Editorial
The Gist
-
Architecture matters. AI decisioning needs a solid data foundation and a layered approach to thrive in martech stacks.
-
Siloed to seamless. Integrating AI decisioning across the enterprise reduces inconsistencies and enhances decision-making.
-
Organizational integration. A seamless, AI-powered decisioning layer helps organizations scale, innovate and adapt to changing demands.
Editor's note: This is Part 2 of a three-part series on enterprise decisioning in martech. Part 1 was AI Decisioning vs. Enterprise Decisioning: What CX Leaders Need to Know.
In my previous article, “AI Decisioning vs. Enterprise Decisioning: What CX Leaders Need to Know,” I delved into the differences between AI and enterprise decisioning, and I provided examples of when each is applicable in marketing.
But how do these two technologies fit into martech ecosystems, from both a technical and architectural point of view? Hang onto your hats as we take a deep dive into this very topic.
As the market continues to shift away from more traditional business rules or trigger-based enterprise decisioning and towards AI infused enterprise decisioning, you should be asking yourselves these questions. Is a siloed or seamless approach feasible for my organization? And where in the stack should AI decisioning live?
Table of Contents
- Comparing Siloed and Seamless Approaches
- The Foundation of Data
- AI Decisioning Across the Enterprise
- Meeting the Demands of an AI-Driven Market
Comparing Siloed and Seamless Approaches
We have talked for years about the concept of a seamless customer experience, in which using data from one interaction channel informs engagement in another. For example, if I browse a company’s website and then contact their call center the next day, they have signals or intent as to why I may be calling.
Unfortunately, in most instances, that isn’t happening because it’s very difficult to architect. In some instances, it’s hard to believe that the concept of an enterprise decisioning layer is feasible to construct, largely due to legacy architectures and technical debt. As a matter of fact, In a LinkedIn discussion on this topic, one commenter asked, “Where’s the ontology?” Where is the proof that this could become reality?
Related Article: How Stack Composability and AI Can Supercharge Martech Stacks
The Foundation of Data
To me, a picture speaks a thousand words: This is my “ideal” (not perfect) martech stack.
Let me provide some context. All good architecture diagrams start with the data at the bottom. Well structured and governed data is the key to AI success.
Above that data layer comes what I call the data app layer. These are the apps that rely on the data layer to operate. Some are channel/function specific (i.e., advertising, customers, assets and products), and these are in navy blue. Others should be more enterprisewide (i.e., MDM, analytics, enterprise data catalogs and CDPs), and these are in royal blue.
Many times, technologies that are purchased in the royal blue end up being siloed or or limited to specific channels or departments. These layers are typically where we see AI introduced as smaller, embedded features within individual solutions.
AI Decisioning Across the Enterprise
Above this, in the aqua blue layer, is where the enterprisewide decisioning layer ideally should be. Over time, this layer will rely less on rigid business rules and more on AI.
In my opinion, this layer should span the enterprise and provide decisioning capabilities across the organization. If enterprise and AI decisioning is siloed into a single application or technology within the data app layer, it’s not enterprisewide, by definition.
(As an aside, AI decisioning follows a technology path that starts with analytics, then artificial intelligence, then machine learning and finally reinforcement learning, all aimed at making decisions based on large datasets. So the idea here is that systems can make adaptive decisions and offer actions and outcomes that improve over time, all with minimal human intervention.)
Webinar
Jul
29
Insights to Action Rethinking the Contact Center for Real Business Impact
Join our exclusive webinar to hear CX executives share their innovative strategies for transforming service delivery.
Register
Webinar
Jul
31
Cracking the Code on Martech Modernization
Register now and learn how to turn martech mess into measurable outcomes.
Register
Webinar
On demand
Personalization at Scale: How Ecommerce Brands Actually Pull It Off
Unlock the secrets to scalable retail personalization. Save your spot now - your future customers will thank you.
Watch Now
Webinar
On demand
Demo Derby DXP Edition Pantheon vs Progress Sitefinity vs Contentsquare
Three platforms, one virtual stage. Experience the demos live - no folding chairs necessary!
Watch Now
Webinar
On demand
From Search to Solutions: How AI Agents Can Power Digital Commerce in 2025
Join our webinar to discover actionable AI strategies that will elevate your brand and boost your bottom line.
Watch Now
Webinar
On demand
Drive B2B Growth on LinkedIn With Video and AI
Join our webinar to unlock AI-driven video strategies on LinkedIn, boost performance and engage high-value prospects!
Watch Now
My ideal stack has this decisioning layer integrated within every department, function and capability across the enterprise for centralized organizational decision-making. This layer will eventually be driven entirely by AI, and it will use AI to make automated business decisions with fewer guardrails or conditions than traditional enterprise decisioning.
Embedding AI Decisioning Across Every Layer
Next up is the sky-blue layer which represents those downstream apps that rely on everything previously mentioned to do their best work. Some tools like BI/REPORTING need the data more than anything else, while others like marketing automation platforms, customer engagement platforms and customer service platforms rely on both the data and decisioning.Finally, we get to the light gray layer, which I call the action or activation layer. This layer is broken into two sections, which we will call the “now” and “near-to-now” layers. The now layer includes native channels such as web, mobile and email, as well as third-party integrations, social platforms, community tools, and software that supports events. These are coined “now” because most organizations are using many of these channels currently.
The “near to now” or agentic AI layer includes all the AI agents brands are using now or in the near future. They may include chatbots, virtual assistants, recommendation systems, sentiment analysis, predictive analytics, content creation and personalization tools. They may also include voice and visual recognition, robotic process automation agents and other AI-driven insight systems.
Meeting the Demands of an AI-Driven Market
You can see by the diagram that my ideal stack architecture institutes a seamless approach. Getting to seamless or truly enterprisewide AI decisioning will require accounting for several key factors. Part 3 of this series will discuss those factors in more detail. Stay tuned!
Learn how you can join our contributor community.
