What the End of Salesforce DMP Means

Salesforce announced they are sunsetting Audience Studio, their Data Management Platform (DMP), purchased only five years ago and formerly known as Krux. Is this a hiccup for the DMP category, or a signal of things to come? 

It’s worth taking a look at where DMP is more generally, how it’s been working, and what the future holds for it. 

Here’s the big picture on DMP:

• Reliance on third-party cookies makes the future of the category uncertain

• It offers benefits in audience management but has limitations around speed, data quality, preventing data loss, and interoperability

• Functions best in an all-digital environment, which misses out on a true omnichannel strategy

• Customer Data Platforms (CDP) account for many of the gaps that have become apparent with DMPs

Baked-in DMP limitations

In some ways it’s not surprising that Salesforce decided to sunset their DMP, because the way DMPs function has some critical limitations. The promise of the DMP category sounded similar in many ways to a CDP: create a 360-degree view of customers, find connections and insights, build segments, and optimize experiences to those customers.   

But the DMP customer 360 never came to life and was built on a house of cards — or cookies.    

DMPs collect a large part of their customer data by scraping cookies across the web — meaning most customers are not aware that this is happening or what is being done with their data.  These massive data sets, collected at best with passive, ill-understood consent, are the fuel powering the cookie-based programmatic ecosystem. This is both ethically questionable and no longer reliable with the impending deprecation of third-party cookies.

DMPs can incorporate first-party data but because onboarding products take days or weeks to bring offline data into the platform (for example, from point of sale transactions) there is always a severe lag between customer interaction and media reaction.

This latency challenge highlights the larger interoperability issue: the customer 360 that DMPs offer has always been geared toward the digital world. Brands who want to truly coordinate an omnichannel experience that accounts for both online and in-person interactions find that the DMP 360 can’t interoperate.  

There is also huge data loss in the DMP customer 360 supply chain. Data gets lost when it doesn't match on entering the platform, and then again when it doesn't match going out of the platform into the buying/selling platforms. Brands often find that with a DMP they’re only reaching 30-50% of a potential audience. This forces brands to index toward mega-segments to absorb this loss, greatly limiting the ability to focus on key subsegments and ultimately hobbling personalization.

What DMPs get right

There is definitely value to be had when using a DMP to power a programmatic media strategy.  (Let’s exclude Google, Facebook, and Amazon from our definition of programmatic, since they neither require a DMP nor function fully with one).   Even if this programmatic open marketplace spend is only 20-30% of a brand’s overall media spending, there may still be ROI derived from audience management, frequency management, and suppression strategies that avoid wastage.  

There are also interesting use cases around third-party and second-party data enrichment to create audiences that compliment first-party data sets. But because a brand’s first-party data gets co-mingled with second-party and third-party data, there are huge usage restrictions that prevent mixed data sets from being taken out of the DMP  — meaning marketing and analytics teams are often forced to work within the confines of the DMP for many use cases, instead of in the tools and systems they prefer.  On top of that, there are now alternatives to DMP’s for second- and third-party data activation use cases — more on that later. 

The future for DMP is uncertain

From the point of view of an advertiser, the DMP really only serves to fuel a piece of media optimization for a fraction of media — and that fraction is based on third-party cookies, which are becoming obsolete. 

Could a DMP product evolve into the new post-cookie world?  It’s possible, and we will need to watch and see what comes next for brands like Lotame, Oracle, and Adobe. But it will require a fundamental overhaul of the way DMP technology works, as well as a choice on the part of adtech providers to continue support the remaining DMPs. 

CDP advantages

The right CDP can solve many of the problems the DMP isn’t able to.  

A CDP enables brands to engage with customers in real-time across a true omnichannel ecosystem: across the advertising ecosystem, but also in-store, at check-in, in a loyalty program, on a website, etc. The online and offline worlds can be brought together for a seamless experience.

It's more future-proof, being a way to build a customer 360 without relying on third-party cookies. Brands and consumers gain control over how the data is used. This includes managing anonymous data and displaying explicit digital opt-in to customers on their websites where this collection is occurring, as well as managing privacy-compliant second-party and third-party data.

Insights and analytics in a CDP are more precise, with workflows designed for customer exploration, dashboarding, and predictive analysis, up to the largest data sets and down to individuals (since activation can happen one-to-one in many of the leading tools). Precision insights also mean that teams have data they can trust, not just for marketing activation but for other critical business cases, like financial forecasting and reporting.   Using one platform for customer data management means numbers align across teams, plans, reports, use cases. 

