Super Bundling of Business Intelligence Tools
Every new feature seem to degrade the overall product
The original idea behind Business Intelligence (BI) tools was to enable self service reporting for non technical users. Over the years BI tools have picked up many more features and feel less and less suited for business users. We went from drag and drop report building to highly complex dashboards that can only be created by platform experts. Vendors sell BI tools with the promise of “self service”, but companies quickly realize many of their requirements can only be fulfilled by experts. We are not talking just about common technical operations like semantic modeling, server deployments, data source configuration etc. Expertise here makes sense. Instead, the core self service utility itself has become over complicated and require advanced skills.
To illustrate, imagine I’m a business analyst in the Customer Service org and I’m coming into an existing Tableau server environment. I just want to know a little more about a KPI. Where do I even start? There are hundreds (often thousands) off data sources and enumerable workbooks. I can’t even start without a guide or manual. Sure sure, you have to do some work, everything is not going to be handed to you on silver platter. But simple things should be simple! This is the kind of inertia you have to get over to make your organization truly data driven. (I don’t mean to pick on Tableau, it’s just popular and well understood. They all have similar issues.)
Vendors often promise enablement for business users, but in reality they are reduced to mere passive consumers of static data. Here experts create reports and dashboards for the masses. This is fine for a class of use cases but it is horrible for your operators. These are the users running your business, purportedly making data driven decisions day in day out. For them this is anything but self service. Basic questions that drill down even to shallow levels often require an engineer in the loop. No question that this is slowing down decision making.
How did we get here?
Feature bloat over the years have largely been thoughtless and reactive. This is the result of a highly competitive and fractured market. You could easily list the top 25 with ChatGPT and you would not have heard of many of the names.
Here is a list of less well known BI tools from Grok:
Cognos
Sigma Computing
Hex
OBIEE
Evidence.dev
Lightdash
Metabase
Pentaho
Pyramid Analytics
FineBI
Some of these are startups and some are legacy tools that still hold significant market share. How many of these have you heard of?
All of the tools start with an amazing core that solves a particular problem in a novel and useful way. But to grow their market share they quickly and thoughtlessly add competitor features. Over my 25 years in the industry I’ve seen many tools come and go and watched vendors adapt. As soon as a competitor shows up with a threatening proposition it is copied. Often, poorly just to check a box during a sales cycle.
Maybe we (buyers) are the problem
Customers go into buying mode with so many requirements that many are at cross purposes with each other. We want data prep, data viz, data exploration, semantic models, etc etc. But, we don’t spend time to understand who will benefit from each of these features. And we certainly don’t think about how these features should be delivered. For example, are the users of data prep capabilities and semantic reporting the same? If not, how will each user interface with a tool that has both in a coherent way. Long story short, they are totally different and vendor implementations of each will be too shallow to satisfy anyone.
Another bigger problem from the buying side is the team/persons with outsized influence on the final decision. Are they the right influencer for the target problem?
Read my post on how Lookers rise is mostly driven by data engineering rather than the end users:
Finally, many larger orgs with two or more existing Business Intelligence tools go into the market overly focused on consolidation. When this is the overriding impetus, you end up buying tools with all the bells and whistles on paper. Only later do you find out that these belles and whistles are too immature to replace the existing tools. Now you have 3 BI tools. Congratulations, welcome to the fortune 500.
Or, Maybe its the Market Research firms
BI tools, according to Gartner must have the following functionality:
1. Automated Insights
2. Data Preparation
3. Data Visualization
4. Manageability
5. Product Useability
The last two are no brainers for every tool. Not even sure why its on a list of must haves. Anyways, each of the functionality above is doing a lot of work. When you dig in, they are super broad and range in capabilities and skills match. Just looking at the top three it’s obviously at least 3 different types of personas. Maybe more when you get into the details.
What skills will the data prep persona bring to the table as compared to data visualization? Automated insights sounds like something fit for data scientists with specific expertise.
It’s probably fine if a platform offered all of these features. The problem is how it’s delivered and for whom. If there were sharp lines between the modules in terms of functionality and user match, then it might work. For example, when you get Google Workspace it comes with Docs and Sheets. But, they are two totally separate modules that serve specific use cases. Imagine if Sheets was a feature of Google Docs. This would be a horrible product. Instead these products are only loosely integrated and not dependent on each other. BI platforms tend to tightly couple all of their functionality. They feel bloated and not targeted for any specific persona.
A Time to Unbundle
You are going to buy more than one BI type tool for your organization anyway. You might as well get the best for each problem you are trying to solve. Imagine a tool that is only focused on delivering the best interactive dashboarding capability. No data prep, no automated insights, no semantic reporting. Just the best dashboarding tool for Analytics Engineers. This will be completely different from the best dashboarding tool for non technical business users. Combining these two personas is how you end up with tools that nobody likes.
The other benefit of focusing on specific tools for specific problems is that you don’t have to pay bloated prices for bloated software. Perhaps your total cost might remain the same but your productivity in each problem space will be exponentially improved.
It’s time to consider unbundling your BI stack. What do you think, should vendors start unbundling core functionality?