Solving the data martech integration challenge

Over the past decade, marketing has become inseparable from technology. From campaign planning to measurement and optimization, nearly every part of the process now relies on a stack of tools. If you want to get something done in marketing today, chances are you’re logging into a platform to do it.

In fact, the 2025 version of the Martech Lumascape contains an astonishing 15,384 solutions providers across 50 categories and many more sub-categories, covering everything from CRM to influencer marketing, project management, marketing analytics and many more.

As a result of all this choice, brands and agencies the world over have found themselves in the misleadingly reassuring position of being able to find a technology-based solution for almost any task, and their martech stacks have grown accordingly.

But growth is not always a good thing, if the various components of that tech stack are not able to properly communicate with each other. Because with each new tool you add, you’re not talking about one new integration, but about seamlessly integrating the new tool with all your existing ones, and that can create an exponential amount of additional complexity. 

This is the problem facing many companies today, and it’s compounded by the fact that, as the saying goes: “If everyone has it, no one has it”. Because the true value of all these platforms and SaaS solutions doesn’t reside in the tech itself, but in the unique data – your data – that it has to work with.

And when I say data, I don’t mean customer IDs like email addresses, phone numbers or cookies; when you consider all the levers a marketer can pull, targeting data like this has relatively little impact. What is far more valuable is business data: merchandising inventory data; third-party datasets telling you the average income in different postcodes; data that comes from sales teams on the ground; or perhaps the dynamic pricing data an airline uses which could, if shared, have value for a marketing team looking for in-demand routes. 

The importance of building a shared business case

All of these data points, and many others, are crucial for steering algorithms toward real business outcomes, not just media performance. Think profit margins, inventory levels, or average product return rates. These aren’t marketing metrics, but they shape how a campaign should behave if it’s going to drive meaningful impact. The problem is, most of this data lives elsewhere in the organisation, out of reach for advertising teams.

So the first challenge for marketing teams hungry for data is not a technical one, but an organisational one: how do we get buy-in from a team outside of marketing to let us use this data? One of the ways to solve that might be to build a shared business case together.

The next challenge is integrating the data, because the granularity of some data sources may be very low, while for others, it’s much higher. You might have access to the overall profit margin on dresses, for example, which it’s helpful for understanding category-level performance. But as a marketer, you’re trying to optimise at the SKU level: Which styles drive the most profitable demand? Without the ability to connect these layers, you’re stuck making blunt decisions with broad averages.

Historically, we’ve relied on shared identifiers to link datasets together; SKU codes, store IDs, sometimes even customer IDs. But today, those connections are harder to make. Data may live in different systems, privacy constraints limit access, and teams don’t always use consistent keys. Even when the data technically lines up, the governance often doesn’t.

This is where modeling helps. Probabilistic methods can infer relationships between disconnected data sets. And thanks to Gen AI, it’s easier than ever to convert unstructured inputs, like images or text, into structured data that can support automated decision-making. For instance, recognizing that this picture shows a green dress becomes a usable input for tagging, personalisation, or even forecasting.

Gen AI also opens up data access by changing how we interact with it. Tools like Google Agent Space let non-technical teams query the data warehouse using natural language, skipping the need to go through data planners or IT.

But then you need the proper governance in place – because when you have users from different parts of the business asking questions, they may have different definitions of what constitutes, say, an active customer, which in turn risks the answer being misinterpreted. I have one client who has seven different definitions of a customer. So we have to accept that every time you solve a problem – or almost every time – you probably create a new one.

Three rules for data integration success

In conclusion then, integrating data across a sprawling martech stack is not without its challenges. But there are three points that underpin every successful integration:

• First, build a strong business case, with multiple teams aligned on the target outcomes.
• Second, make sure you have support across the business for what you are trying to achieve, but allow yourself permission to pivot if the initial approach isn’t working.
• Third, don’t make the mistake of setting and forgetting a data integration project, because data decays very quickly as new features are launched and others are discontinued or no longer supported.

Data needs ongoing care and attention, just like any of the people you hire. But if you get it right, a fully-fledged data integration project will benefit your business in more ways than you can imagine. 

Also published in: Martech360