The issue of customer data is becoming increasingly important in discussions between experts, in studies conducted by certain institutions, and even in company offices. In most of the studies I read about the challenges of marketers, those related to the use and understanding of customer data are always at the top of the list. We are experiencing this with our clients, who are facing increasing challenges and have important questions for the future. As a marketing manager myself, I am constantly asking myself how I can use this rich information … and the challenges on my way are always present.
Why are we so interested in having control over the available information? For many reasons, including:
- The ability to react more quickly to online and offline behavior;
- The willingness to modulate messages, the content that is sent to people;
- The importance of getting a global view, which connects all of a customer’s touchpoints;
- The possibility to communicate in an omnichannel perspective, taking into account the other contexts of the client … thus avoiding working in silos!
In short, all this to better personalize a person’s overall experience with your brand. If personalization is the ultimate goal when it comes to data, it’s hard to achieve it if the foundation is rocky, and doesn’t allow us to start from a solid base.
This base, it relates to a great concept, that of structuring information. In the technical jargon of computing and data, this is called data structure, or data architecture. A recent study, which surveyed over 600 marketers, revealed that the biggest obstacle to personalization is indeed data architecture. In past years, third-party data integration and data quality were cited as the biggest challenges. Now, we can’t deny that the data is there, but the challenge is moving toward how to organize that information to create a successful personalization strategy.
As Amperity’s president has already said:
“Data – from email tools, click tracking, in-store transactions and loyalty program databases to CRM, MDM, EDW, DMP, and other analytics and BI tools – is trapped and compartmentalized. Without a complete picture of your customers’ purchases, preferences, and behaviors, effective personalization is impossible. – Martech Advisor
These silos necessarily lead companies to adopt a piecemeal approach to customization, since data sources are dispersed. Teams have no choice but to make decisions based on incomplete information, based on a tiny portion of the information.
Thus, to personalize well, one must be well structured. For this reason, let’s go back to school together and review a few concepts that will certainly not lighten this article, but are essential to understand in order to do so.
What is a good data architecture?
I have collected a few definitions as stated by some experts:
- It is a common vocabulary that expresses the integrated requirements that ensure that data assets are stored, organized, managed and used in systems to support an organizational strategy. (Dr. Peter Aiken)
- A set of rules, policies and templates that determine the type of data collected and how it is used, processed and stored in a database system”. (Keith D. Foote)
- It’s how to use data efficiently and based on business needs”. (Sven Blumberg, et. al., McKinsey)
- It describes how data is collected, stored, transformed, distributed and consumed. IT is also responsible for setting up rules governing the formats on databases and file systems, as well as the systems for connecting the data to the business process that consumes it”. (DalleMule and Davenport, Harvard Business Review)
- These are models, policies, rules or standards that govern what data is collected, and how it is stored, organized and used in a database system and/or organization”. (Business Dictionary)
If we look at the commonalities of all these definitions, it is important to remember that data structure is NOT just a technical concept that governs how data is transferred from point A to point B, but also a reflection that is part of the organization’s strategy. It forces members who need this information to :
- Know what data they need to retain, transfer, modify, etc.
- How it will be used, which has an impact on storage processes for example.
- Who will need to use it, which has an impact on accessibility.
Once this information is understood, we can then move on to data path mapping. We need to make these flow tangible, in order to better understand them and be able to make decisions about them.
Overall, this is a process that must be systematized, and we must not lose sight of it. It is therefore not a task that is done at the beginning of the year, or a few times to meet certain needs. If we want to ensure this continuity in the processing of an organization’s data, we need to ensure its governance. That’s why we’re talking more and more about data governance … even if it’s not a new concept at all!
Data governance is the process that manages the availability, usability, integrity and security of data in enterprise systems. These 4 elements are based on internal standards and policies that also control the use of data. Thus, where the data structure aims to map flows, governance ensures an ethical use of the data.
Indeed, effective governance ensures that data is consistent and reliable and that it is not misused. It is increasingly essential as companies face new regulations regarding the confidentiality of personal information. This is an issue that has really become important with the implementation of the GDPR, and the soon to be in Quebec with the new Law 64 Project. It forces organizations to appoint a data officer, whose identity must be shared.
A data governance program is supported by one to several people, depending on the size and complexity of an organization. A large organization might have a governance team, a steering committee, or technical managers who take care of the technical side of the data. A smaller scale organization may have only one person dedicated to this position. Again, because of the GDPR, we are hearing more and more the term data architect, which has nothing to do with a BI analyst.
What is a data architect?
This person is, metaphorically speaking, the brain behind the data architecture. He or she translates the needs of the different business units into data and requirements that will be applied to the systems. Based on the business needs and objectives, the data architect creates a technology roadmap to achieve the objectives. He creates plans for data flows and processes that store and distribute data from multiple sources to the people who need it.
