The 90/10 Problem Nobody in Data Talks About

I spent the first decade of my career in retail and growth. I understood the businesses I worked in: the numbers, the patterns, what was working and what wasn't.

But I couldn't access my own data.

Not because it didn't exist. Because getting to it required skills I didn't have, and the people who did were always busy with something else.

I assumed this was my problem. Then I started asking around. It wasn't.

That realisation is what Edilitics is built on. Not a market opportunity. A frustration that turned out to be structural, and that five years of building later, we've finally done something about.

November 2019: The Wall

I joined a fast-moving consumer tech company as a Growth Manager. The first week made one thing clear: if you needed data here, you either wrote the SQL yourself or you found someone who could.

I had never written a line of SQL in my life. Never heard of MySQL or BigQuery. My world until then was Excel, PowerPoint, and just enough SAP to pull a standard report.

MySQL Workbench, a viz tool, Zapier, Asana, Google Sheets. None of it talking to each other, all of it requiring technical knowledge just to get a basic business question answered.

So I learned SQL. On W3Schools. Every day, a little more. Not because I found it interesting, but because without it there was no path forward here.

But something about this bothered me in a way I couldn't quite shake.

I had years of retail domain experience behind me. I understood how a business worked: the margins, the patterns, the levers, the decisions that actually mattered. I could look at a number and tell you immediately whether it was good or bad and why. But I was still dependent on someone else just to pull the number in the first place.

My judgment was only ever as fast as someone else's availability.

The Course: And What It Actually Taught

In late 2019, I came across a Management Development Program in Data Science at IIM Lucknow. The promise was clear: no prior coding knowledge required.

I decided to enrol. The program started with a campus visit in February 2020. Then COVID hit. India went into lockdown in March. Everything moved online and the weekend classes continued.

I left my job in August 2020. Working as a Growth Manager from home wasn't making sense, and the final three months of the course were going to be the most demanding. Pure R, Python, Pandas, machine learning, and algorithms. Everything the course had said you wouldn't need turned out to be the only way any of it could actually be implemented.

The course that promised no coding knowledge required ended up being almost entirely about code. That should have told me something about the industry.

I stayed through all of it.

I graduated on December 23, 2020.

Then I went back to job hunting. And walked straight into a wall I hadn't expected.

What Every Job Listing Was Actually Saying

Every role I looked at demanded 5 to 10 tools. Python, R, Tableau, Power BI, BigQuery, PyTorch, machine learning frameworks. If you assigned a total weighting of 100 to what employers were actually hiring for, tools got 90. Domain knowledge, industry experience, actual business judgment got 10.

This is when the frustration stopped being personal and started being something larger.

I had spent a decade building real expertise in retail and growth. I knew how to read a business, spot an opportunity, interpret a trend. I had just spent a year learning data science on top of that. And the market still wanted someone who had lived inside Tableau or Power BI or BigQuery for years, not someone who understood the business those tools were supposed to serve.

But here is what struck me more than my own situation.

The professionals I knew in finance, operations, marketing, and retail (people with years of genuine domain expertise) were all in the same position. They were Excel-fluent. They knew their craft deeply. But Excel had a ceiling, and above that ceiling sat a world of databases, pipelines, and tools that required an entirely different skill set. Most of them had neither the time nor the inclination to build that skill set from scratch. So they relied on younger colleagues, on data teams, on whoever could write the query.

Their judgment, built over decades, was only ever as fast as someone else's calendar.

This wasn't a skills gap. It was a structural dependency that the industry had decided was acceptable.

January 2021: The Conversation That Confirmed It

In January 2021, I was on a long overnight drive with a group of friends. One of them had spent his career in finance, not in data or tech, but in the craft of financial analysis and business judgment.

I asked him whether he was seeing the same thing I was describing. The dependency on others for data access. The Excel ceiling. The gap between knowing your business and being able to act on it without going through someone technical.

He said yes. Without hesitation. He saw it everywhere. Experienced professionals, people who knew their domain cold, unable to move at the speed their judgment deserved because the data sat behind a wall they couldn't get through on their own.

That conversation did something important. It moved the problem from my own experience to something structural. This wasn't a data industry problem. It wasn't a seniority problem. It affected anyone whose expertise lived in a domain rather than in a tool.

And it was getting worse, not better.

February 2021: The Idea

By the time I got back to Mumbai the shape of an idea had formed.

The fragmented stack (the five or six disconnected tools that most businesses were stitching together with engineering effort and technical dependency) was the problem. Not any single tool in it. The fragmentation itself. The fact that connecting your data, cleaning it, transforming it, and visualising it required multiple platforms, multiple skill sets, and constant engineering involvement just to keep it all running.

What if there was a single platform that collapsed all of that?

One platform. Connect all your data sources. Build pipelines safely. Transform data into something analysis-ready. Visualise what happened. No code required. Point and click. Built not for data teams, but for the people who had been waiting on them.

I spent the following month pressure-testing the idea before committing to it. Not pitching, but listening. I spoke to people I knew across retail, finance, and operations. Every conversation confirmed the same thing: the problem was real, it was widespread, and nobody had built a proper solution for it.

Then I called a senior colleague from an earlier chapter of my career, someone with deep technical experience who had become a mentor, and who had spent years as VP of Engineering watching exactly this dependency play out from the other side. The engineering teams being pulled away from real work to pull numbers for people who already knew what the numbers meant. The tickets raised just to run a query. The hours spent being a data access service instead of building anything.

I talked for 90 minutes straight. The idea, the why, the vision, the shape of what it could be.

He didn't say a word throughout.

When I finally stopped, he told me it made sense. Not as encouragement, but as someone who had lived the other side of this problem for years and recognised immediately what solving it would mean.

That was the day the idea stopped being a thought and became a decision.

What Came Next

In May 2021, I started building.

No technical background. No co-founder. No funding. Two CS interns and one question that nobody had answered yet: could a non-technical founder actually build this without a technical co-founder?

The logical reason I had not gone looking for a co-founder was simple. Why would anyone serious partner with someone who had zero technical experience, an idea so large it sounded impossible, and no proof it could even be built? I was not in a position to make that ask. So I did not make it. I looked for proof instead.

If the proof of concept worked, the conversation would be different. If it did not, there was nothing to partner on anyway.

What the next two and a half years of building looked like, what it cost, and the moment in late 2023 that nearly ended everything — that's the next post in this series.

Sources

McKinsey Global Institute. The social economy: Unlocking value and productivity through social technologies. July 2012. Retrieved May 2026. mckinsey.com