From skepticism to self-service: Mark at Fractory on making analytics suuuper simple

Fractory's Senior Analytics Engineer, Mark, did not fall for Supersimple at first sight. Quite the opposite. His first reaction was skepticism.

We sat down for a conversation with Mark to hear about his take on self-service analytics, data modeling, and how Supersimple has changed the way he works.

Mark, can you tell us a bit about who you are and what you do?

I'm a Senior Analytics Engineer at Fractory. Like in most startups, that means wearing a lot of hats. Primarily, I work on data models and semantic models, and making sure the data foundation is in good shape.

We have many different data sources, our own platform being the major one: user interactions, orders, customer activity, part-level details, and other operational data. Then there's the wider stack around it: Google Analytics, Google Ads, Salesforce, Airtable, and other systems used for operational processes. We also ingest sales calls and transcripts.

Do you remember the first time you came across Supersimple?

Yes, it was actually before Fractory. Someone at another company showed it to me. I checked the website and the docs, and to be honest, I was skeptical.

That changed later, when I started using it in practice at Fractory. Once you actually interact with the product, see how the development process works, and how responsive the team is – your confidence changes quite a bit.

"At first, I was skeptical. My perception shifted a lot once I started using Supersimple."

One specific thing that stood out was the pipeline-based way of building explorations. Initially, I was skeptical there too. I had seen a lot of no-code and low-code tools, and usually those approaches felt worse than just writing SQL or code. But with Supersimple, once you use it a little, it starts to make sense. You begin to think in pipelines, and it becomes a strength rather than an annoyance.

What's good about it is the explainability. You can clearly see how the data evolves step by step. Filters, aggregations, calculations, business logic – they all happen in sequence, and that makes it easy to understand what's going on. In BI, you always end up doing some calculations in the tool itself. No semantic model is ever perfect. So it matters that the logic is visible and traceable.

How do business users like this approach?

At first it feels different from the way most people are used to exploring data in traditional BI tools. But humans are somewhat procedural, so, it eventually clicks with everyone – and once it clicks, it really clicks.

What's interesting is that once they get over the mechanics, the conversation changes. The blocker is no longer "how does this tool work?" – it becomes "what does this business logic actually mean?", and that is a much better conversation to be having.

"People get to the business-logic questions faster. That's a much better place to be."

I noticed that in Supersimple people get to those questions faster than they did in tools like Looker. There, I often saw people get stuck on the technical side of how something was built. Here, they understand the steps more quickly and start asking about the logic itself.

Did that change how you think about data modeling?

Oh yeah, quite a bit. When people can really explore, they show you very quickly what is missing, what is useful, and what gets misused.

That creates a constant tension. On one hand, you want to productize what people use a lot. On the other hand, the whole point of self-service is flexibility. People don't just want the one standardized version of a calculation. Often they want something close to it, but slightly different.

So you end up in this balancing act. What amount of friction is healthy? What should become a reusable semantic object, and what should stay flexible inside an exploration?

Supersimple makes this tension very visible – but it also hits the sweet spot in helping get that balance right. It doesn't force unnecessary rigidity, but it also doesn't turn into an untraceable hell. It gives people enough flexibility to explore, while still giving structure and preserving semantics.

Has that flexibility reduced ad hoc requests to the data team?

Not exactly in volume. But what changed is the nature of the requests.

Instead of "can you change this dashboard?" or "can you build this chart?", we now more often get requests like "can you add this data so we can explore it?" or "can you enable this type of analysis?". That is a much better type of request because the work you do is reusable. You're investing in a capability, not spending the same effort on something that solves only one very specific need.

"The request changes from "can you add this chart?" to "can you enable this analysis?", which means the work you do is now reusable."

We also recently sunset Tableau in favour of Supersimple. Tableau is good for static reports, while Supersimple is best for self-service exploration. And for us, that mattered much more. We need people across the business to answer questions on their own, not depend on us for every dashboard change.

What do you appreciate the most about the developer experience?

Simplicity. So I don't even think about it that much. Compared with older BI tools, Supersimple frustrates me much less. It is simply less heavy. You define your models, push your changes, and the system tells you whether it's good to go or not. There is still structure, validation, a framework around how you build things, but the overall workflow is much simpler.

In older tools, changing things can feel like a ritual. Here, the process is much lighter.

I've also been using LLMs for code generation and model setup. The last time I did that, it took about ten minutes from having the warehouse models for a new domain ready to having them showing up in the Supersimple UI, validated and all.

"It took about ten minutes from having the warehouse models ready to having them show up in Supersimple."

Have you seen an effect on decision-making at Fractory?

Yes, mainly in speed. With tools that are built for self-service, decision-makers don't have to wait for analysts to fetch numbers or build a dashboard every time. They can explore themselves, answer some of their own questions, and move faster.

"Decision-makers don't have to wait for analysts to fetch numbers or build a dashboard every time."

That does not automatically mean every decision is better. Self-service also increases the chance of misinterpretation. Analysts exist for a reason. It's not just technical skill. It's also that they are trained to be critical and careful in how they reason with data. So the quality depends on how people use that freedom.

You mentioned unstructured data. What role does it play?

It's super important. There is a huge amount of insight in things like customer emails and call transcripts. That is where we often find product pain points, process issues, or opportunities that won't show up cleanly in structured data.

It is great to have that context available alongside structured data in Supersimple. If you can explore revenue, customer behavior, and complaint themes in one place, you avoid context switching and think more clearly.

A concrete example for us was looking at customers whose offers were not being processed quickly enough. We wanted to understand, from email and call conversations, what information was missing or what was going wrong in those slow turnarounds? Then we wanted to size that problem by customer value and revenue potential. That kind of question is exactly where combining structured and unstructured data becomes powerful. And Supersimple is where we do it.