How a customer data platform uses data for decision making

The maxim ‘knowledge is power’ has held good for many years but that power also can be democratised when knowledge is shared. Shared knowledge obtained from data, leads to a common understanding of how a business is working, and from this a more effective, faster and smoother decision making process. No more meetings starting with that obstructive statement ‘I don’t recognise the numbers’.

 

So, how does a CDP use data to enhance, accelerate, and democratise decision making?

Several reasons combine to contribute to the result:

1. A CDP will build peoples’ identities using data coming from many different sources.
This means that individuals may be matched on their postal address, their email, their mobile, their cookie ID and more. The identifier, we call it the permanent URN or PURN, for an individual will be fixed in the system and any new data that matches to it added to the customer record. From this we achieve a finite, and at any moment in time, fixed number of known individuals in the system. We can no longer disagree over customer volumes or the data that is attached to them.

2. All the data that arrives goes into a shared data schema.
This means that the data definitions are constant as well as the values within them. A customer data platform can have only one set of values for gender or marital status for instance.

3. The CDP does not dispose of data unless required to do so under GDPR or because it has no conceivable value.
When a CDP is being set up historic customer data is loaded into the system, and from that point on data will be augmented. This data usually includes details of on-line activity, when it can be matched, as well as transactions undertaken, and contacts made through any channel both inwards and outwards.

4. A good CDP will check the data feeds as they arrive for consistency of their layout and content.
Feeds that do not conform are rejected before they go into the customer data platform. In some cases, such as when we are dealing with names and addresses, they may need to be improved using matches to external verification files like PAF.

5. A smart CDP will go far beyond just recording the input data.
It will start to build what we call engineered data. This is data derived from the raw data that comes in to the CDP. It may be a calculation of lifetime value in the first year since acquisition, or the result of applying a customer segmentation or a propensity model. Engineered data fields are then updated each time new data arrives for an individual. It is often the case that charts and reports are built from the engineered data rather than from the raw data. For instance, to answer a question like what is our customer retention rate we need to have an engineered data field that marks whether a customer who purchased in period A also goes on to purchase in period B, the periods being linked to when the individual first appeared as a customer.

6. The CDP will allow access to all users in a company who want to see that single common view of the customer.
This may be by way of dashboards or on-line reports, or because they have the tools to extract data directly from the system. The dashboards may be tailored to each individual user or be common to a department or a whole organisation. But whatever charts or tables they contain they will all be derived from the same common but constantly updated CDP. This means that they cannot disagree because both the data they are drawn from is common, and the definitions that they use are shared.

 

And what types of understanding can be obtained from the CDP?

Having developed a shared, consistent, and complete customer data source, the CDP can be put to support a multitude of uses. What follows is a description of some of the more common ones, but in reality, there is no limit on the kinds of outputs that can be achieved, particularly when the CDP is aligned with a clever data visualisation tool like Tableau.

1. Supporting AI.
Underneath all the hype about AI is the need for good data to support it. If an AI tool is to learn from a flow of data being fed into it, that flow needs to be unchanging in its composition, and accurate to the point of perfection. A CDP is perfectly designed for this role.

2. Tracking the overall customer picture.
Managers looking after customer acquisition and retention will have an almost unlimited number of questions they need to have answered. Some of the most typical are:

  • What volumes of customers are we recruiting through which channel?
  • Do the different channels and media sources provide customers with different characteristics and different longer-term values?
  • How do the different social media channels contribute to my customer volumes?
  • From which geographies am I recruiting my customers?
  • What is my level of second orders and year on year customer retention?
  • What behavioural segments do my customers fall into, and what do they look like when profiled by demographics and lifestyle?
  • Have the characteristics of my customers altered over time?
  • Are there some customer groups that provide such low value that they are not worth the cost of recruitment?

3. Tracking the product picture.
Every business knows the high-level numbers on product sales but beneath this high-level view are many interesting questions, for example:

  • Are different types of customer changing the mix of products they buy?
  • Do different channels lead to different products being purchased, and to different product values being chosen?
  • What is our best recruitment product?
  • How loyal are customers that are first attracted to different products?
  • Given that we know the first product purchased by a customer, can we predict what a customer is most likely to buy next?
  • If we deduct the cost of marketing and cost of goods, how profitable are our different products?
  • Can we build a time series model to predict overall sales of my products based on a combination of customer information, and marketing history, combined with external factors?

4. Next up, understanding the impact of marketing.
A strong case can be made for the assertion that without a reliable CDP the value of marketing will always be an unknown. For instance, if you cannot attribute sales that come in via the internet to prior marketing communications how can you justify the cost of those communications? Or if the longer-term value of customers cannot be measured how can you justify the cost of recruiting them beyond the value of their initial purchase? Some of the many uses of the CDP for marketing have been detailed above under customer and product understanding, but there is an additional layer which comes from being able to link marketing activity in its many guises to an outcome in terms of customer value. Some examples are:

  • Linking expenditure on social media to actual customers recruited and the value they subsequently generate. Very few companies spending fortunes on social media take the trouble to do this, but with the right digital analytics combined with the CDP this can be quite straightforward.
  • Examining campaign results not just from the overall value obtained, but also investigating what types of customers they attracted, and what different longer term values they brought with them.
  • Providing the data to help understand a complex sales funnel where prospects drop out at different stages; in these cases, there is a need not only to understand the impact of the different processes on drop-out levels, but also what types of prospects are better at surviving the overall funnel process.
  • If there is a customer retention problem, then this can be evaluated in terms of lost revenue, and from that budgets provided for retention activities that can then be tracked, using control groups held in the CDP, to find out which provide the best ROI.
  • When a marketing budget is complex and spread over many different activities then there is a need to untangle the impact of the different marketing activities and prioritise the way budget is allocated to optimise results. This requires building an understanding of the relationship between spend in a channel and the usually diminishing results achieved as spend is increased. This is often called a saturation curve. For this kind of analytical activity the underlying data is found in the CDP, although the subsequent analysis is done outside it with different tool-sets.

 

To wrap it up

It would be presumptuous to claim that a CDP can solve all marketing problems, but without doubt it can ensure that knowledge is shared, and the data required for intelligent decision making made readily available. A CDP can be perhaps viewed as the necessary foundation stone for all successful marketing organisations.

 

Contact us if you’d like to know how we can help your organisation.


UniFida logo

UniFida is the trading name of Marketing Planning Services Ltd, a London based technology and data science company set up in 2014. Our overall aim is to help organisations build more customer value at less marketing cost.

Our technology focus has been to develop UniFida. Our data science business comes both from existing users of UniFida, and from clients looking to us to solve their more complex data related marketing questions.

Marketing is changing at an explosive speed, and our ambition is to help our clients stay empowered and ahead in this challenging environment.