From running Facebook’s first ads to improving marketing data with Clarisights : Arun Srinivasan’s journey

This article is part one of a conversation which will expand into our upcoming podcast as Arun is a brilliant storyteller. Stay tuned!

My career in online advertising started somewhat accidentally. I was in college when a friend of my brother introduced me to PartyGaming, a massive online poker platform. They brought me in as an intern to do internet research, which meant finding websites where we could place ads. That was my first exposure to media buying, and it quickly turned into a full-time role.

At the time, digital advertising was still in its infancy. There was no Google Analytics. The way we bought media was different. You would find a bunch of niche websites, negotiate directly with the owners, and place ads manually. It wasn’t as automated as it is today. And this was a performance-driven business, meaning every dollar we spent needed to generate a return. That mindset shaped my entire career.

That was pure hustle. I was already experimenting with community-driven advertising. Forums were a huge opportunity because poker was a game people played in groups, and online poker thrived on community engagement. I was running ads on sites like AnandTech, where people were deeply engaged in discussions, and the results were strong.

Then I came across an article in Wired about a new platform called “The Facebook” that was gaining traction across college campuses. It immediately made sense to me. College students loved poker, and this was the most direct way to reach them. I cold-emailed Facebook to see if they would let us advertise on the platform. They didn’t have a sales team at the time, so they forwarded me to a third-party agency handling ads for them. That’s how the conversation started.

It wasn’t easy. They were skeptical about running gambling-related ads on a student platform. But I knew poker was different from traditional gambling. It’s a game of skill, and I made that argument. I pointed out that poker nights were already a big part of college culture and that our platform was just a digital extension of that. After a few rounds of back and forth, they finally agreed to test it, and we ran what would become Facebook’s first paid ad campaign.

It was an instant success. Facebook was hyper-targeted. Only students could sign up, which meant we had direct access to the exact audience we wanted. The results were incredible. The campaign performed so well that Facebook made fifty thousand dollars from us in just a week. That was the moment they realized how valuable advertising on their platform could be.

The speed of testing and iteration changed everything. In offline advertising, getting a billboard campaign approved could take six months. By the time it was live, you had no way of knowing whether it actually worked. With Facebook, I could run an experiment, check the data in a few days, and adjust accordingly.

That was when I realized that traditional ad agencies, who operated on long planning cycles, were not built for this new reality. The ones who could rapidly test, learn, and optimize would always win. That lesson has stuck with me throughout my career.

This has been one of the most persistent realities in marketing. No matter how advanced a company gets, spreadsheets always make a comeback. When I started at PartyGaming, we were already using macros to pull data from different sources and match revenue to acquisition costs. This was before Google Analytics. There were internal tracking systems, but they were rudimentary. The only way to get a complete picture of performance was to manually collect data and structure it ourselves.

Fast forward years later, and I saw the same pattern repeat at every company I worked with. Businesses would invest in BI dashboards, build central data warehouses, and roll out expensive reporting tools. But when it came down to day-to-day decision-making, teams would always revert to spreadsheets. It was the only way to get the exact view they needed without waiting for a data team to build a new report. And this is not because data teams don’t want to help them, but because the data teams themselves are swamped with requests from across the organisation.

The fundamental issue is that dashboards are static, and marketing is dynamic. When you set up a BI tool, the data team asks stakeholders what they want to see. The team defines metrics, builds reports, and delivers a polished dashboard. In fact, many data teams spend a lot of time in this process of requirements gathering. But this only solves the problems marketers had at that specific moment. 

Marketing changes constantly. Budgets shift, new channels emerge, campaigns underperform, and suddenly you need to analyze data in a way the dashboard wasn’t built for. That’s why marketers go back to spreadsheets. They can pull in raw data and manipulate it on their own terms without being bottlenecked by predefined structures.

Ad-tech is a brutal space. There’s a graveyard of companies that built solutions on top of ad platforms, only to get crushed when those platforms changed their rules. I saw this firsthand when working with campaign management tools for Facebook and Google. At one point, a company called Nanigans was a leader in Facebook advertising. They had a proprietary pixel that helped advertisers optimize based on conversions. Then Facebook rolled out its own version, and Nanigans lost its core advantage overnight.