Finally, a CDP powers measurement based around the customer, focusing on behavior and outcomes, so that brands can have a clear view of what’s working and what isn’t to power adaptation and innovation.     

Some things are still hard 

The dream of a totally comprehensive view into the “entire” customer advertising journey still remains elusive. Why? Customers see and engage with ads and content across a variety of platforms; they search on Google and then see a video advertisement on Facebook. The so-called walled gardens of tech — Facebook, Google, Amazon — still aren’t going to let large parts of their customer data co-mingle in any outside platform. That’s just as true for CDP as it is for DMP.

Could this change? It might, with Google’s Federated Learning of Cohorts (FloC) — a proposal to create customer cohorts to pass to and from other platforms in a way that would protect customer privacy. This solution is in the prototype phase and we are at least a year from seeing what use cases it can enable and how to do it safely for the customer.   

Marketing teams need to add focus on building first-party data sets with opt-in and to continuously quantify the value of those customer names. In the DMP world, marketing teams could use third-party cookies across the open web to amass large sets of customer data and segments into their DMP. Without third-party cookies, this changes. Now marketers can collect anonymous data from their first-party data (typically browsers of their websites who have acknowledged cookies, and customers who provide their PII at some point along their journey). This requires two strategies:

1) Allocating performance budget to accelerate growth of opted-in first-party customer data. 

2) Quantifying the incremental value of each new customer name acquired in order to measure the ROI of customer name collection to the business. This type of analysis often requires designing a control group strategy to understand the incremental customer lifetime value of marketing to newly acquired customer names.   

Need to replace your DMP? What to look for in a CDP for an effective transition: 

Given the similarities between what CDPs and DMPs aim to accomplish, and the ways that CDPs overcome the limitations of DMP, a CDP is a smart investment for brands wondering what to do with their DMP. Not all CDPs are going to achieve this equally though. Here are some key things to look for:

• A CDP with an identity solution good enough to be trusted as the enterprise source of truth. It should be able to create a high quality 360 all of all known and anonymous data available, and it should be equipped to replace the 2p and 3p data needs of your DMP initiative. 

• A CDP should come with rich off-the-shelf audience insights and predictive analytics to produce audiences, coordinate omni-channel journeys, and drive incremental ROI performance, plus the ability for all teams to easily explore customer data and analytics to create unlimited custom attributes, segments, and campaigns.    

• It must have anonymous-to-known capabilities alongside connections and interoperability with the cookie-less ad-tech ecosystem of the future, like Throtle and UID2.0. It should support direct sync to Facebook, Google, and Amazon, as well as continue to innovate new direct sync to emerging data providers built for a post-cookie advertising ecosystem. 

• Your CDP should have the enterprise scale and power to handle the customer data needed to fulfill critical advertising and measurement use cases - think customer interaction data like impression logs or other mega-scale data sets. This can quickly go to trillion-entry scale and surpass the compute capacity of many platforms,  so they can’t process the data and make it available in real-time — which creates implementation and operation headaches or inability to execute use cases.

• There should be dedicated workflows for every team across the business to access and use customer data — whether IT for data management and governance, analytics teams for workflows that are both sophisticated and easy to use, or marketers and media teams for no-code point and click exploration, audience creation, and campaign management. And it should be easy for not only the marketing teams but also the full enterprise to get the right customer data sets into the tools and platforms where they need it (for customer care, website teams, market research, finance). Your CDP investment needs to meet these needs or it will become just another customer data silo amongst others. 

• Your CDP should be independent and system-agnostic. Interoperability is make-or-break for a successful CDP initiative, and a make or break from an advertising perspective. If your CDP is part of a marketing cloud or walled garden, there are going to be issues integrating data from outside that technology provider.  An agnostic CDP should work with your investments in large marketing clouds and also enable the flow of data to any platform or technology system across your ecosystem and to any DSP, publisher, or paid social channel.    

• Your CDP provider should be sufficiently funded with large, long term renewing customers so you know it can stay independent for the long term, protecting you from being forced through an acquisition into a walled ecosystem with misaligned priorities for your data. 

Let’s take a moment to say farewell to Salesforce DMP and its promise. As for DMP as a broader category? Let’s watch and see. In the meantime, CDPs are stretching well beyond the vision of the DMP and into bold new territory. 

If you'd like to talk more about how a CDP can take you further than a DMP, get in touch.