How do you troubleshoot and resolve common issues and errors when using BI cloud and SaaS platforms?

I think it’s important to know also root cause analysis will likely go deeper than a single error - it will be layers. It’s not just the problem detected but every layer. As an example, you have an error message on the data set, you correct it - that’s troubleshooting. It didn’t happen out of nowhere you identify that the data type changed or a field was added or the dev was junior that’s getting closer to the root cause. Let’s say it was a data type change and no one did QA or testing on the reporting. Then that’s the root cause.

I first inspect the fields that go into the result that throw the error message. I’ll then go back to them to check the data types or other problems outcome results (trying to take the square root of a negative number for example). As with other programming languages, my next step is to do a Google search for discussions forums (like Stack Overflow) to see if others have run into the same error message before. Fortunately that’s quite frequently the case and I can translate their problem solution into one that will work in my own model.

Read the error message: Clues as to the initial or 'root' cause of an issue can often be found in the error message. Identifying key phrases or error codes can help you narrow down the root cause of the problem. Check the logs: Many platforms provide logs that record activity and errors. Check the logs to see if there are any relevant error messages or events that may point to the root cause of the issue. Test different scenarios: Once you have some information about the potential cause of the issue, try testing different scenarios to see if you can reproduce the error. This can help you confirm the root cause and identify any additional factors that may be contributing to the issue.

Checking the data source as well as data connection when troubleshooting errors can yield important information. Verify the connection settings: Check the connection settings for the data source. This includes checking the server name, database name, username, and password. Test the connection: Use the provided tools or separate connection testing tools to verify that the connection is functioning properly. Confirm the data schema: Make sure that the data schema matches the schema that is defined in the BI cloud or SaaS platform. Ensure that the data types are compatible and that there are no missing fields or records.

Make sure you have common taxonomy across your data sources, especially as you have turnover in staff or introduction of new data sources.

I worked on big projects with multiple data integrations feeding the BI SAAS database. Having a standard documented process to run the jobs and report issues is crucial. There should be clear requirements with all formatting and data transformation rules. Have clear documented guidelines on who is doing the data transformation , the data owner or the vendor? If possible have the vendor do a compare process to look for data issues prior to loading the data.

Remember that you won’t always have control of the settings. Learning the lingo can be helpful when working with a multi discipline technical team. Knowing what to ask for is important and how to communicate with people who have control of the servers and settings is critical.

Confidentiality and compliance are probably the most critical challenges when trying to resolve technical issues in a SaaS environment: if you can't isolate the root cause yourself and need help from the vendor, how can you preserve the confidentiality of your dataset and respect all compliance rules? One way to do that is by using a synthetic data generator. This tool, when implemented properly, can generate a synthetic (random) dataset that looks like the original one, with the exact same table schemas, all the way down to probability distributions and summary statistics. In most cases, it should allow the vendor to replicate your problem, without leaking any confidential information or IP. But check with your CISO first!