Marketing data technology: making sense of the puzzle

Marketing Data Technology Map

The following is a guest post by the CEO of Sweetspot Intelligence, Sergio Maldonado. I had the pleasure of meeting Sergio at MarTech Europe last month where he shared a sketch of the above “marketing data technology” landscape with me. I found it very interesting and invited him to share his perspective with readers here.

Scott Brinker’s Marketing Technology Landscape Supergraphic has become a global reference for marketing executives, marketing technologists, digital strategists, IT executives, and investors. Having grown in size and complexity, many of us in the “martech” space are often asked to provide our own basic guidance with regards to the generic rules under which its many categories coexist with (or replace) each other.

In my case, rearranging the existing sections and categories on the basis of the manner in which data flows between them or is used by them has done the trick. After all, most of the (as of today) 43 categories listed consist of either making sense of data or using it (directly or indirectly) to create marketing experiences (ads, content) for the consumer.

The above chart and supporting ideas are an attempt at making this same guidance available to others. We could call it a “data-aware marketing technology landscape” or, simply, a “marketing data technology” map.

For the sake of simplicity and ease of consumption, I will kick it off by laying out the seven basic assumptions sitting behind the chart above. I will then follow that up with some overall thoughts as well as a vendor-populated version of the same chart.

1. Provided they are data-centric or data-powered, marketing technologies fall within five main layers: Backbone, Discovery, Delivery, Activation, and Automation.

The Backbone is made of technologies and media that allow us to gather, process, and store data. This will not only consist of stand-alone solutions, but also of the architecture behind many of the solutions clustered elsewhere.

(Data) Discovery encompasses a wide range of human-driven functions, from predictive modeling to digital analytics. Of those, certain functions are not exclusive to the marketing space.

(Information) Delivery refers to the ultimate brain-powered task: decision-making. It represents a bridge between data and organizational change; between performance management and marketing outcomes. While in its purest form it is represented by executive scorecards and dashboards, this layer will also encompass data governance, “insight management” and even “built-in marketing know-how” (in the form of cross-channel intelligence).

Activation relates to putting data to work at the most tactical level, establishing a seamless connection with the marketing experiences it helps generate. As one would expect, it is closely intertwined with the fifth and last piece, Automation, as multiple tasks are eventually systematized, then automated. Many of these tasks will already fall within the realm of the so-called “AdTech”.

(These layers have been labeled as “Backbone”, “Discover”, “Decide”, “Activate”, and “Automate” respectively in our chart.)

2. Data will be centralized by brands and content owners, but not fully integrated under a single roof.

There will be a data lake. And there will be a DMP. And there will be advertiser-side data marts or a data warehouse, but agencies and media will also maintain separate brand-related repositories and systems, directly connected to the marketing campaigns or initiatives they define and manage.

Furthermore, aspiring to a fully integrated scenario (“give me the father of all repositories!”) will pretty much destroy the value of certain categories of data for medium-specific analysis, delivery, activation or automation purposes.

Drilling further on the Backbone layer, the “house and garden” shape shown in the map illustrates other important data-architecture constraints:

(a) SQL and NoSQL databases coexist in most corporate environments, providing alternative solutions for the storage of structured, unstructured and semi-structured (or multi-structured) data.

(b) In the Data Warehouse space, the platform-as-a-service model is particularly well adapted to the marketing environment, with multiple customer interactions already happening and being collected in the cloud.

(c) First-party website and mobile app data have evolved separately, with all web and mobile Discovery tools (“Digital Analytics”) incorporating processing and storage services specifically tailored to the data they collect (directly or via Tag Management Systems). This does not prevent them from feeding raw data into separate repositories.

(d) Social media data is collected, processed and stored by each platform. Some of that data will be exclusively provided to account holders, while some of it will be publicly available.

(e) Tag Management Systems (whether independent or attached to a DMP offering) are equally able to send raw event-based data to semi-structured repositories.

