The MarTech landscape has exploded from a handful of vendors to over 15,000 solutions in just a decade. Organizations that implemented their first MarTech wave now face critical architecture decisions: platform ecosystems that promise integration, composable stacks built for flexibility, or hybrid approaches. In this episode of Data Driven Voices, host Emma Storbacka sits down with colleagues Kalle Heinonen and Gustav Malmqvist for a discussion about why AI is driving stack reconsideration and why metadata management might be the deciding factor for AI-powered automation.
In this blog, we summarize some of the key takeaways from the episode.
From simple vendor choices to strategic architecture
Ten years ago, MarTech selection was straightforward. Kalle reflects:
“There was only a handful of vendors. Building RFPs at the time was a little bit more straightforward. Today we have, I think it’s like 15,000 different vendors in that MarTech report.”
The explosion of options changed everything. Companies once chose a logo and implemented that vendor’s suite. Now they face strategic decisions about architectural philosophy. RFPs pit platform ecosystems against composable stacks, with hybrid approaches as a third option.
This reflects a deeper shift. It’s not just about getting technology in place anymore. It’s about architecting systems that can adapt to rapid change.
The AI wave driving reconsideration
Why now? Gustav points to a shift comparable to cloud transformation:
“There was a big wave 10-15 years ago when everything was moving to cloud. The wave that we have right in front of us is AI that is suddenly impacting all of these tools and making some tools and processes redundant because it can be done in a heartbeat in so many other types of ways.”
Cloud enabled omnichannel experiences and personalization at scale. Now AI promises to scale even further and automate what previously required specialists.
Organizations face a critical question: Is our current stack future-proof for AI-driven automation? For some, the answer is obviously no (still running on-premises servers). For others, platforms work but feel stuck. The technology renewal creates a natural opportunity to also reconsider the operating model, people, and processes that determine actual success.
MACH principles and composable architecture
Gustav introduces the MACH framework. The acronym captures the logic:
MACH stands for:
- Microservices: Smaller, focused tools instead of one solution for everything
- API-first: Data flows quickly between systems in real-time
- Cloud-native: Everything connected in the cloud, accessible anywhere
- Headless: Services decoupled, each with one clear purpose
Composable approaches excel at flexibility. Need to test AI decisioning or agentic personalization? Plug in a best-of-breed tool without overhauling your entire stack.
Organizations with strong data warehouse foundations can make their data platform the system of truth. MarTech tools function as execution layers. All customer and product knowledge lives in the data warehouse. Marketing automation and activation tools pull from that central source.
Emma connects this to organizational impact: does data-centric architecture force teams to work differently? Both Kalle and Gustav agree it does. When data lives centrally and tools access it, silos must break down. IT and marketing need unified workflows.
Kalle shares a client example. The organization has excellent data infrastructure and segmentation capabilities. Rather than committing to full platform migration, they’re adding a composable orchestration layer on top. Trial it in one to two-year sprints. If it delivers value, continue. If not, pivot without massive sunk costs.
Platform power and the metadata advantage
Platform approaches offer compelling benefits for complex organizations. Gustav describes implementing marketing automation with integrated personalization that launched in months.
Platform advantages:
- One vendor to hold accountable
- One ecosystem to learn and master
- Integrated AI engine with full context
- Fast deployment compared to composable builds
Emma brings up a critical insight: metadata might be one of the greatest platform advantages. As organizations move toward AI and automation, metadata becomes essential for agents to query and act on information. Kalle explains the challenge:
“Creating metadata is super time consuming. Every piece of data needs metadata attached to it, and that metadata needs to be correct. If your data changes, the metadata needs to change with it. Otherwise you’re teaching your AI with wrong data.”
Platforms solve this by standardizing metadata across their ecosystem. Content assets, campaign data, customer interactions all carry consistent metadata that AI agents can query and act upon. Salesforce Einstein or Adobe Firefly have full context across the ecosystem.
As Gustav notes, this is incredibly powerful but only within the box. The agent sees what’s in the platform, not outside it. For organizations committed to the full ecosystem, this delivers automation that composable setups struggle to match without explicit metadata management strategies.
The hybrid reality and forcing organizational change
Can you be half platform and half composable? Kalle says absolutely. The choice depends on several factors:
Key decision factors:
- Which use cases drive business value?
- How fast do you need to move?
- What budget can you invest?
- What organizational capacity do you have for change?
Most organizations operate in hybrid reality. They’ve invested in platforms delivering value in specific areas. The question isn’t rip and replace. It’s how to extend capabilities. Keep your platform core and add composable elements for experimentation? Modernize within your ecosystem while maintaining data warehouse flexibility? Both are viable.
The architectural choice also forces organizational change. If you move toward data-centric composable architecture where knowledge lives in the data warehouse, IT and marketing must work together differently. You cannot just implement a system and hope. Data teams, marketing operations, and business stakeholders need unified workflows and shared goals. Whether platform or composable, achieving personalization at scale requires breaking down silos.
Emma frames this as avoiding the “Ferrari in the garage” problem where organizations buy tools, enable a few use cases, and hope something happens. With composable architecture especially, the data team built to power personalization would be asking what to do next. You have to build it use case by use case. It’s not a tool you buy and leave sitting there.
Can enterprises vibe code their own MarTech?
Emma poses the question to Kalle, one of Avaus’s master vibe coders: could enterprises actually build their own MarTech components instead of buying them?
“You can actually build complete systems with Google authentication practically to any SSO with, for example, just with a lovable. And you can do the same within a secured architecture as well as having even having embedded vector databases. So yeah, my answer is yes, definitely you can do that.”
Will Nordic enterprises build rather than buy this year? Kalle believes yes. The technical barriers are falling rapidly through AI-assisted development platforms.
Whether organizations have the appetite and capability remains uncertain. But the tools exist. The modern question isn’t just platform versus composable. It’s whether custom-built solutions become a fourth option for specific use cases.
Key takeaways for MarTech architecture
Kalle and Gustav’s discussion reveals what matters when choosing architecture:
- Architectural choice is strategic: Organizations choose between platforms, composable stacks, or hybrid approaches. Each enables different capabilities.
- AI creates urgency: Just as cloud drove the first MarTech wave, AI forces organizations to evaluate whether their stack can handle automation.
- Composable offers adaptability: MACH architectures let you swap components and trial new capabilities. But they require data warehouse maturity.
- Platforms provide speed: Ecosystems reduce organizational complexity and solve metadata management. But they work best with full commitment.
- Metadata enables AI: As automation becomes central, metadata quality determines what’s possible. Platforms have an advantage here.
- Hybrid is viable: Most organizations will extend rather than replace, adding capabilities based on use cases and capacity for change.
For organizations evaluating MarTech architecture, start with understanding your data maturity, organizational capacity for change, and which use cases drive business value. The architecture follows from there.
Inspiration for marketing, sales, and data professionals
Data Driven Voices is a podcast where Avaus together with industry experts, thought leaders, and partners discuss how to harness data, technology, and strategy to drive meaningful change and business results in primarily marketing and sales. The podcast shares actionable insights, success stories, and thought-provoking challenges to help professionals with new perspectives.