Behind Infosys’ struggles to deliver the income tax project

[By Zorro2212, CC BY-SA 3.0, via Wikimedia Commons]

Everyone has a plan, till they are punched in the nose 

- Mike Tyson

It was supposed to flow like a dream. 

The project the IT services major won from the government was transformative, promising to make things faster and better for users. The ministry was keen to roll it out. For the politicians, a successful launch could create goodwill, improve rankings in the lists that think-tanks regularly publish, and maybe even win some votes in the next elections. All these would have a rub-off effect on the company. More government projects. More revenues. Happy investors. Above all, a chance to demonstrate tech can make lives better, leading to a bigger market for its services.  

However, the development took too long, and there were delays, which annoyed the government. When the project finally rolled out, it didn’t pan out as planned. Government employees seemed to be miffed and let the media know about their feelings. Users complained about glitches, about the fact that it was taking more time to get things done. There were protests in some places. Instead of basking in the strengthened relationship with the government, company executives were fretting over reports that said it might even be booted out of this project. The punches on the nose seemed to land hard and fast. 

One might be tempted to think that the description above is that of Infosys. How the iconic IT services company, the fourth ever Indian firm that breached a $100 billion market cap, is struggling to deliver the all important Income Tax portal project on time. How its sheen is paling next to its rival, TCS, which seemed to have captured everyone’s heart with the passport office project. 

But it’s not about Infosys. The description is in fact about TCS’s famed passport project. Only, all these events took place between 2009 and 2012, during the development and initial months of rolling out. It took time for the new system to work, and for it to become a case study of how to get government projects right.

The striking thing about the Infosys - IT portal launch, though, is what Govindraj Ethiraj, founder of Boom Live, termed as an ‘information vacuum’ in a recent panel discussion anchored by Barkha Dutt. We don’t have enough information to understand what went wrong and who is responsible. The government hasn’t given any information beyond tough statements and deadlines, and Infosys has been silent. (It declined to be interviewed for this piece). Last month, India’s finance minister Nirmala Seetharaman summoned Salil Parekh, the Infosys CEO, demanding an explanation for the glitches. A deadline was set: September 15. While users have reported some improvement in the system, the glitches are still there. A few days earlier, the Income Tax department postponed the filing to December 31, implicitly accepting that the portal is not ready, and deepening the concerns about the lack of information. 

However, as the TCS experience highlights, there are deeper issues with IT services companies taking on government projects. While Infosys will eventually fix the problems with the IT portal, these root causes will persist and bog every IT services company that takes up similar projects. It’s driven by the needs of the IT companies, which also face significant constraints. The key is to balance them.

• One, Indian IT services need large government projects for their topline, but it’s hard to make profits from the projects. 

• Two, it’s important for Indian IT services companies to gain expertise in large complex projects, but the learnings are hard to transfer. 
• Three, these projects can be prestigious but can also cause huge reputational harm.  

As scope creeps, margins shrink

For a long time, the success of the Indian IT sector was defined in terms of the success of IT exports. Exports grew as businesses in the west offshored work to India and got things done at a lower cost. The salaries in India more or less have remained a third of what it is in the US and Europe. Indian IT companies built their business model essentially to fulfill this demand. It was by no means easy. In the early days Satyam had to set up a physical development centre in the US, where it had Indian engineers, working in Indian time, and delivering projects without setting a foot in the customer’s office just to demonstrate it can be done. The model was built around that.

However, things changed by late 2000, especially after the global financial crisis. Many companies saw they had to diversify the revenue pool—to Europe and Japan, and they also realised that there was a huge opportunity right in the backyard. Companies such as TCS and Wipro which were already operating in the market through acquisitions, subsidiaries or group companies (CMC, Wipro Infotech) suggested there’s business there. The Indian government had also become far more open to spending on transformative technology projects.  

In short, the money was huge. Consider the size of some of the projects since that time. India Post's IT modernisation was pegged at Rs 4,909 crore to cover 1.55 lakh post offices. The passport automation project was Rs 1,000 crore. The project in the news—Income tax filing system that Infosys took up—was Rs 4,242 crore in 2019. The market was not all about big projects, it had depth too. Krishnakumar Natarajan, former CEO and co-founder of Mindtree, pointed to ABM Knowledgeware, a Mumbai-based listed company that has been taking up local government IT projects and is also profitable to boot. 

However, the government is a different beast altogether. While the budgeted numbers seem big, the cost structure of Indian IT companies, designed mostly around dollar revenues, don’t result in high margins. In 2015, Infosys founder NR Narayana Murthy said that Infosys hadn’t made a single rupee from government projects. Getting money from the government can be tough too. A senior executive, who led an IT project for a state owned corporation, said while the money might eventually come to you, you have to run from pillar to post to get them released. It has an opportunity cost for the people doing it, because they would rather be working on their next projects.   

Those who worked in such projects say the problems start even earlier. The government’s processes are not designed for transformational projects either. They follow the L1 method, giving the projects to the lowest bidder. It makes sense for commoditized products and relatively simple projects with lesser moving parts. The L1 method is simple to implement and easy to explain to pesky government internal auditors who might question the bureaucrats on the choice of a vendor. But it doesn’t work with complex, transformational projects with many moving parts where an agile approach would fare better.

