into the segment. This is historically why publishers have not gravitated toward DMPs.

That said, private marketplace deals (PMPs) and programmatic direct can and do employ first-party segments.

Here, DSPs/advertisers would work directly with the publisher and their ad server/SSP to target specific audiences. Such campaigns are effectively direct-sold ad buys enabled through programmatic technology.

Can first-party DMPs incorporate third-party data?

A first-party DMP can store any information tied to a user that the publisher wants. Technically a publisher could buy third-party data and attach it to the user’s row in the DMP, which could then be used for targeting.

The problem here is that third-party data is usually (1) not accessible at the user level to the publisher and (2) is likely not tied to the same persistent ID.

A publisher, for instance, may have a DMP tied to a hashed username. Meanwhile, the data provider may send over data tied to IP Address. While the publisher could tie the username to IP and then the IP to the third-party data, you start to enter a grey territory in terms of accuracy and precision.

And assuming you’re transparent about where the data is coming from, offering segments based on third-party data may not be particularly interesting to advertisers due to accuracy concerns.

Are DMPs used for user-level analysis?

Unlike CDPs, which provide macro trends and insights, DMPs are used for ad targeting and are not analysis tools. The way you view DMP success, therefore, is not through complex queries or Tableau visualizations, but through ad campaign performances.

When viewing an ad report, for example, you can analyze how a site-wide ad campaign compared to the campaign targeting just the “High Spenders” behavioral segment.

Can you provide some use cases for first-party data?

Yup!

Let’s take Facebook, whose ad platform garnered an impressive $70B in 2019. The success of their ad business - besides the scale - is their ability to do micro user-level targeting, all derived from their unique first-party data.

behavioral targeting options for facebook
behavioral targeting options for facebook

This targeting is enabled via an in-house, first-party DMP that collects information tied to each user, such as what’s in their profile, what they’ve shared, what they’ve clicked on, and so on.

This data is then tied to their in-house ad server and made actionable by their self-serve ad product, where advertisers can set up campaigns targeting segments of their choosing.

LinkedIn, as well, has an ad platform that turns first-party data like company, job title, and interests into actionable targeting - which has turned it into the $2 billion revenue driver it is today.

linkedin company targeting ad
linkedin company targeting ad

Why are first-party DMPs more important than ever?

Many disparate events have led to a perfect storm for publishers. Privacy laws, ad blockers, and the upcoming demise of third-party cookies on Chrome all can/will impact publishers’ programmatic revenues. Without the ability to augment their buys with third-party data or do retargeting, advertisers will see worse performance and lower their bids, leading to drops in revenue for publishers.

Due to this, publishers are increasingly looking for new, innovative ways to monetize. Some have turned to native ads that can draw higher rates; some have moved to optimized internal promotions; and others have found ways to monetize their first-party data through audience segments and direct-sold deals (many have done all three).

But to monetize one’s data, a DMP is needed. Without it, there would be no way to store and target one’s users.

Why is first-party so valuable?

First-party data is effective because it’s:

  1. Hard to acquire at scale- Consumers don’t share their data with everyone. If you can get their demographic or interest data from a sign-up form or browsing behavior, you’re well ahead of most publishers. Of course, scale is key too. Your niche app may have demographic data on 5,000 users, but the work to set up a campaign won’t be worth it for advertisers.
  2. Accurate - Your first-party data is based on real-world behaviors and/or user-given information. If a customer freely tells you their age, gender, and favorite pastimes, one can be pretty confident this data is accurate.

    Third-party data, on the other hand, while perhaps directionally accurate, is far from trustworthy. If you’d like to test this yourself, you can use the Oracle/BlueKai DMP’s registry to see what segments they bucket you into (they are one of the largest data providers in the world).

