Halloween Special - Data Horror Stories

In this podcast, the team discusses common data horror stories in IT consulting. They highlight the issue of hiding bad data with filters in reports instead of fixing errors at the source, leading to bigger problems during system migrations. Another problem discussed is trying to use tools like Power BI to mimic Excel functionalities, which leads to alignment and maintenance issues since Power BI and similar tools are not designed to replace Excel.

The conversation touches on the issues with manipulating data in Excel before importing it into Power BI, causing inconsistent data and errors. There's also a focus on the importance of cleaning data before migration to avoid transferring existing problems to new systems.

They mention the challenges of treating estimates as fixed prices and the tendency of clients to ask for continuous changes, causing scope management issues. Another point raised is the problematic practice of turning prototypes into production reports, leading to more issues because prototypes are not meant for long-term use.

The discussion also covers the problem of having poorly defined measures across different departments, resulting in inconsistencies in reporting. Lastly, they emphasize the importance of understanding one's own data to ensure accurate and effective data management and reporting. The episode concludes with a reminder to stay safe and avoid nightmares from data horrors.

You can download Power BI Desktop from here.

If you already use Power BI, or are considering it, we strongly recommend you join your local Power BI user group here.

To find out more about our services and the help we can offer, contact us at one of the websites below:
UK and Europe: https://www.clearlycloudy.co.uk/
North America: https://clearlysolutions.net/

Transcript

Andy:

Let's get to the point of the podcast. Today, we're talking about data horror stories. This is the Halloween special. I wish I could do an evil Vincent Price laugh to set the mood. Tom, can you add some spooky sound effects like storms, thunder, and an evil laugh?

Tom:

I'll see what we can do.

Andy:

Great! We're going to discuss data horror stories—common mistakes, bad practices, and specific examples of what goes wrong in data implementations. Tom, you were quite heated before we started. Do you want to kick us off?

Tom:

Sure. If you're listening and this sounds like your company, it probably is. Many places make the same mistakes. One of the worst practices is putting filters into your reports to hide bad data. This happens everywhere. Instead of fixing data errors at the source, companies often ask the reporting team to hide these errors. For example, if a transaction has an incorrect amount or if a barcode is accidentally scanned into a quantity field, they just filter it out in the reports. This leads to more issues, especially when migrating to new tools or systems. Always fix data errors at the source!

Shailan:

Masking data is a common issue, and I've seen it a lot too.

Andy:

I've done it as well. Sometimes we're forced to do it despite knowing it's a bad idea.

Shailan:

Another horror story is trying to make tools like Power BI function like Excel. A senior person used to Excel might want to print Power BI reports and highlight them physically. Power BI is not designed for that, and forcing it to behave like Excel can lead to alignment and maintenance issues. Tools like Power BI and Tableau are analytics tools, not replacements for Excel.

Andy:

I've seen similar issues where organizations manipulate data in Excel before importing it into Power BI, leading to inconsistent data and errors every month.

Tom:

That's common. Many don't want to clean their data properly, which leads to problems when migrating systems. Migrating dirty data transfers all the existing issues to the new system. Always clean your data before migration.

Shailan:

Yes, treating estimates as fixed prices can also lead to horror stories. Clients often ask for continuous changes, thinking it's easy, but it isn't. Another problem is turning prototypes into production reports. Prototypes are meant to be throwaway work, but clients often want to keep and use them, leading to more issues.

Tom:

Another common problem is having poorly defined measures across different departments. This leads to disagreements and inconsistencies in reporting. It’s crucial to have well-defined, consistent measures.

Shailan:

Absolutely. Not understanding your own data can also be a big issue. Clients sometimes expect us to make their data work without proper understanding or clarification.

Tom:

We've covered a lot of ground today. This was our Halloween special on data horror stories. Stay safe and don't have nightmares!

Shailan:

And do sleep well.