Summary: Navigating the Marketing Technology landscape has exploded to include over 11,000 companies, inundating marketers and data teams with disparate data sets. A metrics layer as part of the data stack is proposed to align marketing and data teams towards a unified understanding of metrics. Data challenges include schema mismatch, data type inconsistencies, and data granularity among others. The single source of truth is not a tool but the data itself, requiring a data-driven culture that understands metrics. The solution involves a social contract and a metric model that adapts to evolving data shapes and business needs.
Navigating the Marketing Technology (Martech) landscape is like charting a complex topography that's ever-changing. ChiefMartec's Marketing Technology Landscape is now populated with more than 11,000 companies and new tools are being discovered daily.
This explosion in Martech is something that we witnessed first-hand: In 2014, when there were only about 900 vendors, we wrote an email to Scott Brinker asking to be included in the highly regarded report. At the time, Klipfolio was focused on data analytics for digital marketers and agencies -- so not only were we a vendor, we were also a service provider.
Marketing technology is a broad field; it encompasses any software that marketers utilize in their daily work.
With each new tool added to the Martech stack, marketers and data teams are inundated with disparate, often incompatible data sets. Martech can be highly disposable. This means marketers move from one platform to another, freely erasing rich data legacies while maintaining their reporting requirements.
Are we any closer to solving this data challenge in 2023 than we were in 2014?
In this post, we'll explore the difficulties in aligning marketing and data teams towards a unified understanding of their metrics, and delve into how to establish a 'metrics layer,' a framework that can help both teams navigate the labyrinthine data landscape.
The Big Data Frustration
While waiting to see if we would be included in the 2015 report, we kept busy. Klipfolio is a data analytics company, and our Klips product has found a good fit with digital marketers. This aligns well with the fact that Klips' technical foundation is its data ingestion engine. Klips specializes in connecting to 3rd party APIs, making it incredibly flexible.
Marketers love Klips because they can bring their data from all these marketing tools into one place and actually tell a data story.
When it comes down to it, marketers are not actually data analysts (I know, shocker). Some of us can hold our own, for sure, but data analysis takes a specialized skill set. In parallel to the explosive growth marketing technology, organizations have increased their investments in their business data by building out data teams.
I remember our first hire on the operations/data team at Klipfolio. It was like giving sight to the blind – the data challenges we faced couldn’t be handled just by bringing this data together and making it into dashboards. This problem, in a lot of ways, is somewhat straightforward to solve.
Just a small selection of the marketing technology available today.
Data teams now sit on a mountain of data and marketers are able to request dashboards, reports, or access to self-serve analytics. In theory, this process seems like nirvana. Data teams are able to collect data from all the sources, normalize it, and then ship it back out to the respective tools. Marketers now have a single source for all their data, and data teams have the tooling to give them really actionable insights. This seems like a solved problem, right? Wrong.
In practice, there’s still something missing. Marketers are too often frustrated with the reports they do receive or get lost in the self-service tools; data teams struggle to interpret what the marketing team actually, really wants from their report (this time).
The Quest for a Single Source of Truth
The problem with the quest for a single source of truth is that we always start with technology. There is no technology that can solve this challenge for us. Even the coolest, most capable technology will fall short.
Here’s why.
The single source of truth is not a tool but the data itself.
While we’ve solved a lot of really hard problems with data, such as data ingestion, storage, normalization, and manipulation – we’re still struggling with the meaning of the data. The semantics are off. This explains why the conversations between data teams and marketers don’t always go according to plan.
Since the early days of Klipfolio, we’ve helped businesses with their data. Time and again, the teams that succeed with data have a rock solid understanding of their metrics. This isn’t to say things aren’t in flux or there are uncertainties – it’s just that the common understanding of what certain metrics mean and how they are relevant is critical to unlocking the value of data.
In other words, a data-driven culture that socializes metrics and a common understanding of what they mean.
The technology is almost an afterthought.
On one side, we have an entirely human problem: how to define and stick to our metrics so the data we collect increases in value over time. And, on the other side, we have a technological problem: how do we get data from all these technologies to uniformly work together.
Challenges in Data Uniformity
There are 11,000+ marketing technology vendors and it’s not a leap to think there are more yet-to-be-discovered or yet-to-be-launched vendors in the next landscape. The opportunity provided by all this marketing technology is almost equal to the challenges it creates.
For starters, every martech tool comes with some reporting capabilities and therefore creates a data artifact. Each artifact, however, isn’t necessarily uniform. Based on my experience, I’d say, as a rule, data from different systems is rarely uniform.
On top of this, there is no standard for marketing technology. Every business and marketing team uses a different tech stack based on their unique needs. Fair enough, but it means that the data and metrics produced by these 11,000+ vendors don’t automatically work together.
Even doing something like an Excel lookup between a web analytics tool and an advertising tool can be mind-numbingly difficult. But when you look at data uniformity as a whole within martech, you can see why it's such a problem for data teams.