Google has done the same thing for years. They let third-party tools flourish for a while, then copy their functionality and make it native to their own platform. Marin Software, Kenshoo, and many others struggled for this reason. The second an ad platform decides to absorb your features, you’re done.

That was a key lesson when we started Clarisights. We didn’t want to build another optimization tool that could be wiped out by a platform update. Instead, we focused on something that is structurally resilient—data aggregation and processing. No matter how marketing evolves, companies will always need a way to bring their data together in a way that fits their business.

The biggest difference is that we don’t just visualize data. We process it. Actually, we process millions, if not billions, of rows of data every day. Traditional BI tools focus on reporting, which means they take in data and present it in a structured way. But that doesn’t solve the core issue marketers face, which is making sense of raw, fragmented data in a way that enables decision-making at speed. 

For example, a marketer running paid campaigns needs to see how Facebook, Google, and TikTok ads are performing together. But each of those platforms defines a conversion differently. A sale reported in Google Ads might not match what Facebook attributes, and neither might match the company’s internal sales system. This is where Clarisights comes in. Instead of just showing these numbers side by side, we process the data to unify it. We normalize spend, conversions, and revenue so that teams get a coherent view instead of a collection of disconnected metrics.

That’s why our customers tend to be large, sophisticated marketing teams that operate across multiple geographies and channels. At that level of complexity, you can’t rely on native platform reports or off-the-shelf BI dashboards. You need a system that adapts to how you operate, not the other way around.

When we started, we thought our ideal customers would be early-stage companies—startups that needed better reporting but didn’t have a data team. It seemed logical. These companies had fragmented data, relied on spreadsheets, and needed automation. But we quickly ran into a problem.

Startups don’t stay startups for long. If they succeed, they hire a data team. And the first thing that team does is rip out third-party tools and build everything in-house. If they fail, they shut down or cut budgets. Either way, they churn. We spent months trying to land these companies, only to watch them leave as soon as they raised a new funding round.

The real turning point came when we connected with Delivery Hero. At the time, we had small customers using Clarisights to consolidate ad data, but nothing at massive scale. Delivery Hero had thousands of ad accounts across multiple platforms and regions, and they needed a way to make sense of everything in real time.

The first call was brutal. Their VP of Performance Marketing flat-out told us, “I never buy third-party tools.” He explained that everything was built in-house because they operated at a scale that most tools couldn’t handle. But when we showed him that Clarisights wasn’t just another dashboard—it was a data processing layer that could handle their complexity—his mindset shifted.

That was our wake-up call. Clarisights wasn’t for small companies trying to get their first reporting setup. It was for massive marketing teams struggling to operate at scale. Once we embraced that, everything changed.

Scale introduces problems that don’t exist when you’re spending a few million a year on ads. Delivery Hero was running thousands of campaigns across multiple platforms, geographies, and brands. Every country had different performance metrics, different bidding strategies, and different reporting needs.

A traditional BI tool couldn’t handle that. The data pipeline alone was a nightmare. And it doesn’t matter how hard data or BI teams try to make it work, they are often hobbled by the tech stack. They needed a system that could automatically consolidate all of this into a structure that made sense for their team. That’s where Clarisights fit in.

To give you an idea of scale, when we first started, we were pulling in data from maybe ten Facebook accounts and a handful of Google Ads accounts. Delivery Hero had over two thousand Facebook accounts and fifteen hundred Google Ads accounts. We weren’t just scaling up our infrastructure; we were fundamentally rethinking how we processed and structured data.

Everything has to be built with flexibility in mind. In a smaller company, marketing data can be fairly standardized. You have a handful of key metrics, and your reporting is relatively simple. But as a company grows, everything gets more complicated.

Marketing teams don’t just look at “ROAS” or “conversion rate.” They start tracking things like contribution margin, geo-level performance, and channel-specific attribution rules. One team might want to analyze data by campaign type, while another wants to break it down by city. If you force everyone into a predefined structure, the tool fails.

That’s why Clarisights doesn’t just ingest data—it lets teams reshape it in ways that make sense for them. Instead of being locked into a fixed dashboard, they can manipulate data dynamically.

We had to rebuild everything. When we first started, our system was designed for small teams. It could pull in data at a reasonable pace, but it wasn’t optimized for real-time analysis at scale.

