With regrets to The Smiths and one of the great B-sides of all-time, CDPs are the son and the heir of a Martech shyness that is criminally vulgar.
Why are some brands shy about CDPs? From what I’ve seen in the market there are three primary blockers.
- Businesses are not mature enough with data
- Businesses have Martech SaaS fatigue
- Businesses have issues with talent and organizational design
For some it’s a combination of all three. I’m not trying to minimize these issues. They are real. But when “now” comes, and it will, companies must ensure a most basic requirement is met before moving forward.
The data going into the CDP needs to be high resolution, high quality data for the CDP to work.
I’m not sure how soon is now but if you’re like most, your present state of data has issues. Those data issues need to be addressed before you start plugging in APIs all over your databases. To get high quality outputs from analytics and machine learning you are going to need high quality inputs. CDPs are no different than any other data driven system. Garbage in. Garbage out.
But where are the most important places to start ensuring data quality in order to maximize the value of your CDP? Let’s look at the three pillars of data collection most to start cleaning and prepping before you hook up with a CDP.
Customer Events: This is a core component of the data collection because this is the fabled data on “customer journey.” This data comes from a number of sources. For many site analytics is the primary source of this data. It also can include app data and offline touch-point and transaction data. For most businesses the issues are around site data. Many brands have site analytics installations almost a decade old with a myriad of data quality issues. There’s always been a robust services industry around site analytics and this is not an accident. Proper installations and accurate tagging is incredibly challenging as pages, products, links and the site itself has changed over time. Go ask your site team if there are any redirect challenges or if the directory structure is high quality. Then ask who on the marketing team is responsible for ensure site analytics data is accurate. I thought so…
Customer File: The holy grail of your business. But how often is it scrubbed and washed? When was the last time it was prepped? Who is responsible for ensuring accuracy? How often do you validate the tables? Hopefully you have taken great care with this file but as dimensions have been added over time its imperative to make sure it has been properly formatted. Are there missing values? Some algorithms don’t like that. Getting a customer file ready and using it for merge or modeling and prediction are very different than using it for segmented catalog drops. Your customer file should be well processed, cleaned and prepared for ETL into the CDP. Many are not.
Channel Data: Like our own, the universe of marketing channels has many black holes. The largest channels like Google and Facebook continue to share less and less data while you share more and more with them. Developing a strategy around channel data collection, often including a few hacks, is an imperative. Often time this can be done with macros and pixels. I see too many brands who don’t have any real owned data or rely on agencies for channel data, thus they are unable to ensure data quality. So much of customer acquisition occurs through channels so it is key. This extends not only to visit data but also to bid data and data about creative (often forgotten). Channel data quality directly relates to the quality of your attribution models and your LTV, both key reasons to have a CDP in the first place.
Let’s consider it a forgone conclusion that every brand will have a CDP in 5 years. Like any major advance in data technology there are advantages to having that technology in place before everyone else. So what are the advantages to getting started now?
Moving forward now with a CDP establishes people, process and technology to give you added advantage in the competitive data landscape that has emerged. There are first-mover advantages with the time, attention and effort that will be invested, especially on the back-end.
Data modeling is hard work. These models are not going to be out-of-the box (though some CDPs use that as a selling point I wouldn’t expect these to work well). Every business is different. Every customer base too. Isn’t that differentiation the point of your brand promise? You will need time to tune your CDP for high value outputs. The more optimized you are ahead of your competition the longer lead you have to optimize customer acquisition and retention loop metrics. This is advantageous. This is the “why now.”
So even if you are not ready to start the process of CDP buy/build or vendor review or any of the myriads of other decisions and trade-offs that come with standing-up any large-scale transformative technology, there is plenty of work you can start doing today. Don’t go about things the wrong way!