Building the next generation 'Tricentis': Thesis on packaged application testing

Packaged applications like SAP, Salesforce, and Workday have been indispensable for enterprises, enabling streamlined operations across key functions such as finance, HR, and CRM. Despite continuous reference to these players as “legacy”, “not user friendly”, they are virtual monopolies. Market caps of Oracle, SAP and Salesforce have increased by 25% over the last 12 months. 

These pre-built software solutions offer high scalability and configurability, allowing organisations to adapt them to their specific needs. However, the implementation of these applications is far from straightforward, requiring significant customization, integration, and rigorous testing before going live.

Take SAP, for example—its $37 billion annual revenue supports a massive downstream ecosystem valued at $100-200 billion annually, covering implementation and ongoing support costs. For every product in its portfolio, SAP collaborates with localised implementation partners tailored to specific customer personas. Global system integrators (SIs) like Accenture, Deloitte, and IBM have built extensive SAP practices. For example, Deloitte’s SAP practice generates more than $30 billion in annual revenue.

Testing, often overlooked in terms of its complexity, plays a pivotal role in the implementation process. It is not just a procedural step but frequently the most significant bottleneck in ensuring successful deployments, consuming as much as 40–50% of the implementation budget. Despite its critical importance, testing remains largely manual, labour-intensive, and ripe for disruption through automation and innovation brought by generative AI.

‍

Implementing packaged applications

The implementation of packaged applications involves a multifaceted process. It begins with gathering business requirements and creating a blueprint to map out workflows and identify gaps between the current state and desired outcomes. From here on, there is a development process to customise and configure the application to meet these specific requirements. Throughout this journey, rigorous testing is essential. Unit tests ensure individual modules function correctly. Integration testing verifies seamless interoperability with other systems. Regression testing ensures that everything that has worked in the past continues to. Finally, data migration and parallel runs confirm that historical data transfers seamlessly, and the new system operates as expected alongside the old one before a full switchover.

Testing often spans weeks or even months for each phase of implementation. The stakes are particularly high in "brownfield" implementations, where new functionalities are layered onto legacy systems. These projects demand exhaustive testing to prevent new updates from breaking existing processes. Firms like Deloitte, Accenture etc. have dedicated testing teams, often using automation tools in their manual testing process. Yet, for all their sophistication, even the largest SIs automate just 20–30% of testing efforts, leaving a vast portion dependent on manual workflows.

Testing is a significant component of the overall cost, and can be  upwards of $4 million in a $10 million implementation. This creates a substantial opportunity for innovation, particularly as these applications transition to cloud-based architectures. Shorter release cycles in cloud environments demand more frequent testing, amplifying the inefficiencies of manual processes.

‍

The AI opportunity for testing software

GenAI and advanced vision systems are revolutionising how testing is performed. These technologies can generate test cases from user stories, record manual test steps for automation, and even identify optimal data sets for running tests. The potential to automate significant portions of both test case creation and execution represents a massive shift in how testing is conducted, promising to reduce costs and accelerate delivery timelines dramatically.

‍

Building a successful business in this space requires a product-driven approach. Unlike services firms, which rely on human labour, product companies can scale faster by embedding automation and intelligence into their offerings. Startups should initially focus on a single packaged application, such as SAP or Salesforce, where expertise in proprietary languages (e.g., ABAP for SAP or Apex for Salesforce) can provide a competitive advantage. Unlike Python or Java, there is not enough publicly available data to train models in this space, and therefore, focus will be essential to reach any reasonable level of accuracy. This focused strategy enables companies to refine their product, build a repeatable GTM motion, and establish a strong wedge before expanding into other packaged applications.

‍

Legacy players like Tricentis dominate the market with their SAP-focused testing solutions, generating $330 million in annual revenue. Other players like Worksoft, Mabl and Qualitest also offer automation capabilities, while some SIs have built in-house solutions. However, the market remains fragmented, and new entrants with innovative approaches can carve out a large niche. For instance, GenAI-powered tools that automate testing at scale could provide a differentiated offering, particularly for small and mid-sized SIs that lack the resources to build proprietary solutions.

‍

Challenges and the road ahead

While the opportunity in testing for packaged applications is substantial, it comes with inherent challenges. The primary hurdle lies in the customer persona—implementation partners—and the structure of the market. These partners, particularly large system integrators (SIs), operate in a relationship-driven ecosystem, maintaining strong, long-standing ties with both enterprise clients and technology vendors. Breaking into these entrenched relationships is very difficult. Instead, you’ll need to sell to SIs and it will be their end customers where the underlying value will accrue - a potential incentive misalignment.

‍

Adding to the complexity, many SIs have developed proprietary tools and frameworks for testing and implementation, reducing their dependence on external solutions. For example, large players like Accenture and Deloitte have robust internal testing practices and platforms tailored to their workflows, which creates a significant barrier for third-party solutions. Furthermore, SIs are primarily incentivized to maximise services revenue. Tools that automate processes or reduce manual efforts may be seen as a threat to their billable hours, leading to potential misalignment with their incentives.

‍

Another significant challenge is the concentration of key players in this space. A few large enterprise implementation partners dominate the market, while the rest is a fragmented long tail of thousands of smaller SIs. This creates a risk of customer concentration and makes it harder to scale without securing some of these large players as customers.

‍

Finally, the technical complexity of testing packaged applications poses its own obstacles. Each application, whether SAP, Salesforce, or Workday, has unique workflows, interfaces, and proprietary languages, requiring highly specialised knowledge to build effective testing solutions.

‍

Our belief is that by focusing on a particular packaged application (preferably a large one like SAP / Salesforce) and targeting implementation partners as early customers, a startup could achieve $30–50 million in revenue within a single application ecosystem before diversifying. In fact, we believe that the focus may extend beyond just the choice of an ecosystem, it can be narrower e.g. HR applications or Supply Chain Applications or Marketing Automation within these large suites. Over time, the product could evolve to address more complex testing needs, from integration to regression testing, incorporating AI-driven capabilities to adapt to varied workflows and interfaces.

‍

Founder archetypes

This opportunity needs founders with a deep understanding of packaged applications and a product-first mindset. Strong engineering teams are essential to build robust, adaptable solutions, while go-to-market advantages—such as prior experience in implementation—can accelerate customer acquisition. 

‍

In conclusion, testing for packaged applications represents an underexplored but highly lucrative market. The inefficiencies of manual testing, coupled with the growing complexity of implementations, create a fertile ground for innovation. With the right combination of technology, focus, and team, this space offers the potential to build a category-defining business that reshapes how enterprises approach software implementations. 

‍

If you're building a company that leverages AI to automate testing for packaged applications we’d love to hear from you. Please reach out to us at sayantan@stellarisvp.com or nj@stellarisvp.com.