(f) Most of the tools referred to in (b), (c), and (d) count on APIs for Information Delivery, joint Discovery (where possible), Activation, or Automation purposes.

3. The very nature of each piece of underlying data has a direct impact on our ability to analyze (Discovery) or share it (Delivery), on its own or in conjunction with others.

As it happens, marketing data (mostly digital) represents a serious challenge to traditional business intelligence (“BI”). Whereas the former becomes predominantly semi-structured or unstructured, BI was built around structured data (i.e., data models associated with relational databases). While much of the marketing data now available does not leave room for anything beyond correlations, BI aims for good, old causality.

Various analysis and delivery environments will thus coexist in an apparent show of inefficiency, but in fact reflecting the natural disparities of the underlying data models. Social Analytics functions will, for instance, be best performed by tools specialized in the retrieval, storage, and processing of social feeds. On the other hand, analyzing customer properties in a data warehouse will be best left to predictive, descriptive, and visual discovery tools.

More on this point: How Digital Data disrupts Business Intelligence.

4. The connectors between the last four layers power essential functions within them.

  • Insights are the most important output generated by the Data Discovery functions. They in turn become a crucial input for Information Delivery tasks (out of which stem specific requests for further insights or context on a given status update).
  • Similarly, the outcomes of customers’ marketing experiences (campaign results) power both Discovery and Delivery tasks, while data science-powered models, another important fruit of the Discovery layer, feed Activation and Automation functions.
  • Executive decision-making will result in the longer-term business rules sitting behind marketing Automation, while Automation itself (machine learning, AI) will eventually power some of the diagnostics (and insights) that now see the light in the Information Delivery layer.

5. Human behavior is the limit.

What is the ultimate frontier of marketing actions? The unpredictable mind of the customer.
What is the ultimate frontier of internal marketing processes? The unpredictable behavior of people and teams.

The former explains that “attribution” models are not the holy grail we once hoped for. Or that a true “customer journey” will never become a reality. Unless, that is, we get to fully understand the human brain — at which point the entire marketing process will be ripe for full automation (and human beings ripe for replacement by robots).

The latter explains the differing speeds of technological progress and organizational change. Or the fact that all the metrics in the world cannot replace a strong piece of storytelling (“the API to the human brain”).

6. The privacy-aware customer mindset is unstoppable. As a result, privacy constraints (regulatory and customer-driven) have a much more serious impact than any MarTech vendor will usually concede.

It is a fact that an ever-growing proportion of customers is either disabling cookies (third party or all), or deleting them more often, while mobile access renders many of them useless. People are increasingly becoming fully aware of the data we collect about them — and acting in consequence.

Making it worse, the regulatory framework, and not only in Europe, is shifting its focus from PII (Personally Identifiable Information) data to specific methodologies used in the collection of any data (be it PII or not). Hence the impact of:

  • “Cookie laws”: The EU Privacy Directive in its current status, requesting prior express permission if cookies are to be used for individual profiling purposes (no matter how encrypted or unintelligible) and basically obliging marketers to redefine the manner in which DMPs, Tag Management Systems, or Social Login repositories are managed and put to good use.
  • Fingerprinting (seen by many as the perfect alternative to cookies) being considered akin to cookies in every respect by the same EU legal framework.

The primary consequence of all this is our inability to keep exploring a purely deterministic, “user-centric” approach, giving way to a “user-driven” or “intelligent audiences” scenario in which marketing experiences are automatically personalized on the basis of a combined deterministic/probabilistic approach.

More on this point: Revenge of the silos: how privacy compliance is cutting the customer journey short.

7. The (Information) Delivery layer connects Marketing Technology with the rest of the marketing function. More importantly, it connects marketing with the rest of the organization.

Information Delivery is tied to productivity, marketing performance and even “Insight Management”. Neither are the terms “dashboard” or “reporting” exclusive to this function anymore, nor would they be enough to give coverage to every one if its key components.

