We’ve been working with companies to get their digital marketing setup and measure their efforts for over 10 years. After a while, certain patterns start to emerge. In this post, we would like to walk you through 5 distinct stages that we see most companies go through on their journey towards “marketing data maturity”. Hopefully, this post helps you reflect on where you are as a company right now and inspire you to think about the next step.
Let’s dive in:
Phase 1: A Tool per Task
Let’s get the ball rolling!
Getting from nothing to something is the hardest part for any business. It’s so tempting to get stuck in perfectionism or analysis paralysis before you even have any interaction with your potential prospects or existing customers. Getting the ball rolling is always step 1.
From a tooling and marketing technology perspective, this usually means that every marketer gets carte blanche, as long as the job gets done. A default setup usually includes:
- Google Tag Manager
- Google Analytics
- Conversion Pixels for Google Ads and Facebook
80% of companies don’t make it out of phase 1. And that’s to be expected. For some, digital marketing simply does not require any more effort (or simply isn’t worth it), and others don’t survive long enough to evolve from this phase.
Phase 2: Aggregate Reporting
We need some oversight!
Once a companies’ digital marketing efforts grow, inevitably a digital marketing manager gets promoted or recruited. Their first order of business: getting a grip on what’s happening out there.
Usually, this means creating some huge Excel Sheet where every marketeer is asked to dump their numbers in every week.
Their second order of business: hiring somebody to build a dashboard to aggregate all data from all tools implemented in phase 1 to make sense of what’s happening.
From a tooling perspective, this is usually where you see tools like Supermetrics (affiliate link) and Stitchdata show up. Using it to export everything into Excel or Google Sheets and then visualizing a Data Studio dashboard on top of that.
But, exporting data from tools is a dirty and error-prone job. Soon enough, the organization will look for ways to automate and structure this process.
Phase 3: Birth of the Data Lake
Reporting is taking too much time!
A common move we see companies make is to build a Data Lake. (Often, they don’t call it that, it’s just born out of necessity.) They figure out how to export the data from all their tools, preferably automated, and get it all into a database. In recent years, Google BigQuery seems to be a logical choice, mostly because Google Analytics 4 and Google Ads offer native export features.
Getting all the data in one place and automating the extraction part is an essential step in the journey towards data maturity. It reduces the risk of human errors and sets up the infrastructure for what’s to come in later phases.
Once the data is in BigQuery, building a report on top of that data with Google Datastudio is a breeze. Especially when you enable the BigQuery BI Engine to speed things up.
But it won’t take long until new problems arise on the horizon…
Phase 4: Birth of the Data Warehouse
These datasets all say something else!
Your clever marketers will have many questions. One of those questions will be to explain why Facebook Ads and Google Ads are both claiming 500 conversions and the total amount of conversions in your backend is only 800.
It’s time to take control of your data and to stop relying on third parties telling you what the truth is. It’s time to start connecting your datasets and re-creating your own logic with data models on top them within your data warehouse.
The distinction between a data-lake and a data-warehouse, (for this blog post), is that in a data-lake you simply store the raw exports from all your tools, whereas in a data-warehouse you transform the data into your own desired datasets that can be used within the company.
You will create a User Identity dataset that matches user identification keys from all datasets and allows you to stitch data across per user.
You will create advertising cost datasets that match campaign parameters across different advertising sets to measure return on investment based on campaigns (regardless of sources).
You will create your own attribution logic that takes into account the entire customer journey across multiple advertising sources and platforms to calculate your true return on ad spend.
To be fair, phase 4 is a long and probably never-ending phase. Tools will change, new data will be collected, datasets will be altered and models will be adjusted. By the time you reach Phase 4, you’ll likely have at least one full-time data engineer whose sole task is to maintain the marketing data warehouse.
From a software point of view, DBT is a really popular choice to manage the data transformations and models you’re applying.
Phase 5: The activation of Data
Feed the Beasts!
Once Phase 4 is completed, the insights truly start to pour in. So much so, that your marketers will become the bottleneck for improving your results. Deploying all learnings from your marketing data and optimizing accordingly by manually adjusting campaigns will not cut it anymore. It’s time to automatically feed your data into your tools and algorithms.
It’s time, to feed the beasts.
Phase 5 is where the cutting edge of marketing technology is taking place right now. Feeding high-quality data about your customers and their activity on your platforms, in a GDPR compliant way, to the algorithms of your email marketing platform, your app notification platform, and your advertising suites. Preferably as real-time and up-to-date as possible to be as relevant and timely as possible.
Until a couple of years ago, this meant you’d have to build your own systems. Currently, there are tools popping up in the “Reverse ETL” space that try to fill this gap with SaaS solutions. (Hightouch and Cencus are two top competitors in the space right now.)
If you’re lucky enough to work in an environment that operates in phase 5, count your blessings. This is the top 5% of digital marketing organizations.
In Closing
In the end, like the name of our company suggests, it’s all about getting value out of your data. Regardless of whether you’re working within a “Phase 1” or “Phase 5” organization, if you’re not actively using your data to make changes to your work, you’re not having any impact and you’re not generating any value. You can not increase value without taking action.
The feedback loop needs to be closed, and it can be closed by a lot of manual work and human interpretation (Phase 1) or by a lot of technology and automation (Phase 5). The main difference is going to be the speed and scale, but those only matter once you’ve found the right fit.
Hopefully, this has inspired you to take a step back and evaluate where you’re at in this model and how you might take a step towards the next phase. Like always, if you want to brainstorm on how this could work for your organization, feel free to plan a video call with us using the link below and we’re happy to meet and help out.