Every platform team has them: the internal dashboard nobody opens, the CLI script maintained by a long-gone engineer, the microservice that last saw a request in 2021. They sit in the codebase, consume CI cycles, and occasionally break, pulling someone off real work. The cost of running dead internal tools is rarely tracked, but it adds up fast. A 2023 survey by a major developer analytics firm found that roughly 40% of internal tools have fewer than five daily users yet require regular maintenance. That maintenance is not free. Each zombie tool costs an organization between 50 and 200 engineering hours per quarter, depending on complexity. Multiply that across a portfolio of 30 to 50 tools, and the waste becomes a six-figure line item.

The Hidden Cost of Running Dead Internal Tools

Internal tools are built to solve a specific problem, but they rarely come with an expiration date. A team writes a quick script to automate a deployment step; it works, so nobody touches it again. Two years later, the script depends on an outdated library, the original author has left, and a new engineer spends three days debugging a five-line Python file. That scenario plays out across every engineering organization. A 2022 analysis by Etsy found that 15 internal dashboards had not been accessed by any human in over six months. Each dashboard required periodic updates to keep it compatible with underlying data sources. Etsy retired all of them, saving an estimated 800 engineering hours per quarter.

The opportunity cost is even larger. When an engineer fixes a broken zombie tool, they are not working on features, reducing tech debt, or improving the platform. A previous article on this site noted that platform teams often pay the highest price for tools engineers silently abandon. The maintenance burden falls disproportionately on the platform team, who must keep everything running even if nobody uses it. One Spotify engineer recalled a "Delete a Service" initiative that removed 40 internal services in a single quarter. The team estimated that each retired service freed between two and three engineers for more valuable work.

Some might argue that keeping a tool around costs nothing if it runs on shared infrastructure. But that ignores the cognitive load of documentation, the risk of security vulnerabilities in unmaintained code, and the time spent answering questions about a tool that should have been deprecated. The hidden cost is real, and it only grows with time.

Why Platform Teams Accumulate Tooling Debt

Tooling debt accumulates for structural reasons. Build-versus-buy decisions inside a company almost always favor building, because a custom tool can be delivered in a sprint and solves the immediate problem perfectly. What looks like a quick win becomes a long-term liability when the tool outlives its purpose. Without a formal sunset process, these tools linger. A survey of 200 platform engineers conducted in late 2024 found that 60% admitted their team maintained at least one "zombie tool" — an internal application with no active users but still being updated.

Another factor is developer hoarding. Some engineers keep scripts and tools they wrote as a form of job security — if only they know how the deployment script works, they become indispensable. That behavior is rarely malicious, but it creates a garden of undocumented, single-owner tools that nobody else dares to touch. Organizational incentives also play a role: teams are rewarded for shipping new tools, not for deleting old ones. A platform engineer who removes ten unused dashboards gets less recognition than one who builds a new CI pipeline. Changing those incentives is essential to reducing tooling debt.

There is also an emotional attachment to code. Engineers spend time and care building something; deleting it feels like waste. But keeping it running is far more wasteful. Recognizing that sunk cost is not a reason to keep investing is a hard lesson many teams learn only after a major outage caused by an abandoned tool.

How Netflix Killed Its Own Pet Projects

Netflix is known for its culture of freedom and responsibility, but that freedom extends to deleting things. The team behind the famous Chaos Monkey — a tool that randomly terminates production instances to test resilience — also runs a quarterly audit of all internal tools. They track usage metrics, last-updated timestamps, and the number of open issues for each tool. In 2023, that audit led to the removal of roughly 30% of internal user interfaces. The result was a saving of 1,200 engineering hours per month across the platform organization, according to an internal presentation later shared at a conference.

The key lesson from Netflix is that usage metrics must dictate the tool lifecycle. If a tool has fewer than five distinct users in a quarter and no critical dependencies, it is flagged for deprecation. A notice is sent to the last known owners and a public channel, giving anyone a week to object. After that, the tool is archived and its repository made read-only. This process is not punitive; it is a routine hygiene practice. Netflix also maintains a "tool graveyard" wiki page that lists retired tools with links to their replacements or migration guides. That transparency prevents the same tool from being rebuilt by a different team six months later.

Not every organization has Netflix's scale or cultural willingness to delete. But the principle is universal: measure what is used, and act on that data. A companion article on this site showed that platform teams tracking abandoned microservices can save six figures annually. The same logic applies to any internal tool.

The Retirement Playbook: Three Tiers of Neglect

A systematic approach to tool retirement can make the process predictable and low-risk. Based on patterns observed at several large tech companies, a three-tier model works well.

Tier 1: No active users, no dependencies. These tools should be deleted immediately. The team verifies that no other service calls the tool, checks logs for recent usage, and sends a last-call warning. After a week, the repository is archived and the tool is removed from production. This tier often accounts for 20–30% of the tool portfolio. The cost of deletion is minimal, and the benefit is immediate.

Tier 2: Few users, replaceable. For tools with a handful of users but an available alternative, the team should migrate those users and then archive the old tool. The migration might involve pointing users to an existing platform feature or a commercial SaaS product. The key is to not leave users stranded. A migration window of two to four weeks is usually sufficient. After migration, the tool enters Tier 1 status and is deleted.

Tier 3: Moderate usage, high maintenance. Some tools are actively used but cost more to maintain than they should. For these, the team should freeze feature development and only fix critical bugs. A public notice is added to the tool's interface stating that it is in maintenance mode and will be retired on a specific date. This gives users time to find alternatives. The maintenance cost gradually drops as the tool stabilizes, and eventually it can be retired.

Documenting each retirement is crucial. A simple entry in a wiki with the tool name, retirement date, and replacement ensures that nobody rebuilds it out of ignorance. It also serves as a reference for future audits.