With Delivery Hero, that wasn’t an option. Their teams were making decisions in real time. They couldn’t afford to wait for batch processing or deal with laggy reports. We had to rethink our entire pipeline to make sure data was fresh, accurate, and always available.

This was also a moment where we realized that most BI tools aren’t built for performance marketing. Traditional BI systems optimize for general business reporting, where a few hours of delay might not matter. But for marketing teams spending millions per day, waiting for data is unacceptable.

The other challenge was data integrity. When you’re processing data from thousands of accounts, any inconsistency can break reporting. We built automated validation systems to ensure data accuracy at scale. If a platform API goes down, Clarisights detects the issue and alerts teams immediately, rather than silently letting bad data creep in. We realized very quickly that we had to simultaneously solve the marketer’s need for speed, but also the data team’s need for governance.

Everything. Before, we were selling to small teams who needed an easier way to track performance. Now, we were selling to CMOs and heads of performance marketing at billion-dollar companies. That required a completely different approach.

Selling to enterprises means proving value at multiple levels. The CMO wants to know how it helps drive business impact. The VP of Growth wants to know how it improves efficiency. The performance marketing team wants to know how it saves them time. And the data team wants to know it won’t break their existing stack.

We learned that enterprise sales isn’t about “convincing” someone to buy. It’s about aligning with their existing pain points and showing that your product is the missing piece they’ve been looking for. Once we nailed that, our pipeline grew fast.

The companies that fail in ad-tech are usually those that build directly on top of platforms instead of solving a structural problem for businesses. Clarisights is not another optimization tool or bidding platform. We don’t make decisions on behalf of Google or Meta. We give companies control over their own marketing data, something platforms don’t naturally provide.

If you look at how marketing teams operate, they don’t just work inside one ad platform. They run campaigns across multiple channels, report performance to finance teams, and adjust strategies based on business goals. No single platform will ever give them a complete picture because each one is biased toward making itself look good. That’s why Clarisights is built as a data layer that works across platforms. No matter how the ad ecosystem changes, companies will always need a way to unify, process, and analyze their data in a way that reflects their actual business performance.

BI tools are great for structured reporting, but marketing doesn’t operate in a structured way. Campaigns change daily, budgets shift, and measurement methodologies evolve. Static dashboards can’t keep up with that level of change.

What makes Clarisights different is that we don’t just display data. We make it usable. Most BI tools take predefined data sets and visualize them, but they don’t solve the underlying problem of fragmented, inconsistent data. Clarisights processes data dynamically, letting teams reshape and analyze it on their own terms without needing to rely on SQL or wait for the data team to build custom reports.

Automation is great for micro-optimizations, but marketing leaders still need control over their decision-making. A lot of companies talk about “solving” attribution, but the reality is that there is no universal solution. Measurement will always involve trade-offs.

Some teams will rely on MMM for budget allocation. Others will use incrementality testing. Some will still use last-click data because it’s simple and actionable. The mistake is thinking that one model will work forever. The best teams constantly adapt their measurement strategy as the market shifts.

That’s why we built Clarisights to be flexible. We don’t force a single measurement approach. Instead, we allow companies to bring in multiple sources of truth—platform data, internal finance numbers, attribution models—and compare them side by side. The future of measurement isn’t picking one model. It’s being able to evaluate all of them dynamically.

I would start with first-party data. Any company that relies entirely on platform-reported metrics is giving up control. You need your own source of truth, whether it’s CRM data, backend revenue numbers, or a combination of both.

The second priority is structuring data in a way that enables fast decision-making. Many companies collect massive amounts of data but struggle to use it effectively because it’s scattered across teams.

Finally, I would design the stack to be adaptable. The biggest mistake I see companies make is locking into a single way of measuring success. What works today might be obsolete in two years. The best marketing teams are the ones that can evolve their measurement strategies as the landscape shifts.

One of the great lessons we have learnt over the years is that far too many organizations spend their time and effort trying to build a single source of truth. Often this leads them down the path to a warehouse-driven architecture. But, as I have said before, when it comes to performance marketing there is no single source of truth. There are multiple sources of truth. Each platform is a source of truth. Your internal data is another source of truth. What you really want to do is a single source of insight: is there one place I can go to that will give me insight across platforms? Your tech stack is just a means to an end, and that end should always be data-driven, actionable insight.

Stay tuned for more with our upcoming podcast!