If you’re like me, the memory of playing Snake on Nokia 5110 might raise a nostalgic smile. It was a miracle there was a game on that thing at all. My kids smile too when they see it. They’re baffled: a phone with actual buttons and giant pixels on a tiny screen. They see a relic. What baffles me is if hit 1990s tech seems implausibly outdated, why expect anything more from a software support model developed at the same time?
For decades, the model for software support has followed the same three-tiered approach, designed to triage and resolve IT issues.
Level one (L1) handles basic issues, provides initial support and answers common questions.
Level two (L2) requires deeper technical knowledge and troubleshooting skills; it might entail more in-depth diagnostics and analysis.
Level three (L3) can require highly skilled experts, such as developers, engineers, or system architects to handle the most complex and critical issues.
This model, once a seemingly efficient way to manage the burgeoning world of software, is now facing a critical reckoning. Three-tier software support is no longer fit for purpose in today's rapidly evolving digital age. Thankfully, there’s an alternative emerging, one that uses AI and automation to streamline low-level tasks, freeing up skilled staff to get systems future fit.
It’s worth taking a pause to understand how we got where we are. The origins of this tiered system lie in a specific era of managed services, born some 30 years ago. It was a reaction to the scaling of businesses and the subsequent cost implications of providing widespread software support. The logic was simple: grade engineers based on expertise, price their time accordingly and create a cost-effective funnel for addressing user issues. This resulted in the familiar structure of L1, L2 and L3, a framework still prevalent in many organizations.
While that support model remains largely unchanged, the technological landscape has changed beyond recognition in the intervening decades. In today’s always-on, instantaneous digital age, the velocity, complexity and frequency of change far outstrip the capabilities of this static, reactive model. Sure, we still need bugs fixing, performance tuning and security patches rolled out, but the methods for addressing these needs are lagging behind. And it’s leading to companies pouring money into services that don’t shift the needle for their business.
According to some estimates, companies spent $120 billion in 2024 on software maintenance. By 2033, that number is predicted to top $180 billion. And yet none of that spending is positioning enterprises to be in a better state. It’s money down the drain.
One consequence of the three-tier model is that it’s almost incentivized growth of technical debt. By its very nature, the three-tier system encourages a siloed approach to problem solving. An L1 engineer addresses a ticket, and if it's beyond their scope, it's simply passed on. An L2 or L3 engineer then tackles that specific issue in isolation, often without a holistic understanding of the underlying system. This reactive, ticket-centric approach treats symptoms, not the root cause.
Over time this creates a fatberg of technical debt, an unhealthy mass of unresolved complexities growing beneath the surface.
Over time this creates a fatberg of technical debt, an unhealthy mass of unresolved complexities growing beneath the surface.
The promise of managed services — to ensure efficiency and allow in-house engineers to focus on innovation — is being undermined by a system that inadvertently adds to the technical quagmire.
The traditional three-tier model can also inadvertently create a perverse incentive around change requests (CRs). Today, a significant portion of the managed services industry's profits comes from these requests, which often arise from the very inefficiencies the tiered system perpetuates. By focusing on resolving individual tickets with lower-cost resources, the underlying problems persist, leading to a continuous stream of CRs for workarounds and fixes. It’s robbing Peter to pay Paul.
It can be hard for companies accustomed to operating a traditional three-tier support model to instigate change. Fundamentally, it’s an economically sound response for a reactive operation: and within its bubble, the system believes that it is effective. Meanwhile, outside, the times are changing
Agent-based support
The increasing capabilities of artificial intelligence — particular in spheres such as agentic AI — fundamentally redraw how we should think about support. Imagine what you could do with the money you save from automating just 10% of L1 work. This could create a significant shift in the L1 workers’ responsibilities.
You might not stop at L1. As AI and machine learning become more sophisticated, aspects of L2 work can be streamlined or automated, leading to a blurring of the traditional lines between support tiers. This shift necessitates a move towards more horizontal skill sets within support teams, where individuals possess a broader understanding of the system and can leverage automated tools to address a wider range of issues. The strict hierarchical structure of L1, L2, and L3 becomes an artificial barrier in a world where technology can handle routine tasks and insights from across the system are crucial for effective problem resolution.
And by freeing up your technical staff they can focus on not just the quick fix, but addressing technical debt, enabling your enterprise to be fighting fit for tomorrow.
Whenever I speak to clients about this, they recognize only too well the inefficiencies and potential cost savings that automation offers. If they have concerns, it’s not from resisting automation; rather, they are often asking about effectiveness — the AI accuracy and speed of implementation.
Take the work we did with MYOB, an APAC-based accounting software company. It partnered with Thoughtworks to optimize its complex tech landscape. An application assessment identified key areas for improvement, leading to a phased managed services implementation. This centralized operations, automated tasks and boosted efficiency — resulting in a 10% cost reduction in the first year, with projections of 20-40% savings in the coming years. "Thoughtworks’ DAMO™ Managed Services meant our teams could share the load… creating more time for MYOB developers to concentrate on future-facing features," says Simon Noonan, MYOB's CTO, highlighting the program's value in freeing resources for innovation.
The ticking time bomb of tiered support
While CTOs might champion automation, procurement processes often default to the traditional tiered model, seeking specific numbers of L1, L2, and L3 personnel. The ingrained mental model of how to support software is blocking the change at the gates — procurement. The systematic adoption of automation as a fundamental way of working, is relegated to a tactical add-on rather than a strategic overhaul of service delivery.
The need for a paradigm shift in software support is becoming increasingly evident. The old model, with its rigid tiers and reliance on manual processes, is struggling to keep pace with the complexity and demands of modern software ecosystems. The rise of cloud adoption, the thirst for AI-enabled transformation and the pressing need for robust cybersecurity — all significant drivers of software maintenance spending — demand a more agile, proactive and automated approach.
While the three-tier structure might have offered a semblance of order in a less complex era, it now risks becoming a drag on efficiency and innovation.
The focus must shift from simply responding to tickets to proactively ensuring the health and evolution of software systems, ultimately delivering greater value and lower long-term costs for businesses. The ticking time bomb of outdated tiered support is a signal that the industry must adapt or risk being left behind by today’s rapid currents of technological advances.