The Data Architect is therefore the chief collaborator who coordinates departmental stakeholders, business partners, and external suppliers around the organization’s goals to define a data strategy. He must then :
- Define the data vision: moving from business requirements to technical requirements, which become the basis for internal data standards and policies.
- Define the data architecture: including standards for data models, metadata, security, reference data such as product catalogs, and master data such as inventory and suppliers.
- Define data flows: these govern which parts of the organization generate data, which parts use the data, and how data flows are managed humanely and technically.
In the next few years, there will be a real recruitment challenge for this type of profile, as we are looking for a technical individual, able to understand IT jargon, without applying the IT processes that support architecture decisions. He/she must, however, be able to understand the business challenges, and those of managers to make decisions that will address both areas, which often do not agree.
Characteristics of a good data structure
Going back to the main subject, we have already explained that the data structure has a direct impact on the ability of companies to personalize. Indeed, data is the fuel behind personalization, but using the wrong data can derail all personalization efforts. It is therefore essential that data be accurate and relevant to the customer – otherwise the relationship may collapse.
If the impact is that great, then it would be interesting to understand the characteristics of a good data architecture … here are 5!
- User-oriented: In a modern data architecture, we need to think about who will use the information. The architect must therefore create solutions to facilitate accessibility, while respecting security. Business objectives will determine this orientation.
- Built on shared data : An effective data architecture is based on data structures that encourage collaboration. It eliminates silos by combining data from all parts of the organization, as well as external sources if necessary, while also eliminating competing versions of the same data in one place. A Single Customer View like the one offered by Dialog Insight is a good example.
- Is elastic/scalable: This refers to the ability to adapt to the growing needs of the business when it comes to data. This is one of the reasons why companies are moving more and more towards the Cloud, since it takes less time to set up than on-site servers.
- Simple: Simplicity outweighs complexity in an efficient data architecture. You need to aim for direct paths, whether in the movement of data, the tools chosen, the assembly frameworks, etc.
- Secure: Security is intrinsic to a good modern data architecture. It is probably the point that will become the most important in the coming years. This ensures that data is available on a need-to-know basis, as defined by an organization’s data governance. A good data architecture also recognizes existing and emerging threats to data security and ensures compliance with legislation such as HIPAA and the GDPR.
What are the impacts of a good architecture on personalization?
There are many, and some will be specific to an organization, but in general we know that :
- Information is collected more easily, breaking down silos between departments, business units, and even with suppliers/partners. Better quality data = better personalization.
- This creates much more complete customer profiles, which take into account the complex and multiple paths a customer can take. This makes it easier to act on global knowledge, and not just on 1 channel.
- There are fewer organizational silos, since information is transmitted between departments. It is easier to personalize if the necessary information is accessible, such as inventory, logistics, accounting, store visits, purchasing, etc.
- There is better monitoring of regulations and restrictions: by eliminating the risk of using data in the wrong way, marketing teams focus on communicating well, without having to worry on a daily basis if they are breaking any laws.
This is a very quick overview. I will cover the challenges of personalizing in a future article, and I will go into more detail on how to approach them. The solutions, not surprisingly, are often found in the structure of the information.
Now where to start to improve?
The first step to building a good, solid data architecture is to ensure that teams that need certain information can actually access it. If not, you have an accessibility challenge. And remember, making it accessible doesn’t mean giving all employees access to the buffet. Rather, it’s about putting in place regulated processes that will make data available to the right people.
Once the mapping work is started, there will certainly be decisions to be made at the governance level. So a data governance framework will need to be put in place. To do this, you need to assess the basic principles that ensure how well your information is being used.
In addition to the tools, you’ll also need to think about the people. Who will be responsible for leading this project? Do you need to recruit externally or do you have an employee who would be able to start this project bit by bit, starting with the most urgent data needs?
You will also need to incorporate an internal communication strategy to demonstrate that data will be better processed and managed internally. This project is an organizational one, and employees must be driven by the potential of the data to meet the standards that will be put in place.
Finally, other initiatives may result, some more technical, such as the implementation of a new tool, or more strategic, such as changes in work methods or the creation of new positions.
Phew! Quite a topic. I think we hear very little organization talking about data structure. In a way, it makes a lot of sense because it’s very unique to an organizational context. But on the other hand, there are certainly some guiding principles that deserve to be shared, and that need to become standards so that companies can respect new laws, and the privacy of their customers. This article is just the tip of the iceberg!
I’ll end by saying that as a marketer, we need to democratize the issues surrounding data. Although marketing has become largely digital, and this has made people a little more technical, many people feel intimidated when we talk about this subject. It’s not something that can be ignored. On the contrary, it’s a subject on which we need to gain competence, since it will be an essential arrow in our bow to do our job well in the future.