“Data integration is the process of combining information stored in various sources to provide a unified view of the data, increase accessibility, and decrease inconsistencies. Data integration facilitates a single version of the truth.”
Data. It is the best thing since sliced bread. Experts tout the great feats you will be able to overcome. That is, if you are able to wrangle, tame and combine data so that it tells an insightful story. That is powerful stuff, the kind of stuff that make marketers salivate.
The problem with data is that it likes to hide in a lot of unexpected places. It also has a lot of “personality.” Think: a mix between a teenager and a toddler. Once you find it, you’ll realize it is a mess (like a teenager’s room); is incomplete (you can only ever find one sock) and has no convention (decides to wear polka dots, plaid and a bathing suit to school). If you really want to make good use of data as a marketer, you are going to be forced to deal with the nitty gritty of data and its integration.
Data, Data Everywhere.
If part of your job is to manage your website or marketing automation applications, data integration will be essential for you. Some of the systems you may want to integrate data from include your CRM, HR, Email Marketing, CMS (content management system) and possibly proposal automation system. You need to start by determining which source will be the “authority” source for the specific data you are integrating.
For instance, if you have a biography on your website of a professional that works at your firm, you probably will want your HR system to be the “authority” for his/her name. If at some point their name changes, you would update it in the HR system and through data integration, this change would appear on the website. If part of your digital strategy is a content audit and strategy, it is worth aligning this effort with the integrations you will need.
Let’s ETL, Shall We?
Another consideration once you get a little further into the realm of data integration is to determine how you are going to actually extract and load the data from and into various systems. You might hear the words extract, transform and load as it relates to this.
When you begin, you extract data and put it into a common format like an XML file. This basically is putting the data in a format both systems can understand and agreed on so that it can be easily imported. Transform refers to the application of “rules” as it relates to the data that has been extracted that prepares it to be loaded into the system where it will be used.
Let’s say we extracted all the data from an HR system (you would never do this, btw), but we only really needed the name and title of the person. During the transform phase, rules would be applied that basically say to load only the name and title data from the XML file.
The load phase is exactly as you would expect. You load the data into the system you are using, like a content management system. Seems easy right? Not so fast.
One important step is to determine how often you want to reload this data from the authority source. Do you want it every hour, day or week? Also, what happens if the load doesn’t work? You need to have a system in place that alerts someone when the load isn’t processing. An alert should lead to an investigation, fix and then reload. If you don’t have this important step, things can get ugly. Mass chaos, even (teenager’s room).
Lock it Down. Serve It Up.
No discussion about data integration would be complete without some mention of security and performance. As it relates to security, you need to work with your tech team or agency to determine what data is accessible, how it should be encrypted and where it should be housed during the “transform” phase.
Finally, you need to consider that the more complex data integrations are, the greater possibility there is of performance impacts. More data means more strain on applications and environment, which could impact the experience you are trying to create. It is for this reason that you need to be in-tune with your tech team and tech partner, so that you can make the required investments early on.
As marketers, we often jump to the end state of a project and don’t consider the effort required to get us there. Data integration is not one of the places where you want to jump ahead. You should make sure that you are extracting good data, keeping it secure, ensuring it loads properly and is ultimately used/presented in a way that provides value to your business. Having a good relationship with your team helps make sure this happens. With that, go forth and integrate some data.
Written under the guidance of Data Integration Extraordinaire, Michael O'Laughlin.