On the dream of instantaneous BI reporting

On the dream of instantaneous BI reporting

Tableau, Power BI and ThoughSpot are attempting to develop the means whereby end-users can simply type in, in natural language, the business question they want answered, and an appropriate chart will automagically appear.

See the others' sales videos by Power BI and ThoughtSpot

Whether these services are quite there yet (or not) and the degree of work that would go into setting something like this up, is probably something I would have to do a bit more research on.

But on the whole, I am skeptical.

These services reminded me of something I vaguely remember about how Google's translation service works:

Rather than understanding how all the languages work, there was simply enough data out there that every sentence had already been translated somewhere, so it was just a matter of collecting it all together and establishing the links.

For most business reporting contexts, we have something similar. There are heaps of effective charts, relating to finances, marketing and sales, that have already made and publicly available.

A quick search for "analytics dashboards" (in this case for "retail" ones) will yield plenty of inspiration for ways of charting your (retail) business data.

Whether it's tracking sales, customer acquisition, social media / marketing engagement, finding a couple of excellent charts, out in the public view, is very achievable. The challenge is probably that there are too many.

It is probably just a matter of gradually refining an algorithm, so that the chart (or charts) they present appropriately answer your question.

One step further, are the services out there that supply the complete pipelined solution, provided your needs matches up nicely to their specific use case.

All-In-One Reporting Platform for Agencies - AgencyAnalytics
AgencyAnalytics helps digital marketing agencies build automated reports for clients. SEO, PPC, social media and more (40+ integrations). Try it Free!
Real Time Google Analytics Dashboard Online - Octoboard Demo
Use this real-time business performance dashboard to display your key metrics on office TVs, share your data with team members and management team online. Includes Sales, live website traffic, leads, live payment transactions.

Simply add in the relevant authentications to these service and you are (largely) good to go.


But often, the specific aspect of the business that requires analysis is not something that can be readily integrated into a service, such as this.

If it's something related to an online service, there might be some reporting functionality offered within that platform. But what if you're wanting summaries that extend across multiple services. Or what if the specific metric that's important to you is not part of their out of the box offering?

As it result, it is not surprising to still see collections of Excel files, downloaded from the various platforms, being manually copied and pasted together, aggregated with a few functions and then put into a chart. Oftentimes, due to the manual nature of this method, there is the risk of some rows being missed or being pasted twice.


The leading dashboarding tools are better than they've ever been. With enough training, you can become a reporting wizard. But the time investment required, to become proficient to produce non-standard metrics / charts, on the fly, is a fair bit more than you might be expecting.

People often resort back to Excel, even though it is not automated and is more susceptible to manual errors, because they know they can be relatively comfortable and proficient with piecing together the specific stats they are after.


Concluding remarks

Getting to some standard, "vanilla" charts shouldn't be underestimated. Setting up a good process/pipeline, so that key metrics can be reported on, on a regular basis, isn't always easy. They present the macro view, from which all subsequent, more detailed investigations are usually validated. So getting them right is important.

These charts usually aren't looked at on a day to day basis but they have value in how they guide you into deeper analysis, down to the nitty-gritty of things, where actual decisions get made.

Services that automate simple charts, from natural language might be a cool novelty. But working out much more bespoke metrics that actually guide decision making, usually require a fair bit more nous and a deeper understanding of the business, than these algorithms will be able to supply.