    According to Oracle, I am a 76-year-old man from Denver with a Graduate Degree and teenagers and a veteran in my household. For the record, none of these are true.
  3. Precise - Third-party data relies on approximations based on cookies, IP Addresses, mobile identifiers, and various other anonymized IDs. As compared to first-party data tied to, say, a username, these methods are imprecise (except, perhaps, mobile IDs - but those may be deprecated soon). If a large apartment shares an IP, for example, all five residents may get targeted when only one of them should have. This doesn’t make the IP Address inaccurate, per se, but it’s imprecise in that targeting it may lead to false positives. First-party data tied to a persistent ID like username or a browser cookie, on the other hand, ensures that advertisers are reaching the right person.
  4. Unique - Your data will be unique to your product - giving it extra value. LinkedIn, for example, knows where users work, what their job titles are, where they got degrees, and so on. This data is what powers their successful native ad product, since it’s information that advertisers can target against almost nowhere else.

It’s important to note, though, that just because the data is unique doesn’t mean it has to have value.

Take Dictionary.com; with over 100 million monthly pageviews, they have both scale and unique first-party data (in the form of dictionary searches). Unfortunately for them, this data doesn’t translate into premium value: would a behavioral segment around “People Who Searched For Adverbs” be of interest to advertisers? Probably not.

First-party data also needs to be actionable to be valuable. If you’re able to track user behavior but then can’t tie it to your ad server, it’s useless for monetization.

What data does a first-party DMP usually hold?

The data will be information that the publisher has collected on the user (with consent where applicable), including:

  1. On-Site/In-App Behaviors: Actions that users have taken on the publisher’s digital properties, such as what products they put in their carts, what articles they click on, terms they are searching for, how many times they've purchased, etc.
  2. Real-world Behaviors: Actions that users have taken off-line, such as point-of-sale data like purchases at a brick-and-mortar store.
  3. Demographics: What groups the user falls into; specifically - age, race, religion, gender, family size, ethnicity, income, and education.
  4. Psychographics: Similar to demographics, but focused more on values and interests, such as “introverts”, “mediators”, poker lover”, etc.
  5. Frequency Capping: How many times the user has seen an ad from a given ad campaign / advertiser. Tracking this enables publishers to set caps so users aren’t bombarded by the same ads.
  6. Consent Info: In order to honor consent opt-in and opt-outs, brands need to tie this data to a particular user. DMPs can store this information, which can be used in targeting to exclude them from ad buys.

psychographics and demographics
psychographics and demographics

(Source: CB Insights)

How is the data tied to the user?

The information will be tied to a persistent ID, such as a hashed email/username, mobile identifiers like the IDFA or Google AAID, or a cookie-stored randomized ID.

In order to make a DMP actionable, this ID has to be passed in the ad request at time of page/app load. The ad server would then cross-reference this ID with the DMP. If it sees that User123 is a member of the “Shoe Lovers” audience segment, then that user would be eligible to see an ad targeting “Shoe Lovers”.

How does the DMP get connected to my ad server?

The main purpose of a first-party, sell-side DMP is that it connects to your ad server so user-level information can be factored into the ad decision engine.

For this to occur, one of the following is needed:

  1. Your CDP/DMP vendor needs a pre-built integration with your ad server, which usually involves updating JavaScript code so that the ad code pings the DMP before the request is then sent to the ad server
  2. You ad server has built-in DMP functionality
  3. You've built your own data platform that's connected to your in-house ad server (whether you've build it yourself or with ad serving APIs)

What are the best first-party DMPs for publishers?

Below touches upon the industry leaders in the first-party DMP space. The best option for you will depend on what ad server you use, engineering resources, and whether you want CDP functionality.

The Best Ad Servers

1. Adzerk

Adzerk is the market leader in server-side ad serving, enabled through APIs. With Adzerk, brands can launch custom, fully-bespoke ad servers in a fraction of the time and cost of trying to build it from scratch.

Adzerk's clients include Yelp, Bed Bath and Beyond, Ticketmaster, Edmunds, Mozilla, WeTransfer, and more.

Adzerk’s tools include a built-in first-party Data Management Platform called UserDB. Brands can send their data via APIs and have the ad platform they’ve built reference it at time of ad request.