Yet, IT companies still sign up for projects confident that they can make money by efficient execution. Tyson’s punches, then, come in the form of scope creep. A typical bureaucrat—not only in government organisations but even in large private sector companies—believes that he can get maximum bang for the buck by adding as many features as possible. In a government set up, as a document moves from desk to desk, more features get added till it gets too unwieldy. S Ravichandar, who has worked with the government on many projects, including tendering processes and Jawaharlal Nehru National Urban Renewal Mission  (JNNURM), etc said it is in the nature of the bureaucracy to add more features. He cited the example of a project he worked on, the urban renewal mission, where the initial draft had just four conditions and a simple structure for funding from the centre to the state. But as each department weighed in, he said, “by the time that train left 20 more bogies were attached to it”. A recent report in ET Prime points out that the income tax filing process had to handle a checklist with over 85 items, which increased the complexity of the project.

Often such additions happen even after the project is signed off, and when a new joint secretary or an additional secretary assumes the post. 

A project manager in a successful tech company will do the opposite. Sharad Sharma, co-founder of technology think-tank iSpirt, who has worked with the government and with tech companies, said on a Zoom call that he can think of five features that Zoom should incorporate in the product. “I have no doubt people inside Zoom have thought about them as well. But the job of a product manager is to select, prioritise. His job is to say no, so what’s shipped works and serves the purpose.” It might be easy for a Zoom product manager to say no. However, it tends to be harder for an IT services company to say that when the request comes from a high-ranking government official. That is where expertise comes in. 

Transformative capabilities and institutional amnesia

Indian IT companies’ key innovation was to unbundle software development processes, get them executed in different centres, at different time zones, and stitch them all into a working application. This helped them bag multi-million-dollar projects, and Fortune 500 clients. Yet, even in the biggest of the assignments, the key skills they brought in was project management—following the functional and technical specification, and delivering it on time. The design of complex systems was often done by others. Taking on a complex project at scale poses a different set of problems. 

The Indian government was trying to achieve something ambitious with income tax filing. It was not just about filing, the goal was to provide the refunds to tax-payers within a day (it took 63 days earlier). It’s no easy task, because to provide it on the same day, the system has to ensure that all the data is right. Contrast this with transferring money through a UPI app. The app essentially has to connect five entities: sender’s bank account, sender’s service provider, receiver’s bank account, receiver’s service provider, and NPCI’s unified payment interface itself—while also checking if the sender has enough balance in his account. In the case of income tax filing, the moving parts are higher by an order of magnitude.

Now, taking up large transformative projects in India meant developing those capabilities. Ramamoorthy Ramkumar, former managing director at Cognizant’s Indian operations, referred to a joke in Chennai that goes like this: ‘if you know how to drive in Ranganathan Street (a narrow, busy street in T Nagar, Chennai's shopping district) you can drive anywhere’. That applies to Indian government projects too. If a company has learnt how to design and implement a large complex Indian government project, it can pretty much do everything. 

There are many reasons why. The customer can be very demanding. The design has to be frugal and has to cater to a diverse set of people, whose skill levels on a scale of 100 can vary from 0 to 100.

For many companies, N Dayasindhu, CEO of Itihaasa, said, “if you do a good project in India, it opens doors for several geographies with untapped opportunities—Africa, Asia, Latin America”. India is perceived as a country that has mastered frugal innovation, which would work well for other developing countries, rather than solutions that were developed for countries which had already invested heavily in infrastructure. For example, the World Bank has been recommending digital identity for African countries, and any IT services company that has worked as a vendor to India’s massive digital identity project will have an advantage. 

However, it also comes with a risk. The risk is well captured by a key criticism leveled at Infosys. Infosys has been criticised for glitches in the project to handle the website of the Ministry of Corporate Affairs (MCA), which it took over from TCS. Many users complained they couldn’t access documents from the site. Many couldn’t even log in. Similarly, Infy's implementation of the GST project was also mired in glitches. 

Here’s the moot point: Why did Infosys—and the government—not learn from past mistakes?

The answer lies in the very nature of how these complex projects stack up. One doesn’t get to make a cookie cutter that will make the next project a breeze. Each project is unique. The documentation doesn't capture the entire complexity. The transition of the MCA project from TCS to Infosys exemplifies that. The handover got many things right. After its contract ended, TCS transferred the entire documentation to Infosys. They had factored in time for the transition. The incentives were right—Infosys made money based on the performance of the site and not merely handing over the project. Infosys didn't fundamentally change anything after it took charge. Yet, there were glitches. 

People within both Infosys and TCS agree that knowledge sharing was not complete—because there are inherent limitations. Mindtree co-founder Krishnakumar says during a project, the team absorbs a whole lot of tacit knowledge. Some of them are about knowing the right person—who would either solve the problem, or suggest an alternative so the issues don’t flare up. That cannot always be transferred. So, for some time, there will always be a learning curve during which things can go bad.

