This is part 2 of the series titled Data Activation Is Not The Goal. Here's part 1.
I ran a poll on LinkedIn asking folks what Data Activation meant to them…
And I was glad that a majority of people agreed that Data Activation is equivalent to running data-powered campaigns and not just running reverse ETL jobs or building and syncing audiences. It’s worth noting that “campaigns” is a catch-all term that also includes growth experiments.
If you disagree, I want to share a perspective that might change your mind. And even if it doesn't, it might give you something to think about.
This quick guide aims to provide a robust overview of the activities that lead to Data Activation, and the activities that follow. Please note that this will only make sense to you if you’re familiar with the basics of data activation in the context of the modern data stack – if you’re not, I encourage you to begin here.
And if you’re ready, let’s jump in.
Here's how I define Data Activation:
Data Activation refers to the act of running data-powered campaigns and experiments to deliver personalization.
This implies two things:
- Before data is activated, multiple teams need to collaborate to collect, model, and analyze data.
- And after data is activated, teams must collect more data, build more models, and run more analyses.
Data Activation, therefore, takes place only after the foundational infrastructure has been set up which entails a series of activities, none of which amounts to “activating” data.
Whereas, once data activation is set in motion, well, that’s when growth and data teams can begin to work their magic by collaboratively building the most efficient experiences powered by the freshest data
In simple terms, a lot needs to be done before and after data is activated to deliver personalization that has a measurable impact on the business – personalization that drives growth.
Let's dig deeper into the activities that lead to Data Activation, and the activities that follow.
For the lack of better terminology to describe everything that happens before and after data is activated, I’ve taken the liberty to come up with the following:
This stage includes all the activities that need to be performed before data can be activated – activities that aren’t equivalent to “data activation”:
- Collecting data from all relevant sources
- Defining and modeling the core metrics
- Performing analyses
- Building segments or audiences
- Syncing data downstream where it’s activated
These are some of the main activities that teams need to collaboratively perform, using a host of tools, to enable data activation. Moreover, this is a broad categorization, and depending on the use case, there are many other things – like enrichment and identity resolution – that the data might need before it’s activated.
Collect the right data or just activate what's available
I want to highlight that data collection is a very important step in the pre-activation stage and teams cannot always rely on the data that’s already in the warehouse.
The availability of data is often taken for granted as any organization looking to activate data ought to be collecting data. However, organizations are unlikely to have the right data already collected – waiting to be activated.
Unless growth teams know what data they need in order to set up activation workflows, and unless they collaborate with the data team to ensure that the right data is collected and the required models are created, the data that’s already available might not even be usable.
Additionally, before activating data, growth teams also need to list down the problems they’re looking to solve – problems like:
- What are the points of friction preventing user activation? (Getting users to the aha moment; not to be confused with “data activation”)
- What’s preventing active users from upgrading? Is there a feature gap or is it a pricing challenge?
- What’s causing active customers to downgrade? Has usage reduced or are there external factors at play?
To narrow down the problems that data activation can potentially solve, growth teams need to derive some insights, and in some cases, those insights might only be derived from data that’s not already available, leading to more data collection and modeling efforts.
Depending on the people involved, the culture, and the use cases, organizations can end up spending a ton of resources on these activities, sometimes even forgetting why they decided to prioritize those activities in the first place.
It’s not so uncommon for organizations to invest in a plethora of pre-activation activities and the necessary tools, of course, and never really get to activating data in a meaningful manner.
“Set it and forget it” – beyond marketing-speak – is just bad advice.
The goal, after all, is to go from activating data to driving measurable growth for the business.
And without iterating, failing, learning, and measuring outcomes, can we even drive sustainable growth in a difficult market with buyers who have high expectations?
This is where a data practitioner steps in, ideally someone who likes to think like a growth person – someone who cares about the impact of their work and wants to make measurable contributions to business growth.
External tools that power activation workflows produce a ton of valuable data, and it takes a really motivated data person to quickly extract data from those tools and combine it with first-party behavioral data to build performance reports, enabling the growth team to measure the impact of their campaigns and experiments on user engagement, activation, conversion, retention, and churn.
And for B2B SaaS businesses that need to report on accounts instead of users, that data person needs a double dose of motivation because their job just got a lot harder.
As a former growth person who also took charge of the pre-activation stage, I know how great it feels to set activation workflows in motion. Growth people thrive on building campaigns and experiments aimed at delivering personalization.
This is when they get to test ideas, iterate, fail, learn, improve, and eventually prove that growth is possible when good data is made available in the right shape in the right tools.
But it’s also easy to get carried away – and for a minute, forget that “activating data” is a job half-done.
Just like producing content at a dizzying pace and publishing it once is only 50% of a content creator’s job (distributing, updating, and redistributing the content being the other 50%), running data-powered campaigns and experiments is, after all, a means to an end.
The ultimate goal is to go from activating data to driving measurable growth for the business.
Lastly, and I think we can all agree here, activating data to deliver personalization for the sake of delivering personalization is no good. Brands must invest in finding out if their personalization efforts lead to positive outcomes (more usage, more, more revenue, higher retention, etc), neutral outcomes (no real impact on growth), or maybe even negative outcomes (annoyed or creeped out customers).
In the final part of this series, I'll dig deeper into what comes after the successful implementation of data activation workflows.
The section below contains some material on CDPs which are essentially the tools that enable Data Activation.
It seems like the Modern Data Stack is being rebranded as the Composable CDP by the folks at a16z. In June 2023, they published a post that received a lot of strong reactions from a lot of people. As someone who is fiercely passionate about the space, I had to share a few words in the comments.
Another industry stalwart, Jonathan Mendez penned down a quick piece as a follow-up to the a16z post. I thought I disagreed with some of his points until we caught up and he helped me understand his point of view better.
We both are actually on the exact same page and believe that the definition of CDP is being stretched a little too much and there’s no end in sight to the confusion that has ensued.
The only sane way forward is for the industry to agree upon a new term altogether, one that accurately describes the data platform built on top of the cloud.
Moreover “services” or “apps” shouldn’t pose as “platforms”.
A composable customer data platform can only be assembled using various services or warehouse-native apps.
What remains debatable is whether the cloud data warehouse by itself is the customer data platform or does it only become a platform once multiple warehouse-native apps are layered on top of it.
Now, can a CDP be called a CDP without behavioral data?
A common argument is that you already have a data warehouse so you need not invest in yet another customer data platform. The argument has merit but holds true only if the warehouse actually has the data that non-data teams need to elevate the end-user experience.
And behavioral data or event data is key for GTM (go-to-market) teams to understand user behavior and nudge the user in the right direction via event-based emails and in-app messages.
So, what do you think?
Can a CDP (Composable or Packaged) be successful without behavioral data? An audience member had asked this question during the CDP Rapid Fire but this 4-minute clip wasn’t included in either of the published episodes.