Combining your DMP and Search strategies: how (not) to get lost along the way

Why use a DMP?

Before getting to the heart of the matter, let’s revisit how to exploit the richness of available data, depending on its source:

• First of all, there is first-party data: this is data about visitor behaviour that advertisers collect from their own sites. This data is collected most frequently via a web analytics tool (site-centric).
• Next, third-party data: data collected by a third party about user behaviour on a network of sites beyond the advertiser’s site. This data is treated, packaged, and made available to advertisers for advertising. For example, this can include socio-demographic or intentional data (based on types of pages previously visited). Providers of this type of data are generally tech publishers. Access to this data is sometimes free. For example, Google Display Network makes these types of “ready to target” segments available.
• Finally, advertisers are more and more interested in second-party data. This data is collected and shared under an agreement between two advertisers, via a shared tracking system (see the article on this subject on Tea House).

A DMP centralises all these types of data on a single platform, from which marketers can then combine different criteria to easily and intuitively create customised audience segments.

These segments can be injected across different marketing platforms for advertising purposes, which allows for specific messages to be sent and a bidding strategy to be adopted which will be more or less aggressive based on the audience. It should also be noted that certain marketing platforms like Google or Facebook are able to create audiences with similar behaviour to your own DMP audiences (called “similar audiences” or “lookalikes”).

How specific, or even complex, is leveraging a DMP for Adwords campaigns?

We have given an overview of different types of audience data to which publishers have access. Unfortunately, not all data sources (and thus not all resulting audience segments) can be exploited through RLSA campaigns, unlike display campaigns, which are less constraining.

• Audiences based on third-party data or combinations that use third-party data may not be exploited for Search campaigns
• The use of second-party data is vague as yet, and so we strongly recommend publishers to discuss with their agency or Google account manager
• For first-party CRM data, the email-cookie matching rate for the DMP is often lower than that of Adwords’ Customer Match function, which loads the CRM database directly in the Adwords interface and matches data natively.

Also, reporting data might be incomplete in several specific cases:

• For publishers who do not want to track conversions into the AdWords interface, reading performances for each audience is very complicated or even impossible if campaign structure is not suitable.
• For advertisers using a third-party bidding tool (such as Marin or Kenshoo), it is not possible to read RLSA campaign performance if the campaigns target both the queries of DMP audiences and those of users who are not included in DMP segments. Currently, the only way to read DMP performances with a third-party bidding tool is to create campaigns that target DMP audiences only. This means duplicating a non-DMP campaign, and, within the new DMP campaign, duplicating the same ad group as many times as the number of DMP audience segments targeted by the campaign. It also would mean that DMP audiences must be mutually exclusive, and excluded from the standard non-DMP campaign. In other words, it means yet another labyrinth!

Setting up dynamic ads is also still a complex task. Here, too, campaigns have to be duplicated to address specific messages, which can heavily impact campaign structure, and even more so day-to-day operational management.

The good news is that at the end of 2016 Google announced a new feature called Ad Customizer, based on audiences. Ad Customizer is not a new product, but now it allows marketers to send a specific message using a data feed or dynamic parameters based on a remarketing audience. It can thus be used (among other things) to set up advanced dynamic messages, without having to duplicate campaigns.

Useful tips from the teams at 55 about using your DMP with Adwords and measuring performance

For Tracking and Measurement: we recommend using the Adwords conversion tag, or importing certain goals from Google Analytics into the Adwords interface to precisely measure each audience’s performance.

If you have an Adwords remarketing tag, or if you have Google Analytics audiences shared with Adwords in addition to audiences injected through your DMP, it is important to compare reachable volumes across the different tools. A choice must be made regarding which lists should be migrated to Adwords to avoid juggling among several audience sources. Otherwise, managing mutual exclusion can become complicated.

For the operational side of things:

• Find a balance between your media tactics’ business impact and its operational impact. Displaying a specific message for all keywords and all audiences systematically must be avoided at all costs. For example, for a brand request or a generic request, the request’s intention is relatively less explicit, and we can thus test several messages of varying specificity. On the other hand, for a request that is already very specific, it is not necessary to systematically create a message dedicated to DMP audiences because they are not necessarily fundamentally different from non-DMP audiences.
• Assign audiences to campaigns (this is possible since December 2016), as this will simplify the structure of your account
• Use Ad Customizer to send specific messages when necessary, which avoids multiplying the number of campaigns
• Adopt a bidding strategy in function of audience performances, and combine this with Adwords’ Smart Bidding feature which enriches the strategy in real time, query by query, while taking multiple signals into account (including historic performance indicators like click-through and conversion rates, but also geolocation, query as typed by the user, browser, language, advertisement type, and OS).

At fifty-five, we think it is essential for brands to make data the focus of their advertising strategy. Capitalising on a DMP while respecting operational best practices will help to refine campaign monitoring and lead to better performance. Yet, as you can imagine, combining an Adwords strategy with a DMP can be quite complicated. We recommend that advertisers work with professionals with experience in this realm.

Translated from the original French by Niamh Cloughley.