This is further complicated by the fact that while the organisations might be the same, people within organisations change. Software engineers and managers move on to new projects, or leave the organisation. And the project team and the companies are deprived of tribal knowledge, crucial to get things done. Similarly, in the government, secretaries change. While files are permanent fixtures in all government offices, their primary purpose is often to fix responsibilities. They capture even less information than a typical documentation system in an IT company does. 

Even globally, government projects have struggled with similar issues. For example, www.healthcare.gov was down within two hours of launching the site, and other issues emerged too. A study published around that time by Standish Group found that 94% of all US federal government projects were unsuccessful, facing cost and time overruns, and failing to meet user expectations. Four out of 10 failed completely. 

While the numbers can be edifying, at a company level, institutional learning does take place, and eventually turns into organisational capabilities. Leaders, processes and entire organisations evolve—and the subsequent projects benefit from them.

Boasting rights and curses

India’s first attempts at building a metro rail project were mired in problems. Kolkata Metro faced huge cost and time overruns, and was so affected by several accidents that other states were reluctant to take up a metro project for several years. Till the Delhi Metro project came up. While its execution surprised many, it was not an overnight success. E Sreedharan had spent his lifetime in railways, including being a part of Kolkata Metro project, repairing Pamban bridge, building Konkan Railways, all of which helped him execute Delhi Metro better. The execution, and the real benefits it offered to people, helped in the re-election of its chief minister Sheila Dixit.   

Indian IT companies seek out transformative projects for the prestige it gives. For the politicians it is more than boasting rights. If the project delivers on the promises, it means gaining goodwill of the public, which could eventually turn into electoral votes.

It's not restricted to people who are on the top, but also to people who are lower down in the hierarchy. Most of those who contributed to the Aadhaar project wear it as a badge of honour in their resumes.

A project manager working in a leading IT company said it’s tough to get people into an India project especially when they are in the early stages in their career because the promise of onsite work and lure of Fortune 500 brand names are bigger than having to work in India. However, for those who have gained some experience already working on large complex projects can give a leg up.

The mood in the country has also changed. While IT companies were praised for bringing in export revenues and generating jobs earlier, there are expectations around doing something for the country. Getting on to these projects helps in that narrative. 

However, the visibility comes with its own risks—especially if it’s a project that directly impacts people’s lives. When things don’t go well, it results in public shaming. 

Citizens complain to politicians and politicians—who are more aware of the issues than they are credited to by the media and the public—make attempts to stand up for the citizens by getting tough on the companies. It’s this that resulted in the summoning of the Infy CEO to the finance minister’s office.

There is also additional risk, as evidenced by the emotional debates around the accusations of Infosys being anti-national by RSS-affiliated magazine Panchajanya. In a cover story, the author accused Infosys of scuttling the government’s technology projects. While RSS has distanced itself from the comments, in some ways it has normalised this type of criticism.

The debating cycle has also moved the problem to the top of the mind of people, that one cannot think of Infosys without thinking about the income tax portal glitches. While this is likely to fade away in the coming days, right now it is in the viral cycle. Recently, when Nandan Nilekani endorsed The Solutions Factory, Arun Maira’s latest book, on Twitter, the tweet received three replies—and all the three questioned him on Infy’s IT filing project.

Just as a successful project can make an imprint on everyone connected with the project, criticisms can also have the same impact. As a result it has raised macro questions on the IT sector industry itself, and micro questions on the employees involved in the project. “Even your cousins and uncles, who might generally look up to you for being a world traveller, start wondering what you are up to,” an engineer working in another company, but has gone through a similar experience, said.

So, what can companies do, given all these constraints, given that one cannot change the bureaucratic way of working overnight? There are three common themes that emerged from the conversations.

Focus on the upstream: The Infosys glitches are seen as the system not doing rigorous user acceptance testing. Mohandas Pai, the former CFO of Infosys, said that a team of testers comprising chartered accountants and startups in the domain should have done complete testing before it was rolled out. There is merit to the argument. However, testing can identify the problems later. The key is to invest enough time upstream. A person who has seen TCS work on a government project was surprised to see the time the company spent in the early stages of the project getting feedback and buy-ins. In his book TCS Story, former CEO & MD C Ramadorai writes that when they won the National Stock Exchange project, the entire team underwent training on how the stock market works.

Build lateral capabilities: There is a term that's popular in IT services companies—the second mover advantage. The second mover learns from the mistake the first mover makes, and does a better job. The second mover also comes in when the business model is clearer and therefore need not invest in capabilities that don't have ROI. If TCS is relatively better at executing some of the government projects, it is because of the investments it made earlier in projects that others wouldn't touch. Ramkumar pointed to projects such as traffic management and forestation that a TCS wing did for the government in the early 80s and 90s. It gave the company a feel for large complex problems, and the capabilities it built during the peacetime often comes to help during the wartime.  Infosys and other companies that came later into the game in the 2000s are still developing and institutionalising the mindset and tools for such projects. It’s not rocket science, Krishnakumar said, but companies have to do it.

Be transparent: It’s true for any project, but especially so for government projects, which use public money, which touch citizens’ lives—transparency is the key. It brings all stakeholders to the same page, and there will be no surprises, an IT executive said.