Business Intelligence
The Goal of Business Intelligence and the Problems of Business Intelligence.
Close connected is Data Engineering, which begs the question of Business Intelligence Engineer vs Data Engineer.
# The Funny Reality
See also BI Tools.
# Other Resources
# Awesome BI
Linked Content:
|
|
# The BI Storyline: “If You Give An Executive A Dashboard”
Instead of rewinding the long history of BI, from relational databases to data warehouses with dimensional modeling and much more, which I’ve done here and Tech Target here, we can go through the story every BI engineer knows: “Executive and a Dashboard”.
The story is a way of explaining the natural order in a funny way, but that is very accurate if you have been working in BI, from getting a simple executive dashboard and making it real-time, to exporting it to Excel - Bill Yost tells the story in this way:
If you give an executive a dashboard, they’ll ask if it can be real-time.
When you make it real-time, they’ll ask if it can refresh faster.
When it refreshes faster, they’ll ask why one number flickered.
When you explain the flicker, they’ll ask if the data is trustworthy.
When you explain the data sources, they’ll ask if we can add just one more.
When you add the new source, they’ll notice the numbers don’t match.
When the numbers don’t match, they’ll ask which one is correct.
When you explain that neither is wrong, they’ll ask which one they should use in the meeting.
When you ask what decision the dashboard is for, they’ll say “just generally.”
Which means they’ll ask for a filter.
When you add a filter, they’ll want another one.
And another.
And another.
When there are twelve filters, they’ll ask why it’s so complicated.
When you simplify it, they’ll ask where the filters went.
When you put the filters back, they’ll ask if we can break this out by region.
When you break it out by region, they’ll ask for a global rollup.
When you add the rollup, they’ll ask why the totals don’t match the regions.
When you explain aggregation, they’ll nod and say that makes sense.
Then ask if we can export it to Excel.
When you export it to Excel, they’ll email you a screenshot with a circle around one cell.
When you answer that question, they’ll ask if the dashboard can show that instead.
And chances are, once you give an executive a dashboard, they’ll want a new dashboard to go with it.
The funny thing is, in this short story, we have all components and key tasks and different stages of BI such as ingestion increasing to real-time or better aggregation and rollups, usability with filters or just making it visually more appealing, summarized in one story.
# With AI in the Picture
With AI, we approach this story in almost reverse order. Give the executive the full worksheet in the form of a prompt, customizing the data, adjusting the filter, or changing the look.
It’s fully customizable like the Excel sheet first, which makes it so powerful for the business person and executive, and so hard for the BI engineer, as most of it might just run on local data with the chat using pandas or DuckDB to apply filters. Every insight seems just one prompt away.
The hard part is going up the stack (or left), adding filters, adding new data centrally, making it fast for everyone, making the numbers correct, verifying them.