Elaborating on the first point, “dashboards” are now everywhere, at every level and every category. Every tactical function has a dashboard for human supervision or direct control. Every Data Discovery tool (medium-specific or generic), provides a means to reach out to data consumers, most likely in the form of a dashboard. However, none of these belong in the performance management (or marketing performance/intelligence) space. As a result, the term has ceased to be a valid label for any given category or task.

Further to the second point, it is the following additional pieces that really complete the Information Delivery picture — even more so when we consider the role of this layer as a bridge to the rest of the organization:

  • “Insight Management” (or “Digital Insight Management” in the especially dynamic digital marketing space) is about channeling analysts’ input to decision-makers and other data consumers in a way that their value-added conclusions can be acted upon (by tying it into various decision-making workflows).
  • “Built-in Know-How”: As it once happened with CRMs (embedding knowledge of the sales process within a relational database) or ERPs (industry-specific resource management know-how), Information Delivery tools are best positioned to incorporate recyclable pieces of “marketing know-how” – and therefore remain closer to Marketing executives than any other layer. This could come in the form of KPI libraries, cross-channel campaign management, pre-built scoring systems or ROI attribution formulas.
  • “Performance Management” is about tying marketing goals, forecasts or benchmarks to overall corporate goals. This is the “scorecard” soul of a Delivery tool, much more concerned about the “what” (performance over time) than the “why” (dimensional breakdowns of a given KPI).
  • “Data Governance” has different meanings at different layers, but the essence of it at Delivery level is a proper distribution of metrics and objectives amongst team members.
  • “Team Productivity” should be more about building an effective bridge to collaboration platforms currently in place (integrated messaging, agile marketing) than entirely replicating what they already provide.

More on this point (I): Data Visualization vs. Dashboards; (II): Commitment, Time, People… Dashboards.

Overall thoughts and specific technologies.

In essence, although it may be true that the marketing technology space is gaining complexity, it would also seem like its key components are simultaneously maturing, with many functions quickly falling into place under an ever-clearer landscape that cannot escape some basic constraints: data (it comes in limited forms); people (individuals and teams); media/services (the canvas of our experiences); maths; and technology itself.

As points 2 (irreconcilability of data types) and 5 (inevitability of the privacy-aware customer mindset) above reflect, obsessively breaking data silos is not the answer. I believe we should instead adapt ourselves to working with multiple silos: connecting them where possible, overlaying them when not.

All that said, let’s now share a vendor-populated version of the same “marketing data technology map.”

Marketing Data Technology Vendors

Please notice that this is only provided for illustrative purposes, adding the specific tools I am most familiar with. I look forward to receiving your comments.

Sergio Maldonado is CEO at Sweetspot Intelligence (New York City). A published author on Digital Analytics and E-commerce law, he is passionate about data and marketing technology. Sergio holds an LLM in Internet Law (Queen Mary’s University, London) and a JD (UPV, Spain).

Share

Comments

  1. Brilliant. Thank you.

  2. Great stuff Sergio. Where do offline data and offline connectors such as Adometry come in to play in your opinion? Another tree in your data garden?

  3. Hi Jasper

    Thanks for your interest.

    The way I see it, they already have a place under “built-in know-how” (there is an “ATT.” -attribution- block in the Backbone layer). That piece can then power tools in any other layer. Their output could also be consumed on its own (eg. correlating TV ad events and their associated digital impact) and appear next to Visual IQ (delivery). No need to plant another tree 🙂

  4. I like it. Great syn-thesis Sergio!

  5. “There will be a data lake.” Excellent analogy. The challenge I see time and again are the API’s, or the tributaries that connect the lake with other bodies of water. Integrations are often difficult and unreliable, but they will (must) get better.

    Excellent post!
    -BH

  6. Hi Sergio, Thanks for this great landscape! I’ve shared it already many times here in the Netherlands and they love it!

    Because i’m dutch 🙂 and English is not my native language I need to have an explanation what you mean with ATT, M.I and MM you put in the CRM/know how block.

    thanks!

    Rob

Leave a Reply