India's could-be flagship software product company

In this series, Sramana Mitra shares chapters from her book Vision India 2020, that shares 45 interesting ideas for start-up companies with the potential to become billion-dollar enterprises. These articles are written as business fiction, as if we’re in 2020, reflecting back on building these businesses over the previous decade. We hope to spark ideas for building successful start-ups of your own. 

At the end of the first decade of the twenty-first century, India was riding high on outsourcing. But for all its advancements, India was yet to produce a global software company with either original technology innovation or productization. Barring a few exceptions, the glut of venture capital chasing India found it difficult to be deployed. There was way too much money and way too few deals. Instead, tech-sector VCs were diverting capital to retail, real estate, hotels, and other non-tech sectors.

Our response: Taxonomy was founded with the ambition of becoming no less than India’s flagship software product company.

To determine its market focus, we looked at the largely unaddressed problem domain, unstructured data management. Within the context of business intelligence, this was an open problem that I identified as the premise for a substantial company.

The most potent competitor was Europe’s second-largest software company, Autonomy. At the time of our initial research, Autonomy’s numbers were impressive. Revenues were up 47% to $500 million in 2008, even as the global recession raged. Their pre-tax profit was $208.8 million, up 84% from 2007. They had 20,000-plus customers, 400-plus value-added resellers (VARs) and, 400-plus OEM companies with licenses to more than 500 products. And most importantly, the size of the problem they were addressing continued to escalate. Enterprises were facing a deluge of unstructured information, with 80% of total business data falling into this category. Gartner observed that the volume of this data was doubling every month. Most of the business intelligence market, however, was based on structured data management and manipulation technologies. So as the amount of unstructured data continued to grow, so too did the challenge for the modern enterprise in trying to understand and extract meaning from it.

I had substantial experience in the enterprise software space, having worked both with startups and large ISVs like SAP. In fact, when I worked with Nimish Mehta, an SVP at SAP in 2006, I had the opportunity to examine enterprise search from their vantage point. With this knowledge base, I minutely evaluated the opportunity to take Autonomy on with a core suite of search technologies, and I concluded that offering it to the world of developers as free, open-source building blocks would dramatically change the rules of the game.

India did not really have experienced product managers when we started, which meant we had to recruit qualified people from Silicon Valley and move them to India. The 2008 financial crisis and resulting recession proved unquestionably beneficial for us from the recruitment standpoint, as we moved hundreds of very experienced product people – managers, engineers, architects, algorithms experts – who had found themselves out of work in America. And they, in turn, groomed the next generation of talent, eager to learn, anxious to move beyond the achievements of their predecessors. In fact, many of the product managers who came to work in India were not even of Indian origin. They came, instead, towards a world-class career opportunity.

Our search technology suite encompassed various different flavors of algorithms, recognizing the long-overlooked fact that different applications needed different algorithmic and architectural underpinnings. We studied applications ranging from compliance, archiving and retrieval, product catalogs (including recommendation engines), tacit collaboration, knowledge management, customer support, and content management. The building blocks included automatic classification and taxonomy generation, clustering, natural language processing (NLP), sentiment analysis, various types of query processing and management, collaborative filtering, and speech analysis. Both architecturally and algorithmically, we achieved a world-class product suite.

We introduced these blocks to the public in 2010. A highly targeted Twitter and blog campaign generated buzz about our arrival. One such leveraging point of this campaign: open-source guru Brian Behlendorf. Brian, a member of our advisory board, tweeted about Taxonomy to his followers, and immediately, 25 of the most influential open-source bloggers ran with the thread.

Within two years, we had 3.3 million downloads by over 200,000 active users. Almost 40,000 applications were built on top of our building blocks, and we generated 500 paying customers following the classical commercial open-source business model – free basic software, with paid premium modules, as well as paid support and training. Open source allowed us to be more a marketing-oriented business than a sales-oriented business – although we retained both a substantial Indian telesales organization, as well as a direct sales force for deals above half a million dollars. In many of these deals, we competed head-on with Autonomy and won. Word was spreading in the developer community that an inexpensive yet comprehensive technology option, capable of bringing unstructured data management applications to market rapidly, had emerged in the open-source domain.

We also took advantage of our generous India operation of 3,000 people, offering services alongside our products. Often, our customers would not only buy our products, but also a development team from us. This allowed us to create a tremendous exit barrier in our customer base, which Autonomy could not compete with. It also enabled us to innovate in every project, while also learning in close cooperation with our customers what innovations would further enhance our core engines. For example, a customer in Romania working on a mobile vertical search application brought to our attention the specialized requirements of mobile interfaces, which meant we had to process more in the back end. Other key learnings were as varied as device-specific nuances and domain specific requirements, even social media–elated data structures –all of which broadened not only our competencies, but also our reach.

We grew to $30 million in 2012, $240 million in 2015, and all the way to $1.2 billion in 2020. We maintained a product-service mix of 70%:30%, ensuring us an immensely attractive P&L structure with 20% net profit. All in all, our market cap in 2020 grew to $7 billion.

Some of our OEM partners have also built very successful independent software companies. One of them, an e-commerce personalization ISV, is now doing $120 million a year in profitable revenue. Another compliance vendor is doing $76 million, catering mostly to global 2000 enterprises. A third customer, a self-service vendor, is already up to $67 million. This year, we look forward to rolling up a large chunk of the application layer developed on our platforms. Each of them already includes hundreds of our developers as part of their extended engineering organizations, so integration will be relatively smooth. But we can aggregate almost a billion dollars in additional revenue by assembling 27 of these segment-leading ISVs.

It has been a very interesting journey so far, leading the movement of productization in the Indian software industry. A fundamental industry-level cultural shift is no small task to take on. Of course, the strength India has traditionally possessed in services is something we and our compatriots leveraged extensively. But the open-source model has now been added to the mix, proving wildly beneficial and scalable for many of the companies now ramping up alongside Taxonomy.

Some 13 years ago, I wrote my highly controversial column for Forbes called "The Coming Death of Indian Outsourcing,” taking to task the industry’s captains for being complacent and lacking imagination.

Well, today in 2020, India has moved unambiguously out of that uninspired rut. A thinking, thriving Indian software industry has emerged. Products are being built. Innovation is rampant. And at the head of this renaissance, Taxonomy is striding ahead.