How to build the best martech stack for data-driven marketing: A guide

How to build the best martech stack for data-driven marketing: A guide Tido Carriero is chief product development officer at Segment.


20 years ago, when Salesforce launched its SaaS CRM platform, times were simpler. The majority of customer relationships were managed by a single sales team who manually added their (copious) customer notes to the CRM. 

We’ve since evolved to a multi-channel world, where customers engage with brands across everything from mobile to social media. The sales team’s input into a CRM platform no longer captures the whole customer relationship and doesn’t give marketers what they need to deliver a tailored, thoughtful experience.

Even so, marketers have done what they can with CRM as their data foundation, patching tools together to manage the customer experience – but this has inadvertently created a tangle of data integrations that now make it harder to add and remove tools and adjust the marketing stack as needed.

To deliver the sophisticated, personalised interactions that customers now expect, a new approach is needed—and it starts at the foundations.

A tale of two system designs

If you look under the hood of any digital-first consumer service like Amazon, Spotify, or Twitter, you’ll find a client-server architecture

Here’s what you need to know. A client-server architecture stores all information about customer interactions on a central server, which then enables each device (be it a web browser, an Android phone, an iPad, etc.) to connect and get information about a specific customer. With a client-server architecture, the server acts as the “brain” that talks to every single client, or node, to make sure it knows the right action to take. 

But your own business is probably built on quite different foundations – a peer-to-peer architecture. This requires that all devices talk to each other to get updates about the current state of the world. That means all connected services must engage with each other when managing and acting on customer interactions. 

Not so bad, right? Wrong. Using a peer-to-peer infrastructure means that even a simple instruction – like  “don’t send a marketing email to the same customer twice within a week” – is comparable to a game of Telephone, over-complicating a simple command.

Losing sight of your customers

Unfortunately, most legacy businesses (including most of the FTSE 100) have evolved a peer-to-peer architecture for managing their customer relationships – exactly the wrong data infrastructure for the job. 

How did this happen? From my experience speaking with CIO and CTOs, three themes come up again and again:

  • Quick fixes. “When we started building our data architecture, we needed a quick solution for a short-term goal"
  • Inconsistent ownership. “We’ve had a revolving door of teams working on this problem with no long-term ownership”
  • New customer channels appeared over time. “When we started, we didn’t realise the long-term impact that this data architecture could have on customers’ experience on the shop floor, as email was the only way we communicated at the time”

Building the right foundations

The good news is that, if your company builds its data architecture in the right way, you need never encounter the problems of a peer-to-peer architecture. 

Take your cue from the new breed of digitally-native, direct-to-consumer companies like Peloton, Amazon and Netflix. These companies built their customer data infrastructure in a very different way, knowing that the secret of any data-driven business lies in its foundations. 

These companies have disrupted major, established lines of business by focusing first and foremost on an incredible customer experience across every channel. In short, they have invested heavily in a central “brain” that understands the customer, and from which marketers can deliver exceptional, tailored experiences. 

Here are three ways you can clean up your mess and follow their lead:

  • Create a system of record for each customer. The simplest way to do this is to get everything into a single data warehouse or data lake, where it can be joined across a customer identifier. You’ll also need to put summary information into a data store that can be accessed in real-time, so you can get the promised 360-degree view of the customer whenever you need it
     
  • Treat your data infrastructure as the business-critical system it is. Sure, they’re not as exciting as the marketing tools you connect them to, but your customer data pipelines are essential – and that means they should be managed like any other business-critical system, with the care and attention to match
     
  • Balance flexibility and standardisation. As a marketer, you’ll want to choose the customer interactions you monitor and build your own stack. At the same time, it’s important that some things are standardised for all business units, so that your company shares a solid, centralised understanding of how customers are engaging with your products. 

Once all of this is in place, you’ll be free to build the marketing stack that suits your needs, with tools you can integrate with your first-party customer data at the flip of a switch. As your focus evolves, you’ll be free to switch them out for new choices in a matter of minutes. 

The wise marketer builds his house upon the rock

If you’re stuck with a peer-to-peer architecture, your technology stack needs to evolve. Otherwise, it’s only holding your marketing back. 

Ask your tech team to look at the customer data architecture you have today and strategise how to move forward with you. Try whiteboarding out every system that touches customer data and the customer experience. If it starts to look like a huge tangle of pipelines, then it’s probably trapping your company’s potential – and the only way to solve it is to go back to square one.

Interested in hearing leading global brands discuss subjects like this in person?

Find out more about Digital Marketing World Forum (#DMWF) Europe, London, North America, and Singapore.  

View Comments
Leave a comment

Leave a Reply

Your email address will not be published. Required fields are marked *