A business can have a very expensive argument with itself. Marketing arrives with one number, finance brings another, and operations has a spreadsheet that would prefer not to be questioned. Krishna Poda has built a career around getting these departments to agree on what happened before they decide what to do next. The arithmetic is only the beginning. People must agree to use it.
The co-founder and CEO of Saras Analytics works at the junction of enterprise data systems and the considerably less orderly lives of growing ecommerce brands. His subject is profitability, his material is fragmented information, and his recurring complaint is that possessing reports can become a remarkably effective substitute for making decisions. A dashboard may be handsome company. It still cannot run the meeting.
A database plan takes a different turn
In 2006, Poda moved to the United States intending to pursue a doctorate in databases. The plan changed, and he went into work instead. Data remained the thread. Before Saras, he was an enterprise architect at Oracle, where his work involved integration, warehouses and business intelligence. The enterprise world gave him a close view of what companies could do when they had the resources to organize their information.
It also left him interested in the businesses outside that comfortable circle. Smaller companies had questions worth answering, too. Their budgets and available teams were another matter. Poda's eventual business would try to carry some of the capabilities of enterprise analytics into the middle market, where assembling the machinery could be an undertaking in its own right.
He has a dry way of describing the decision to start. Looking back on the leap, he called himself “stupid enough to get into it.” There was a broad intention before there was a precise customer profile. The admission gives the founding story some welcome texture: even someone who sells the benefits of informed decisions can begin with an incomplete spreadsheet.
“So fear not so much, but a huge learning curve.”
Krishna Poda, on becoming a founder
Three cousins, one shared problem
Saras began in 2016 with Poda and his cousins, Abhiram Kolli and Balaji Kolli. They brought different professional histories to the same problem. Abhiram had built data products for investment teams at Credit Suisse. Balaji's experience crossed analytics and business development. Poda brought the Oracle background. A family connection supplied the introduction; their work supplied the common language.
The early consulting work exposed a frustration they could recognize from inside the job. Collecting information from different systems produced repetitive manual work. The team decided to build software to address it. Daton became the data pipeline; Halo was an early analytics and insights product designed for ecommerce businesses with small analytics teams, or none at all.
By August 2021, Saras reported a team of 110 in Hyderabad, profitability and more than $2 million in annualised recurring revenue for the 2021 financial year. These are historical markers, rather than a description of today's balance sheet. They show the business had moved beyond an idea into products, customers and the obligation to keep everything working.
Software and consulting grew together around the practical work of making business data usable.
Bootstrapping matters to Poda's account of the company. He describes building a profitable business as an accomplishment in itself. On his own writing profile, he connects that experience with a continuing interest in sustainable unit economics. That phrase has little glamour at a dinner party. It becomes much more attractive when the bills arrive.
The profit hiding behind the sale
Poda's public writing keeps returning to the economics underneath growth. His stated interest is helping omnichannel brands understand their unit economics by joining information that lives in separate systems. The ambition is practical: a founder should be able to connect activity across the business with the money that activity produces or consumes.
In a contribution to the Operators newsletter, he sets out how business variables affect one another. Customer acquisition costs influence the profit and loss account. Discounts change customer value. Fulfilment expenses move margins. More sales can also mean more cash committed to the business. A decision that appears attractive in one department can send an unwelcome invoice to another.
His proposed use for AI includes exploring those interactions through scenarios. What happens when acquisition becomes more expensive while discounting changes? Which channel mix copes better with higher fulfilment costs? These are questions he wants businesses to test using their own information. The value would lie in seeing the consequences before committing the money.
That helps explain the breadth of his conversations. For the company's podcast series, he has discussed attribution with Fairing co-founder Matt Bahr and Amazon agency operations with Adverio's Mike Danford. Attribution and catalog management have their own specialist vocabularies. Both ultimately meet the same commercial question: where should the next hour or dollar go?
He has a classification for the confusion
Poda organizes the journey toward useful analytics into five stages. It starts with instinct, moves into dependence on reporting, then encounters departmental silos. Later stages involve experimentation and alignment across functions. The model gives a name to an awkward middle period: a company may have accumulated a great many reports without acquiring a shared understanding.
In his account, the next step depends on how people use the information. Teams should be able to test assumptions and consider how a proposed move affects other parts of the company. The progression runs from describing events toward examining consequences. A report earns its place by helping someone choose a course of action.
- 01Instinct
- 02Reporting
- 03Departmental silos
- 04Experimentation
- 05Shared alignment
A conceptual progression, not a measured score or growth chart.
There is a recognizable generosity in giving operators a model they can inspect. It makes the discussion more concrete than an invitation to buy yet another tool. Someone can ask where their own company sits, which recurring task consumes time and what would have to change to move forward. The conversation has somewhere to begin.
A little less theatre in the meeting
His writing on a “culture of intelligence” brings the responsibility back to leadership. He divides the work among people, process and technology. Leaders need to ask questions, teams need the skills to interpret information, and regular conversations need to produce action. Buying software covers only part of the assignment.
That emphasis gives his technical work a social dimension. Shared definitions matter because different functions can otherwise look at the same business and disagree about its performance. Poda wants marketing, sales, operations and finance to work from an agreed foundation. An elegant system has limited usefulness if everyone continues to bring a private version of reality.
He also writes about the habits that make a workplace function. His LinkedIn articles include a piece on the signals of good culture and the costs of bad culture, alongside another on why analytics takes a team of specialists. In a recommendation, former Oracle colleague Alex Baird describes his willingness to teach and eagerness to learn. It is a colleague's assessment, and an apt complement to a career spent translating between technical and commercial teams.
The invoice deserves a second look
Some of Poda's examples are pleasingly unromantic. In his writing about hidden costs, he applies the known-and-unknown framework to shipping, customer churn, inventory and marketing. The exercise asks which costs are understood, which questions remain unanswered and which patterns might already be present in the information without anyone noticing them.
Shipping offers a useful illustration. Carrier fees and standard rates may be familiar, while unnecessary expedited orders, long routes or charges buried in bulk invoices receive less attention. He advocates periodic audits and connecting information across systems. The implication is almost mischievously ordinary: before searching for a spectacular new growth tactic, examine what the business is already paying for.
Inventory creates another version of the problem. Stock ties up working capital; shortages carry their own costs. His interest is in bringing those operational consequences into the same conversation as marketing and sales. The business gets easier to understand when the departments stop treating their numbers as separate autobiographies.
AI arrives, and the foundation still matters
By March 2026, Poda was discussing Saras IQ with Ridge CEO Sean Frank and HexClad president Jason Panzer on the Operators podcast. The conversation covered employees asking questions of business data, agent workflows and the role of governed information. IQ was described as an ecommerce-focused layer built on Claude. The setting had changed from enterprise reporting to conversational software; reliable underlying data remained central.
His writing on a single source of truth explains why. Departments need consistent definitions and information organized so that business users can ask useful questions. An AI interface adds another reason to get those foundations right. A quick answer becomes commercially useful only when the organization can understand and trust the numbers behind it.
The future he discusses includes agents interacting with other platforms and software handling recurring work. Those are ambitions under discussion, rather than promises that every company already operates that way. For Poda, the route toward them still passes through the less photogenic work of preparing and governing information.
Across his career, the question has acquired new interfaces, new products and a wider audience. The underlying concern remains wonderfully familiar to anyone who has sat through a meeting about conflicting figures. Before a company lets its machines answer more questions, it needs a dependable account of its own business. Poda has made that stubborn bit of housekeeping his occupation.