For a technological solution to exist, it needs to solve the problems of data uniformity in martech. Here’s just a glimpse into the types of challenges that exist.
Data uniformity is a challenge in marketing technology adoption.
Interoperability and Scalability
One of the greatest challenges in a data environment is the interoperability between different measures. As an example, temporal issues such as how different systems track date and time can create headaches for data normalization; but, what happens when a metric doesn’t have any date or time information?
It’s not uncommon to have a 3rd party service only update a raw value, such as Twitter/X with the number of followers metric. From a technology perspective, we actually have some inferred information about date/time – for instance, we know when we last queried the API and could use that as our time series.
This small detail throws a big wrench into our plans to scale self-service analytics to the marketing team. Without complete information – arguably, the metadata of a metric is part of that info packet – then it’s impossible to do apples-to-apples comparisons. This is critical when trying to make data-driven decisions about your business. Context is king – and if your metrics can’t communicate and be compared, then your data system is fragile, a far cry from a single source of truth.
Scalability of your analytics, then, has a lot to do with the interoperability of the underlying metrics and data.
The Social Contract Between Teams
What’s so often missing in the conversation between data teams and marketing teams is a common understanding of the meaning of metrics. I don’t just mean the initial definition – but the process of continuously monitoring, updating, and reviewing key metrics.
Both teams want the same thing: a codified library of business metrics that everyone has agreed to. The reality on the ground for both teams is more difficult than that.
First, marketers are constantly tweaking our tech stack – introducing new tools, removing old ones, and adding in new processes. Along with that comes a continuous change in our methodology. In 2014, growth hacking was still in vogue while now in 2023 it’s more about product-led growth. Our metric definitions, at a minimum, have been heavily modified to keep up with times, tech, and methods.
For data teams, the challenge of getting the data into a central location is often vigorous enough that they’d rather do it once and be done. Revisions, while expected and accepted, are exceptions to the process, rather than the rule. Data, by its definition, should be immutable and unchanged – so tweaks violate some basic principles.
The disconnect gets played out in the analytics layer where marketers request metrics, data teams provide them, and both sides wonder: is this what I wanted/is this what I asked for?
It’s a tricky problem – marketing wants to play it fast and loose with something that, by its very definition, is meant to be immutable.
Both teams are right.
The solution is a social contract – and, no, not another weekly recurring meeting. A social contract enshrined in the data itself.
The Metrics Layer Solution
In the realm of data management, the metrics layer serves as a contextual framework that organizes and gives meaning to raw data. It acts as a semantic layer, a sort of "metadata wrapper," that makes data more actionable and interpretable. Just as you can't simply throw text and images onto a screen and expect them to make sense to everyone, raw data also requires a certain structure and contextualization to become useful information.
If you’re familiar with HTML, then you may already understand the idea of semantic markup. The visible content on this page such as the images, headings, and font is the data. It’s raw and requires interpretation by the browser to properly render on your screen.
HTML adds a metadata to the content on-page and uses markup to add meaning. A Heading 1, like the one used at the top of this blog post, receives a certain style but also expresses important information for accessibility. What’s more likely to have the page title in it: A Heading 1 or a Heading 3?
This type of metadata is incredibly useful in HTML and web development, and the same concept applies to data. In a web page, you can change the content on-page or the markup – and the page itself could remain at the same URL location. One could argue you’ve changed the shape of the page, while the underlying data is still the same – if the page is a metric, it still exists.
HTML adds the semantic meaning to the content to help browsers correctly render content on-page.
This problem exists in a different form between data teams and marketing. Data teams advocate for immutable metric definitions because they want to preserve insights for future analysis. Changing the metric definition throws the data out-of-whack and gives conflicting signals that drown out insight.
The marketing teams need to change with the times – marketing is ultra competitive and the reality on the ground is changing every day. Definitions do need to be tweaked and revised to reflect business needs.
The meeting point is the metrics framework. The metrics framework is, in a sense, detached from the underlying data. The data fed into the model is only one part of the metadata associated with that metric. The data source itself could change but the metric would remain exactly the same.
The thing that trips up both teams is the question: what happens when the underlying shape of the data changes? For instance, you change your MQL model or your move from one marketing automation tool to another.
The metrics framework is opinionated in the sense that the metric, by design, tracks not only the data fed into the model but the shape of the data itself. While the data source and data collection mechanisms may change, the metric itself continues to do its job – sometimes, that’s reporting volatility in the number.
While at Klipfolio we saw that bringing together raw data at a dashboard is solved with our Klips product, the “social contract” between data and business teams about the meaning of data and a single source of (metrics) truth was not. This is why we built PowerMetrics.
The final challenge is getting those pesky definitions on the table and discussed. Ok, this will likely require a few meetings. Defining these metrics, why they’re important to your business, and then using technology to enshrine that agreement is the path forward. The social contract that needs to exist between data teams and Martech teams starts with a common understanding of business goals, departmental goals, and then data needs.