Openness and optionality benefit both martech customers and vendors

The research featured in today’s newsletter is one small excerpt from our upcoming Martech for 2027 report that will be released on December 1. To get one of the first copies — as well as to catch our discussion about it live or on-demand — sign up here.

Signed up? Trust me, you’ll be glad you did. It’s a big release. Moving on…

I’ve long been an advocate for open platforms and ecosystems in martech. Since, as some of you know, I led the creation of HubSpot’s technology partner ecosystem from 2017 to 2025, you might assume that’s the origin of my bias.

But it’s actually the other way around. The reason I joined HubSpot to tackle that mission is because I strongly believed that open ecosystems were crucial to the future of martech. At the time I joined, most martech was still relatively closed. I was eager to see that change, at least for one platform, and prove that openness was not only better for customers, it was better for vendors too.

I’ve got the receipts of that proof. And now, thanks to AI, the advantages of open ecosystems have grown even more as the martech industry transforms from an app orientation to an infrastructure one.

(Breaking news from HubSpot as I write this: Brian Halligan has returned as CEO to “refound” the company. I have enormous respect for Brian — one of the most remarkable leaders I’ve known — and am excited to see the new direction he and Dharmesh Shah set. It’s a challenging time, but I am bullish on the opportunity they have to reshape the company and its platform for this next era of AI-driven go-to-market. The Customer Relationship Harness? Sounds promising. Godspeed, Brian!)

Give credit (or blame) to Claude and ChatGPT — but most of all to MCP, the nearly overnight standard created by Anthropic to let AI easily connect with data sources and software services across your tech stack.

When marketers were able to connect their favorite AI assistant to the apps in their martech stack and just ask in plain language for answers, insights, and actions, the heavens opened up. Or, less poetically, their attic of underutilized data and capabilities opened up.

You didn’t have to navigate Byzantine menus and configuration screens. You didn’t have to think through which pieces of data were sequestered in which different apps. You could focus on questions you wanted to ask and work you wanted to accomplish, and your AI assistant (originally a co-pilot and now an agent) would pull together your disparate data and develop a plan of action that spanned the silos of your menagerie of martech apps.

At least with those apps that could be accessed via MCP, APIs, or a CLI (stands for command-line interface, which is how most developers work with tools in a terminal window — yes, today’s state-of-the-art developer interface looks a lot like an old UNIX box from the 1980’s).

Once they had a taste of this magical experience, marketers began demanding MCP access (or similar) from all of their martech vendors. And many listened.

In our research for Martech for 2027, Frans Riemersma and I examined 2,203 real-world martech stacks and cataloged the 3,529 martech products we found in them. (That’s just a fraction of the 15,536 products in the full marketing technology landscape because the long tail distribution of those products is highly dispersed.)

Products that appeared in 100 or more martech stacks we considered to be in the “head” of our long tail distribution. Products that appeared in 10-99 of them we bucketed into the “torso” — popular, but far from ubiquitous. All the others in fewer than 10 stacks were placed in the “tail.”

We then programmatically evaluated those products to see which ones have documented APIs, which ones offer MCP support, and which ones have native integrations with one or more cloud data warehouses or lakehouses (e.g., Databricks, Snowflake, BigQuery). That last one is a little different than the first two, but related to this apps-to-infrastructure shift.

First, with APIs, we were pleased to see that 87% of the most common products in the head have documented APIs, as well as 76% of the products in the torso. Even 61% of the tail. Across the board, the majority of these products have APIs.

This is important, because APIs are a bellwether for how “open” a martech product is — which determines how much you can do with it through your own AI agents and assistants.

Of course, having an API doesn’t necessarily mean you have a great API or that you provide full coverage of your product’s data and functionality via that API. Candidly, most martech products are lacking here, requiring you to use their UI — like an animal — to access their full capabilities. But that subtle (or not-so-subtle) “UI lock-in” moat is now a liability, as it hinders them from adapting to the infrastructure opportunity looming in front of them.

Back to the data: we were also pleased to see 40% of the products in the head offered MCP support, as well as 32% in the torso and 17% in the tail. Less than APIs, but pretty impressive given that MCP is barely two years old. This level of adoption in such a short time period is a testament to how much demand there is for it.

Of course, as with APIs, not all MCP implementations are equal. Coverage is the key differentiator — how much of a product can you work with via MCP? Other factors include how performant it is, how efficient its token usage is, and how granular the governance controls around it are.

