One of the smallest things you can learn that has one of the biggest impacts is knowing the level of granularity you need in your data to draw out the insight you need for decision making.

Too often I see marketers at conferences talk about the need for a “single customer view”.

It’s often spoken about as this utopia. As if marketers can log into a platform, whether their CRM, CDP or DSP, search for a contact or a group of attributes and see immediately the move they’ll need to make and then by extension, be able to activate on that insight at the click of a button.

This is of course, what many tech vendors sell to us as the dream so it’s easy to understand why this is mentioned so frequently at networking events… 

Wouldn’t it be nice if everything were really that simple?

The truth is, we don’t need a single customer identity to execute all of the important use cases we need. Sometimes we only need aggregate, anonymous, big data groups. Sometimes we need a link to channel source (attribution). Sometimes we need user level but not identity level.

If I had a dollar for every image like this I’ve seen over the years… Thanks zeotap for the fantastic example.

It’s great news really that you DON’T always need this. At least for insight. Of course if we are talking insight and activation – you will but instead of collecting everything for everything’s sake (especially in an increasingly privacy focused world) we should be relying on insight first to guide what we collect, knowing there is opportunity and a value proposition in it for our customers. 

So with that, you don’t always need a single customer view

The trick is in knowing the insight you you are looking for, and then knowing how to get it.

Marketing analysts hone this understanding over years of experience and exposure to a multitude of marketing problems. Finding clever and creative ways to connect them to analytics solutions.

Image item

It’s a skill that can be taught though, and it starts with understanding the types of data we can collect and the kind of analysis that data can unlock. 

The first step though is to establish your use cases.  

Once this is done, they can be ranked from easy to hard to implement depending on the data you have and can enrich and the data you are yet to collect but have the infrastructure to do so.

So here’s a little guide to build up your understanding of the connection of data types to data opportunities in the key use cases I come across.

Then, you can determine whether those marketing metrics you’re being asked to provide really do need to await a utopia state, or if you might be closer than you think.

Pssssst! Like I always say in our team, even without your desired data there’s almost always a proxy that can be used. If you don’t have exactly what you need, always think about what might be used as an “implicit” data trait. Or perhaps even the “next best indicator”. These can help you unlock value faster than if you were to await the capability to uplift.

While there is some overlap in some of these groupings where types of analysis can fall within more than one group, (they’re not totally MECE), most commonly, we see the following. 

Single Customer View or Single Identity Analytics

From my experience, insights that genuinely need a link to a consolidated identity-based-profile are actually used for very few use cases. 

These include things like:

  • Segmentation (things like RFM)
  • CLV
  • Propensity models
  • Survival or cohort analysis
  • Churn prediction
  • Cross-channel or cross device attribution

Anonymous user level analytics

Then there are insights that help us make decisions on a user level (without the need for a consolidated identity). This includes things like:

  • Website page path
  • Funnel and fallout analysis
  • Heat and click maps
  • A/B testing

Aggregate anonymous data

  • Trends in content and products
  • Website event health metrics 
  • Media mix modelling
  • Market research and competitive analysis
  • Customer satisfaction and NPS

Now I know what you’re thinking.

“My analytics platform gives me a bunch of these? Why do I need to know what data type is used for what?”

Here’s where it can come in handy.

This week I’ve been working on an attribution methodology for a content platform.

We need a level of granularity in data to unlock the ability to run a media mix model, but those familiar with media mix modelling know, we only need this data as an aggregate anonymous snapshot as we allow the power of the regression analysis to do the rest of the heavy lifting on attribution.

So NOT needing this data at a user or identity level will save them a boatload of money. This is because analytics and data platforms usually charge per event. Instead, sending data directly into a warehouse only comes with a small storage fee.

I know what you might be thinking “But my GA4 instance is free?” Well, is it though? With cardinality and thresholding challenges, it isn’t always easy to get the data you need and Google itself still has limits.

Knowing your data types and what you can do with them can be a fundamental building block on literacy that can make you a dangerous marketer – enabling you to empower others with the knowledge of what’s possible.

Hi I'm Kate! I'm relentlessly curious about the attribution and origin of things. Especially as it relates to being a corporate girly balancing ambition and a life filled with joy.

Leave a Reply

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