Capterra: 4.5 stars | G2Crowd: 4.5 stars

adzerk
When to consider using Adzerk
  1. You don't want to slow down your site. Adzerk's DMP is linked to their ad decision engine automatically and optimized for fast ad response times
  2. You want to do server-side ad serving, so you can speed up your site, show native ads, and circumvent ad blockers
  3. You're in-market to build an innovative, custom ad server
When to think twice
  1. You are looking for CDP functionality like data analysis/unification
  2. You are using a third-party ad server, such as Google Ads Manager

2. Google’s Analytics 360

Once referred to as Audience Center 360 but now rolled up into their Analytics 360 product, this Google offering allows publishers to use Google Analytics to build behavioral segments, which can then be targeted as line items in Google Ads Manager and sold against.

Capterra: 4.7 stars | G2Crowd: 4.5 stars

google analytics 360
When to consider them
  1. You are already using Google Ads Manager as your ad server
  2. You are already paying for Analytics 360
  3. You want an easy way to turn Google Analytics browsing data into segments to target
When to think twice
  1. You want to create segments beyond browsing-based behavioral ones. It does not ingest demographic data that users provide via registration forms
  2. You are not using Google Ads Manager already
  3. You want to do private marketplace deals with a DSP, as Google's data can't be exported to ad exchanges/SSPs
  4. You don't already pay for Analytics 360. Pricing starts at $150,000 a year
  5. You want to do server-side ad serving in order to circumvent ad blockers, allow for native ads, and speed up ad response times

3. Blueconic

Blueconic is a major CDP that works with the likes of Hearst, T-Mobile, ING, and more. In addition to unification and marketing activation, they offer integration with Google Ads Manager and Xandr/OpenX for PMP deals.

Capterra: 4.4 stars | G2Crowd: 4.3 stars

blueconic dmp
When to look into them
  1. You're also looking for a Customer Data Platform
  2. You are using Google Ads Manager as your ad server
  3. You want to do private marketplace deals, as they have integrations with multiple SSPs
When to think twice
  1. You don't want to pay extra for CDP functionality
  2. You are using an ad server that's not GAM
  3. You care about page/app load times. Integrating with GAM requires an update to the ad code where the tag pings Blueconic before pinging Google - which increases ad response times
  4. You want to do server-side ad serving

4. Tealium

With $170 million raised and over 500 employees, Tealium is a powerhouse in the CDP realm. Including in this offering is a robust DMP arm.

Capterra: 4.4 stars | G2Crowd: 4.4 stars

tealium dmp
When to look into them
  1. You're also looking for an all-in-one solution for GDPR compliance, tag management, data unification, and audience activation
  2. You are using Google Ads Manager as your ad server
  3. You want to do private marketplace deals, as they have integrations with multiple SSPs
When to think twice
  1. You don't want to pay extra for CDP functionality
  2. You are using an ad server that's not GAM
  3. You can about page/app load times. Integrating with GAM requires an update to the ad code where the tag pings Tealium before pinging Google - which can greatly increase ad response times
  4. You want to do server-side ad serving

5. 1PlusX

1PlusX is a pure-play first-party DMP. Their focus is on turning first-party data into actionable segments - for both marketers and publishers. Because of this, they have integrations with multiple ad servers, including Google Ads Manager, Smart Ad Server, and Adform.

1plusx dmp
When to look into them
  1. You're looking for a pure-play first-party DMP without any other functionality
  2. You want machine learning tools to predict user attributes
When to think twice
  1. You don't want to pay extra for CDP functionality
  2. You are using an ad server that's not GAM
  3. You can about page/app load times. Integrating with them will lead to lower ad load times. This is true for all third-party solutions besides 100% in-house platforms, Google Analytics 360, and Adzerk - where the first-party DMP and ad server are linked in the backend
  4. You want to do server-side ad serving

What are my next steps?

If you're still unsure of what to do next, we are more than happy to set up a 15-min consultation to walk through your needs and discuss the best path forward. We wish you the best in your quest to activate your first-party data.

Meanwhile, if you had any additional questions or thoughts, feel free to join the discussion below!

Join the Ad.Product community

Sign up for our monthly newsletter and to be notified of member-exclusive events and opportunities.

Ad.Product is the first community for product managers, engineers, and others to discover and discuss how to build innovative, user-first ad platforms.

Chris Shuptrine

Recommended Articles