I’ll give Salesforce props: Salesforce Headless 360 and Claudeforce are pretty good examples of what this looks like when you’re serious about it.

Finally, we looked at built-in support for cloud data warehouses/lakehouses (CDW). 38% of the products in the head, 30% in the torso, and 18% in the tail claimed to have this capability. That’s almost as much support as for MCP.

There’s an interesting parallel between CDW and MCP support. While MCP lets marketers bring their own UI and workflow to their martech stack, CDW gives them a similar flexibility with their data. They can push data out of a martech app into their CDW to do with it what they want — and increasingly take data from elsewhere in their CDW and pipe it into any receptive martech app. This freedom opens up a ton of possibilities for innovation. And as with MCP, once marketers get a taste of it somewhere, they really want it everywhere.

But also like MCP, which breaks the “UI lock-in” that many martech products enjoyed, CDWs break their “data lock-in.” This caused many martech vendors to hesitate at first. If you’re wondering what’s left for martech vendors to own once their data and UI are fully open — the exact question that, I assure you, they have all been asking themselves — the answer is actually quite a lot.

I’ve written about the opportunity for context-as-a-service several times this year as one framing for the deep, rich layer that needs to sit between your data and your AI. In Martech for 2027 — you did sign up for that, right? — Frans and I will introduce a more advanced framework for orchestrating “coherence” in an AI-explosive marketing environment. So much of this is underserved today, because many of these desired (and now necessary!) capabilities have emerged, or been amplified, by AI.

There’s a world of opportunity in martech’s new infrastructure layer.

It’s pretty obvious why martech customers would want open systems. It gives you optionality. That’s always been a useful position to have when negotiating contracts with your vendors — having choices gives you leverage. But in an environment like today, where things are changing so rapidly, having options to bring in new technologies, swap out old ones, or even build your own gives you the competitive advantage of agility. More important than ever.

But less intuitively, embracing an open ecosystem strategy can be enormously beneficial to martech vendors too.

First, it gives customers what they want — that optionality. Ironically, that’s a powerful reason for customers to stick with a vendor. It gives up some of the vendor’s leverage at renewal time, but it trades it for customer love. Nobody likes to be held over a barrel. People will pay a premium to work with a vendor that they trust, that doesn’t corner them, that focuses on winning their loyalty by delivering great products and services.

A vendor says they’re customer-centric? Openness is a litmus test of that claim.

Second, ecosystems are incredible go-to-market engines. And I mean that in the broadest possible sense of go-to-market. It’s not just explicit co-selling or co-marketing. It’s the myriad of ways that different kinds of partners, adjacent and complementary businesses, and community participants and stakeholders influence and impact the success of a platform’s customers — and therefore the success of the platform itself.

This is a massive — and, sadly, often underappreciated — moat. It takes years to build this kind of interrelated web of mutually reinforcing connections. It’s built on trust and respect. If you’re measuring its value on a kind of last-touch attribution metric of who registered the deal, you’re missing the big picture. At a time when the chessboard is being reshuffled and customers are turning to many, many different sources for help in their AI transformation, the “sale” is much greater than the transaction. If you’re not truly open though, you block off a huge portion of that potential ecosystem from aligning around you.

Finally, an open architecture imparts great optionality to the vendor itself. It makes it easier to bring in new, emerging capabilities quickly — whether it’s through partnerships, acquisitions, or internal disruptive innovation projects. We’re seeing this right now, where vendors who invested heavily in APIs have been able to move the fastest in wrapping their programmatically-accessible capabilities with MCP, riding the tidal wave of new AI interfaces and workflows.

I’ll stop here, as I could write a book about ecosystem strategy. (That idea may fester.) But the tl;dr is that openness, when done right with the right business support around it, can be a major strength, not a weakness, for the vendor too.

Mark my words: the martech infrastructure wars will be won by those that are most open, with the most enlightened ecosystem strategies. Because what makes infrastructure valuable is everything you build on and around it.

Keep an open mind,

Scott

P.S. The matrix at the top of this post is a view of the relative “openness” of martech categories across the full 15,536-vendor landscape, ranked by their documented APIs and MCP support. You can get an advance copy of a How Open Is The Martech Stack brief immediately when you sign up for Martech for 2027.