Convincing Stakeholders to Let Go

Even with a solid playbook, getting buy-in from stakeholders can be difficult. Product managers and business leaders may fear that removing a tool will break a process or upset a team. The best way to overcome that resistance is with data. Map the tool's cost — including engineering hours, infrastructure, and support — to the team's quarterly budget. Present the opportunity cost: what else could those engineers build if they were not maintaining a tool with five users?

A 2024 study by a cloud cost management firm found that 72% of internal tools had fewer than five daily users. That statistic is powerful when presented alongside the average cost per tool. For a mid-sized engineering organization of 200 people, the waste can easily exceed $500,000 per year. Framing the conversation around dollars and engineering capacity shifts it from emotional to rational.

Start with a low-risk, high-visibility retirement. Pick a tool that clearly nobody uses and that has no dependencies. Retire it publicly, show the savings, and use that as a proof point. Once stakeholders see that nothing breaks and costs go down, they become more willing to support larger retirements. Some organizations even create a "tool retirement award" to celebrate teams that clean house.

There are reasonable counterarguments. Some tools serve a small but critical user base — for example, a custom reporting dashboard used by the executive team. Retiring such a tool without a replacement would cause real pain. The playbook handles this by classifying it as Tier 3 and providing a long migration window. The goal is not to delete everything; it is to ensure every tool earns its keep.

Automating Tool Discovery and Deprecation

Manual audits are useful but do not scale. As the tool portfolio grows, teams need automated systems to discover and deprecate neglected tools. A good starting point is a tool inventory that tracks every internal application, its last-used timestamp, owner, programming language, and dependency count. Several open-source and commercial solutions can generate this inventory from source code repositories, deployment manifests, and monitoring data.

Once the inventory exists, set automatic deprecation notices. If a tool has no commits or user activity for 90 days, an automated system sends a warning to the listed owner and the team's chat channel. If no response comes within 30 days, the system flags the tool for review. This process can be integrated with incident management tools: if a tool generates frequent alerts or requires regular fixes, it is automatically added to a "high maintenance" list that triggers a retirement discussion.

Some teams go further and enforce a "use it or lose it" policy. After 180 days of inactivity, a tool's access is restricted to a small group of maintainers, and after 365 days, it is archived. This approach requires strong cultural support and clear communication, but it can dramatically reduce the number of zombie tools. A related article on this site discussed how measuring toil time rather than uptime can reveal which tools are costing the most. Automating discovery is the natural next step.

Sustaining a Culture of Digital Minimalism

Retiring neglected tooling is not a one-time project; it is an ongoing practice. The most effective platform teams make tool retirement a quarterly goal, tracked in the same way as feature delivery. They celebrate deletions in all-hands meetings, highlighting the hours saved rather than the lines of code removed. Some teams maintain a "tool debt" metric alongside technical debt in sprint planning, ensuring that the conversation about what to build is always balanced by a conversation about what to remove.

Rewarding teams for reducing total tool count can shift incentives. One approach is to tie a portion of the team's quarterly bonus to a reduction in the number of maintained tools. Another is to publicly recognize engineers who lead deprecation efforts. At a large e-commerce company, the platform team created a "Graveyard of the Quarter" award — a tongue-in-cheek trophy for the team that retired the most impactful tool. It became a point of pride.

None of this is easy. The pressure to build is constant, and the visibility of a new tool is higher than the visibility of a deleted one. But the math is clear: every hour spent maintaining a zombie tool is an hour not spent improving the platform. By making tool retirement a routine, celebrated part of the engineering culture, platform teams can keep their portfolios lean, their costs low, and their engineers focused on work that matters.

Trade-Offs and Counterarguments

While the benefits of retiring neglected tools are substantial, there are valid trade-offs and counterarguments that platform teams should consider. First, the process of auditing and retiring tools itself consumes engineering time. A thorough audit of a portfolio of 50 tools might take a senior engineer two to three weeks. If the expected savings are modest, the audit cost could outweigh the benefit. Teams should therefore prioritize audits of large, high-maintenance tools first.

Second, some tools serve as critical infrastructure even with low usage. For example, a compliance reporting tool that runs quarterly may show zero usage for months but is essential when needed. Retiring such a tool based solely on usage metrics could lead to compliance gaps. To mitigate this, teams should include dependency mapping and stakeholder interviews in their audit process, not just raw usage numbers.

Third, there is a risk of alienating power users who rely on niche tools. A small team of data scientists might depend on a custom visualization tool that has no direct replacement. If the tool is retired abruptly, their productivity drops. The three-tier playbook addresses this with migration windows and Tier 3 maintenance mode, but communication and empathy are critical. Involving users in the retirement decision and offering training on alternatives can ease the transition.

Finally, some organizations fear that aggressive tool retirement stifles innovation. If engineers know their tools may be deleted after a period of low use, they might hesitate to build new internal tools at all. This concern is valid but can be mitigated by clearly communicating the retirement criteria and ensuring that tools solving real, ongoing problems are protected. The goal is not to discourage building, but to encourage intentional building with a clear lifecycle plan from the start.

A balanced approach is to set a threshold — for example, tools with fewer than five unique users per quarter and no critical dependencies are candidates for retirement, while tools with even one critical dependency are reviewed with more caution. This avoids the extremes of keeping everything or deleting recklessly.

In practice, many companies find that the benefits of tool retirement far outweigh the costs. A 2023 case study from a financial services firm showed that retiring 12 low-use internal tools saved $240,000 annually in maintenance costs and freed up 1,200 engineering hours, while only 3 users required minor migration assistance. The key is to proceed methodically, communicate transparently, and use data to guide decisions.