Time to Stop
the Delusions

Everywhere you go in the analytics world today, you’ll hear the same term repeatedly: “self-service.”
It’s the new hotness.

Think it’s an exaggeration? Search Google for the term, and you’ll get (in .46 seconds) 776,000,000 results. Nearly a BILLION results.

And the titles:

“The Rise of Self Service Analytics.”
“21 Self-Service Analytics Platforms.”
“What is Self Service Analytics.”

You’ll be hard-pressed to find one of the 350+ analytics platform vendors that doesn’t claim to be self-service somewhere on their website.

Clearly, something is going on.

The world is being taken over by self-service analytics — everybody wants them; every vendor has them. It’s a fantastic time to be alive! Any average business user, faced with a challenge that requires data to solve, can instantly access the data that they want, format it in the required way, and perform the necessary analysis to solve their challenge. What a time in human history!

Or is it? What if it’s all the fallacy? What if it’s all a lie, sleight of hand, a misdirection that we’ve been led to believe is real? What if this isn’t the state of the world?

While calling the claim of “self service analytics” might not be an outright lie, it
is often misleading at best. And it falls into three categories:

“very narrow
view of the world”

“we know
what’s best crowd
for you” group

“deniers of

The “Narrow View
of the World”

You know the saying that “if you’re a hammer, everything looks like a nail?” That’s what’s going on with the first group of self-service self-proclaimers. They tend to be visual analytics firms — or at least, that’s their heritage — and everything is viewed through the lens of charts and dashboards.

To this crowd, “self-service analytics” is defined very narrowly: if the average business user can create new charts, move them around a page, change the colors and filters — boom! You’ve got self-service analytics. It’s not that easy creation of analytic visualizations isn’t a necessary element of self-service, it’s just that it’s not nearly enough. What if you need to augment the data you’ve been provided? What if you need to add new fields, new tables, or entirely new sources? Can you do this yourself? Unless you’re familiar with data models like star and snowflake schemas and do a little data
engineering in your off-hours, you’re probably going to need some expert assistance.

And that’s not self-service.

The ability to create and modify visual analytics might be held out as self-service by some vendors, but that’s like completing a color-by-numbers book and calling yourself Picasso. Close, but not quite.

The “We Know
What’s Best For

“Hahahaha! That’s not us,” I can hear many analytic vendors exclaiming.
“We’re not just a self-service chart-builder. Our users can add or modify data easily!” How do you know if you’ve found one of these vendors? It’s easy — they tell you in the first 15 minutes of any meeting.

But it takes a bit longer, a bit more digging, to find out what they mean. To interpret their “marketing-speak.”

This group really means “you can add or modify data as long as it’s in the model that we’ve preconfigured for you.” That’s right — they’ve taken a shot at guessing what you need to manage your business with data, and as long as they’ve assumed correctly, you’re fine. Of course, we know that reality is often quite different than initial assumptions.

The second an executive asks, “sure, but what if…” everything falls apart. If the data you need isn’t in the model, you’re out of luck. It’s like riding on a rollercoaster and extolling the freedom you have to travel anywhere you want — as long as it’s the pre-defined path.

Like the first scenario where self-service is viewed as the ability to manipulate visualizations, having limited data capabilities is nice, but not sufficient to be called self-service analytics.” Yet again, as soon as you want to deviate from the tracks that have been laid out, you need to call in the experts. The average business user doesn’t have the skill set or the time to ingest new data sources, transform the data as needed, prepare a data model, and begin analysis. It just isn’t the real world.

The “Deniers of Reality”

Finally, we have the “deniers of reality” — perhaps the most insidious of the three categories. The problem here is that it’s hard to see what’s missing with these analytic platforms.

Do they have self-service visualizations?
Yes, they do.

How about the ability to add new data sources without expert support?
They have that, too.

So what’s the issue? The problem with the deniers of reality is that they have an outdated view of reality. They are self-service — as long as all of your data is in a friendly, structured format. If you’ve got well-understood SAP data, they’ve got you covered. If it’s self-service analytics from Oracle Business Suite — you’re good to go. To the deniers of reality, living in a world that existed five years ago, the data world is always highly-structured. It’s a fantasy world.

