How Can I Break Down Data Silos?

Simply put, a data silo is when you have two or more data repositories that do not coordinate or integrate with each other because they were never intended or designed to do so.  Siloed data occurs naturally and you should not be embarrassed.  Generally, a group or division created a data repository to solve a specific problem.

An email marketing list is an example of a common data silo. It was created by the marketing department, may have a low level of accuracy over time, and it is focused on email addresses, but not other contact data.  On the other hand, the sales CRM systems may have overlapping data, but the contact data will be maintained. It will not include all the addresses in the email list.  The CRM is designed for sales to manage relationships.

Data silo problems happen when you want to coordinate or integrate the databases.

Often, data repositories are controlled by different groups.  Luxury retailers often have the following common types of customer data silos:

• E-commerce customer data is often separate from in-store customer data.
• Country and other geographic business regions tend to have their data silos.
• Business units may have their data silos. For example, high-end fashion departments may maintain customer lists that are different from the rest of the organization.
• Marketing data is often in silos, such as a social media silo and an email marketing silo.

CRM data errors

(back to top)

To get a 360 customer view and to develop amazing, personalized customer experiences, you need to have a complete view of a customer. The data silo problem happens when unintegrated customer data is spread across multiple databases.

Omnichannel and cross-border shoppers tend to be your best customers.  A recent report from eMarketer stated that although omnichannel shoppers make up less than one-tenth of customers, they account for 27% of all sales.  Your customer shops where and when she wants to shop, and she expects a consistent, outstanding customer experience every time.  She doesn’t know and doesn’t care that your back-end systems contain data silos.

Here are problems that you can solve when you break down data silos:

• A VIP customer in one channel or location can receive the special treatment she deserves when her data is available at every touchpoint.
• Marketing campaigns will be more professional because she won’t receive multiple copies of an offer, or even slightly different offers, at the same time. (This becomes especially noticeable when the name or address differ somewhat for the same person.)
• You will be able to target marketing correctly or make the right recommendations based on her complete purchase history.

Breaking down data silos leads to better decisions.  Here are some examples of decisions that are improved with better customer insights:

• You will know who your best or most profitable customers are, even when her purchase history is spread across different databases. You can delight her with special offers that build brand loyalty.
• You will be able to build “look-alike” models for market segmentation or product definition.
• You can more easily find synergies across channels, locations, and product.
• You won’t waste energy and resources on bad insights created by bad data, which avoids damage to your customer experience and brand image

To create an emotional connection with your customer, you need to offer an outstanding omnichannel customer experience that is personalized and consistent everywhere.  You also need to know her very well to have the proper customer insights to identify trends.

(back to top)

Data silos tend to happen naturally. A person or group within an organization has a specific requirement and needs a database. The database is often designed with a particular purpose in mind.  The creator may be unaware of another database or may find that the other database is not fit for the needed purpose.  Sometimes, the reason a data silo is created is because the person or group wants to control the content.

For most brick-and-mortar retailers, channel-based data silos were created when the e-commerce group was set up as a separate operation. E-commerce was seen as a competitor in those early days. Now, we know that many in-store retail customer journeys start with e-commerce.  But, the data silo happened anyway.

Regional and business unit silos are very common in luxury retail.  For example, different countries or regions will establish a database for their market. Again, that is very natural.

Silos are often created when organizations merge or are acquired.  Each company will have built its own systems with structures and data formats.  In many cases, the application software used will be different as well.

(back to top)

Data silos are almost universal in retail.  Most are now trying to build a 360 Customer View or Single Customer View to build more personalized experiences and gain better customer insights.  To do this, they want to create a single source of truth.

The data silo problem retailers face is that each silo is likely to differ and some of the data will be out-of-date.  Lack of coordination means that it is difficult to know which silo contains the most up-to-date version of the information.

Data silos are the leading hurdle in customer data management, according to a 2016 survey by the Aberdeen Group.  The percentage of respondents naming data silos was double the number naming lack of budget as a hurdle to meeting their customer experience management goals.

How Big Is My Data Quality Problem?

(back to top)

Data silo integration is difficult.  Retailers face both organizational and technical problems. However, difficult does not mean impossible. Retailers are doing it. There are ways to overcome data silo problems.

The organizational and behavioral issues are challenging.  On the business side, the problem almost always comes down to the fear of losing control of the application.  It requires leadership to change behavior and convince all the parties that they must work together.  Country and regional managers often have considerable concern about integrating systems with a “corporate” mandated solution.

The benefits to the business always outweigh the concerns.  At one of our clients, there was fear revenue from a key customer would be allocated to another business unit.  For example, if a client lived in Germany, but purchased in Paris, which country manager (Germany or France) would get the revenue credit?  In many companies, the Paris revenue would be attributed to France because that is where the activity took place.  However, that might not be appropriate. The customer relationship may have been developed in Germany.

The corporate business management concern is that data silos prevent knowing how important that customer might be to the overall business. If the German customer purchases in multiple countries, her total purchases in any one market might be small. However, taken together, she might deserve higher recognition.  Will she get the experience she deserves?

There was also fear that one business unit would spam the customer list with marketing materials so that the other business units would not get through the clutter.  The spamming issue was resolved with data governance and business rules.

The technical reasons for not integrating data silos are also significant.  Different databases may organize customer data differently. That is natural.  Name, address, and phone number formats are different in different locations. A system optimized for one cannot just be merged into another.  Character sets may also change. For example, a Greek customer in Athens would likely use the Greek alphabet.  However, in the rest of the EU, the Latin (Western European) alphabet is common.  There are also detailed technical issues around database schema and storage formats.

How to bring together differently formatted databases.

(back to top)

To change behavior, it is best to demonstrate the nature and extent of the problem.  A data assessment, such as this customer data assessment performed by Global-Z, uses a sample to show the quality of the data.  When we did this for a well-known global retail brand, we discovered that nearly half of their customer data (45%) contained errors.

The assessment showed the extent of the customer overlap between divisions.  This customer discovered that there was a much more significant opportunity to upsell and cross-sell than they had ever expected.

(back to top)

Don’t try to eliminate data silos.  Data silos are not bad.  Your “data silo problem” is that you need to integrate them.

The best way to integrate data silos is to create a system of reference, not a new system of record.  In doing so, you will create a single source of truth, populated with “Golden Record” view of your customer data across your enterprise.  This is often referred to as Master Data Management (MDM).  You’ll get a holistic view of your customers for better insights and the ability to personalize interactions.

With a system of reference, your existing silos may remain. The sources do not need to be changed, which is critical for certain types of data, such as permissions granted related to privacy rights (e.g., GDPR regulations in Europe).  The silos become integrated by adding keys, such as a persistent key for the customer, that is appended to the source records and used by the master record.

Another advantage is that you can enhance information when records are matched using a combination of algorithms and business rules.  For example, if one business unit has an updated phone number for a customer who recently purchased, all of the business units can have access to the new and more accurate data.  (One needs to be careful to make sure that all privacy rights are followed.)

Once your silos are integrated, you can provide exceptional customer experiences at every touch point and have greater customer insights and analytics.

(back to top)