3 practical steps to reduce tech debt

Tech debt is holding you back (and we brought receipts)

Tech debt is a significant challenge facing many organizations, and the evidence is clear. Organizations often build up so many custom objects, fields, and workflows that managing them becomes almost impossible. As time passes, the complexity increases, and changes become more costly and risky. A shocking statistic reveals that 51% of all custom objects never get used in Salesforce implementations. Additionally, 41% of custom fields on these custom objects are never populated. This data is from our Change Intelligence Research Series.  We analyze 1.4bn metadata items per month so we have some great insights.

Many orgs are overwhelmed by the amount of unused or underused components in their Salesforce instance, leading to costly decisions like abandoning systems altogether. This inefficiency is a major part of the overall tech debt burden. Back in 2017 the analysts Forrester, called it the “Salesforce @scale dilemma”. The more successful Salesforce is, the more it is used, and the more it is configured. This impacts the agility that made Salesforce so compelling in the first place. In 2025, it has only got worse.

“Think of the time wasted on build and test. Obviously no time is wasted on documentation.”

Agents are the future. But not for you, yet.

This is not only a drag on current enhancements, but it is a significant barrier to implementing agents. The power of Agentforce is its ability to reuse existing metadata – Apex, Flow, Prompt Builder. But to do this it needs good documentation of how that metadata works. And agents need access to well-governed data.  Finally, we’ve discovered that agents need to be very accurately and explicitly directed. And that means accurate descriptions of operations processes.

These are the prerequisites to have effective working agents. The current state of technical debt in orgs suggests that there is a lot of remedial work required before you can deploy agents. If you don’t have these 3 aspects – metadata, data, processes – under control, the agents you build will not work consistently or reliably. That is a huge risk to the business. You wouldn’t hire a young, new excited intern and let them start work with no real understanding of how the business operates.

Before agents, orgs “got away” with poor implementations. It is showing up as technical debt and poor ROI. But it is not killing you. But try and implement agents on top of this house of cards and it will highlight the current limitations of your org. 

The great news is that the preparation for agents will make your overall org far more effective and force disciplines on you that will make your org better and better over time.

Business Analysis is not done (rigorously)

One of the core reasons tech debt accumulates is the lack of rigorous business analysis during the implementation process. Teams often rush through requirement gathering, diving into building solutions without fully understanding the business need. As a result, they create unnecessary fields, objects, and workflows that only add to the complexity without delivering value.

Business analysis tools such as process mapping and entity relationship diagrams are underutilized. These techniques can help teams understand what is truly needed, and in some cases, show that no new additions are required at all. This approach, if embraced, could prevent much of the tech debt from being created.

“You start developing. I’ll go and find out what they want.”

Build a business case to reduce tech debt

In many organizations, the focus is on delivering new features rather than reducing tech debt. As a result, teams are under constant pressure to deliver quickly, often without the time to properly assess the impact of adding new features. This “build more, faster” mentality exacerbates tech debt, making it increasingly difficult to manage as time progresses.

Building a business case to reduce tech debt is essential. Leaders need clear evidence and quantifiable data to prioritize tech debt reduction over new feature development. Tools like metadata dictionaries and dependency analysis can help illustrate the cost of ignoring tech debt and make a compelling argument for dedicating resources to address it.

“First, you need to stop the bleeding.”

Take action

To address tech debt effectively, the following actions are recommended:

Understand the scope of tech debt 

Establish a metadata dictionary. This foundational step is critical for understanding and managing tech debt. It is an up-to-date view of all metadata (fields, objects, workflows, permissions etc) and their dependencies. Without this teams are fighting in the dark. And metadata dictionaries that have AI can dramatically reduce impact analysis effort by allowing you to “talk to your org”.

“A spreadsheet is not a metadata dictionary. You wouldn’t manage opportunities in a spreadsheet.”

Evaluate the cost of tech debt

Not all tech debt is equal. You need to understand the impact, and cost of tech debt so you can build a business case that management understands. There is clearly time that is wasted on development. But a higher cost is the time wasted by large user teams (sales, service) on page layouts with 100+ fields. Then add in the impact on data quality, which feeds into dashboards that management are using to make decisions. Using this evidence you can build a case to invest in reducing tech debt. 

“If your opportunity page layout looks like a CSV receipt you have confused users.”

Enforce a structured implementation lifecycle

Ensuring that changes follow a rigorous process that includes proper requirement analysis can prevent new tech debt from forming. You need commitment from management to allow teams to follow a structured implementation lifecycle, giving them the necessary time to complete each step properly.

“If you want to know which process to map first, it is your implementation methodology.”

These actions, paired with clear management buy-in, are key to reducing tech debt and preventing future accumulation.

Final word

Reducing tech debt is not just about fixing the past, but also about preventing future issues. As a foundation, you need a metadata dictionary that accurately reflects the org config and dependencies. You can then build a business case for removing the highest impacting tech debt. Finally, you must establish a rigorous implementation lifecycle, and continually improve it. Without it, tech debt will continue to accumulate, ultimately leading to inefficiencies that can cripple the organization. And it will stop you from benefitting from agents.