If you’re not familiar with the concept of “structured” and “unstructured” data, here’s what we’re talking about:

Structured Data

Structured data is data set up in the familiar tables, columns, and rows format with which most of us are familiar. Customer ID connects customer address, product purchases, and payment information in a very straightforward manner.

Unstructured Data

Unstructured data is different. Data might all be stored in a single object, with the fields containing the information necessary for your analytical insights nested deep down within the structure. Although it might sound complicated, it describes how much modern data — like social media posts — is stored. Most analytical platforms are designed for structured data and they do a great job with tables, rows, and columns but fall down spectacularly with unstructured data.

They simply aren’t designed to traverse down the nested structure, find all the fields, and understand all the relationships necessary to do analytics.

So what happens with semi-unstructured data that is required for analytics?

You call in the experts. Data engineers dig into the data, find the fields (hopefully), and convert the object or array into multiple “structured” table/column/row type files. Then the analysts get involved. Now that they have a workable, structured format, they can get busy. They take the multiple files that were derived from the original unstructured data file and start piecing them back together. All the relationships that connected the data were lost when the conversion occurred — the analysts have to rebuild them so that the data can be consumed by the analytic tools.

Doesn’t sound very “self-service,” does it?

The world today isn’t so structured as it was a decade ago. Today we’ve got JSON mixed in with traditional relational data making things more complicated. Those platforms that are easy for the average person to use with structured data often fail when the structure is stripped away. They don’t allow the user to traverse complex, nested data structures and turn them into insights without significant — and often highly technical — manipulation of the data. And that’s the problem.

The purveyors of these platforms, when they call themselves “self-service,” are willfully ignoring the reality that exists today — the new, unstructured world in which we all live.

And they’re hoping you won’t notice.

Let’s Reset the Definition of
“Self-Service Analytics”


What we need is a redefinition of the term “self-service analytics.” It’s been used too loosely, too broadly, and too casually and, as a result, has lost much of its meaning. When we say self-service analytics, we should be conveying a common understanding about the capabilities being discussed.

Here’s how we need to define self-service analytics for today’s world:

Self-service analytics describes those data analysis platforms where the average, non-technically trained business user can:

Add new data sources
without expert assistance

Create metrics, combinations of metrics, charts, filters, and other analytical elements without expert assistance

Combine the analysts into dashboards or reports without expert assistance

Make the data ready for analysis by
performing any needed transformations without expert assistance

In a nutshell, if you claim to be self-service analytics, then all parts of the process should be self-service. We shouldn’t let the pretty chart makers claim to be self-service when they require hours of data modeling, and we should balk at data transformation tools making the claim when they have no way to visualize the data.

Self-service should mean a complete, end-to-end, do-it-yourself experience.

Where Do We Go From Here?

We need a revolution. We need you — and the analytic users like you — to stand up and stop accepting the self-service claims generated by the marketing teams at analytic vendors the world over.

Each time you hear the term, stop the presentation, and start asking questions:

  • Can I add new data sources myself without a data expert?
  • Can I do things like suppress zeros and negative values or convert dates without an expert to help?
  • Can I combine data from multiple sources on my own?
  • Am I able to build new dimensions and filters on my own?
  • Will I be able to start visualizing, drilling down, and sharing the analytics on my own?


If they answer “no” to any of these questions, it’s not truly self-service. You’ve found one of the pretenders, a “narrow view of the world,” a “we know what’s best for you,” or a “denier of reality.”

And that’s a sign that you need to walk away.

Luckily, you don’t have to walk too far to find vendors that can help you in your quest to deliver self-service analytics to your organization. While few (if any) vendors can individually deliver on all the elements required to fulfill the self-good promise, there are options. Look for vendors that can connect to any data source —structured or semi-structured — and make it ready to use without requiring the user to write a bunch of code. Seek platforms that allow the user to begin analysis without the development of a complex, brittle data model. And find analytic tools that work together without the need to have data engineers develop complex scripts to pass data from one phase to the next.

Self-service analytics platforms aren’t easy to find despite all the claims. In order to find these elusive systems, you need to understand what to look for — and what to avoid. Do your homework, look for the shortcomings and avoid the pretenders and you’ll be able to deploy true, self-service analytics to your business.

Ready to Start?