6 Tips for Successful Data Integration
"The big focus, the No. 1 data-centric area that will get the most significant investment over the next 12 to 18 months is data quality."
Mike Leone, quoted in Data Forecast for 2022: Data Quality
and Cloud Convergence, published in Tech Target.
As Leone's data quality prediction signals, a firm's data is valuable as a business asset and needs to be accurate.
Organizations are creating a deluge of data. "More volume, in more sources, spread across more clouds, makes it challenging to discover, manage and control data" according to a recent Informatica Global CDO Survey.
The survey showed that 79% of organizations use over 100 data sources, with 30% using over 1,000 sources. And the trend is toward more hybrid and multi-cloud platforms.
With so many sources of increasingly complex data, integration solutions are even more critical for firms trying to optimize their systems and investments.
But some data integration projects can run adrift. Some go on way too long. Others shift goalposts so often that the original reason for the integration remains unaddressed.
Here are some common pitfalls:
- Bad initial data quality.
- Choosing the wrong solution for your needs.
- Failure to win executive support.
- Inability to define clear, measurable criteria for project success.
- Misjudging total project costs.
- No long-term vision for data needs.
Let’s take a closer look at each of these, so you can avoid them and achieve success in your projects.
1. Start with Data Quality
Quality data is the cornerstone of any system integration project. You need good, accurate data as the basis for sound decision-making.
But not all data is good data. Bad data can include missing, wrong, or duplicated data, spelling variations, incorrect formats, and typos caused by human error. These things waste time and can lead to mistakes. Gartner's research found that organizations believe poor data causes $15 million per year in losses.
Long before any integration project begins, you should clean up your existing data. Data cleaning is the process of ensuring your data is correct, consistent, and usable.
It’s a good idea to create a data cleanup strategy, and the place to start is to stop errors at the source.
Here are some best practices shared by Geotab:
- Monitor errors. Keep a record of where most errors are coming from and address them at the source. You will probably find that errors will stop when people have responsibility and accountability for quality.
- Standardize your process. Standardize the point of entry to help reduce the risk of duplication.
- Validate data accuracy. You can research and buy data tools that let you clean your data in real-time or hire a consulting firm to handle it for you.
- Scrub for duplicate data. Identify any duplicated data, either manually or by using automated data tools to analyze raw data in bulk.
- Analyze your data. After data has been standardized, validated, and scrubbed for duplicates, you can then use third-party sources to organize its metrics.
- Communicate with your team. After cleaning up your data, it’s essential to keep it clean. So, share the new standardized cleaning process with your team and encourage them to adopt it.
2. Get the integration you need
There are many ways to integrate data. The wrong decision could be costly, so understand your needs and goals before starting.
Ask yourself: What is your company’s overall goal? What exactly do you want to integrate, and why? What problems or pain points are you trying to address? Will this integration solve that? Who will be involved?
Also, consider whether you need to integrate your data, your applications, or both. These are not the same thing. Data and application integrations involve different types of processes.
Get your stakeholders together and brainstorm to clearly define and agree on your integration scope, goals, and metrics. Assess the feasibility and ROI of different solutions or approaches. Listen to expert, impartial advice. Then choose your solution wisely, within your budget.
3. Ensure top management support
No data integration project will succeed if it does not have the support of your top executives. These usually include the CIO or IT director, the CEO, and the CFO.
You also need support from the department representatives who will be directly affected by the new system. For example, to integrate your core HRIS system with a payroll system, you need buy-in from your CFO and your CHRO.
Remember that any data integration project is more than just an IT project. It enables the automation of processes and the sharing of data. If some people have an “ownership” approach to their data, preferring to control access to it, it’s essential to understand why and address it.
4. Define your success metrics
How will you measure the success of your project? Collaborate with stakeholders to arrive at common, logical, effective assessment means.
Use any yardstick for success you think is necessary. Success criteria might include, for example, a better return on investment of the new approach, or cost savings due to reduced administrative needs. Perhaps better customer service will be one of your criteria.
Long-term success criteria for the integration solution differ from short-term criteria for the implementation.
Project implementation will probably involve validation, choosing a vendor systems expert, and testing before any new data system goes live.
The longer-term success of your data integration project is arguably much more critical. How will you assess if your new integrated approach is demonstrably improving your operations or outcomes?
5. Don’t misjudge the costs
Misjudging total project costs for data integration is a common mistake. In addition to the cost of the new tool or data-sharing system, you also need to consider the cost of operating the finished solution. How expensive will updates and maintenance be?
Also, consider the end-users. If they find your solution too hard, they won’t use it. Take the time to understand your users. Align your solution to their skills, work context, and needs.
6. Plan for the future with good data governance
Do you want to integrate just two systems right now? Or might you need to incorporate more systems down the road? Custom integration can be a quick fix—but may not be a good decision for your long-term data needs.
So, consider what the organization may need in the future. Data is dynamic and tends to grow. It’ll also change structures and formats. Your processes, too, may change over time.
Data governance helps a company manage its data assets wisely by providing company-wide guidance in data policies. It can help prevent the chaos of unplanned and incompatible “mini-integrations” and multiple hybrid cloud solutions erupting in different departments throughout the company. It outlines roles and responsibilities for managing data.
Having good data governance from the outset will give you a head start in the success of your data integration projects.
Plan for the future. Your project will not be one and done. Technology will change, and your governance and execution need to change with it.
Use your governance framework to guide your future endeavors.