Do all the things like ++ or -- rants, post your own rants, comment on others' rants and build your customized dev avatarSign Up
From the creators of devRant, Pipeless lets you power real-time personalized recommendations and activity feeds using a simple APILearn More
SortOfTested24535169dBecause business degrees are easier than engineering degrees.
heyheni22603169dBusiness Bi products are for middle managers to feel powerful by clicking on some charts while showing off to higher management.
The real intelligence work is done beforehand by data scientists. Creating data lakes.
bioDan5674169dMaybe you can build something better?
vane9311169dCause most of the world was drawing those tables and charts on toilet paper 40 years ago.
So clients demand the same quality in digital products right now.
bittersweet43099169dThe more generalized/dumbed down the package is, the shittier it will always be for your specific use case.
At its core, all Analytics/BI is about logging events, storing it in a database, and allowing people to query (and maybe visualize) that data.
Products from Google Analytics to Salesforce wrap that up in a package -- so that a non-technical manager can "administrate" it, at least at first.
Of course companies eventually build custom stacks on top of that because the stack is missing a feature, so they're exporting their data from GA to MongoDB, parsing it through three Python scripts just to visualize it in Tableau.
If you go back to what Business Intelligence is at its core though, you can easily create elegant solutions from the ground up.
🔴 Decouple the logging, storage and visualization parts.
🔴 This applies to tech stacks: https://en.wikipedia.org/wiki/...
🔴 Use open, replaceable technologies (ELK, SQL, Spark, Superset, Metabase)
bittersweet43099169dAlso, ideally -- hire BI Analysts who can do the visualization part by themselves on the fly.
Devs make sure the right data is logged and stored, Analysts just need to be pointed at that data.
The ideal process means that an Analyst just fires up their IPython/Jupyter Notebook or whatever, joins three database tables, renders it into a scatter plot and a country heatmap, and tells the marketing: "We're selling a lot of purple dildos in this zipcode, but no plastic tentacle vaginas at all. Maybe we should rent a billboard next to this highway"