The innovation tax audit

Welcome to the Executive Briefings. If you clicked over from LinkedIn, it is because you understand an uncomfortable truth about the modern tech economy: we have optimized for building, but we have completely lost the discipline of editing.

Most product teams today operate as feature factories. They measure their success by story points burned, velocity metrics, and the sheer volume of code pushed to production. But velocity is just speed. And speed without financial direction is just a highly efficient way to burn through your operating capital.

As I recently detailed in my piece for Built In, Innovation Requires Deleting Code, the biggest waste in business today isn't poor execution; it is successfully solving the wrong problems. Teams build impressive solutions in record time, yet many end up unused, forgotten, and quietly dragging down the P&L.

Let’s talk about why your most profitable move this quarter might be deleting 20 percent of your codebase.

To understand why building is dangerous, you have to understand the unit economics of software.

Every feature you ship carries an invisible, perpetual tax. It adds to database storage requirements, increases compute load, expands the surface area for security vulnerabilities, and dramatically increases the cognitive load required to onboard new engineers. In financial terms, every line of code you refuse to delete increases your Cost of Goods Sold (COGS) and shifts your budget from Capital Expenditures (CapEx building new value) to Operating Expenses (OpEx keeping the lights on).

I recently applied this philosophy at a healthcare SaaS company that was struggling with severe margin compression. They were actively exploring pricing increases and layoffs because their cloud costs were scaling faster than their revenue.

Instead of cutting headcount, we audited the product. We scanned their codebase, cross-referenced it with their usage logs, and found that nearly 22 percent of their features met the criteria for complete removal. These weren't obscure back-end scripts; they were customer-facing reporting modules, legacy export tools, and bespoke dashboard widgets that had simply fallen out of fashion.

They were funding a museum of old ideas.

When I presented the recommendation to delete 22 percent of the product, the engineering and sales teams were terrified.

Engineering argued that because they had spent millions of dollars and thousands of hours building these features, they had to keep them running "just in case." Sales argued that removing anything would cause churn. This is the Sunk Cost Fallacy working in reverse. They were using past expense to justify future waste.

Building stuff is the easy part. It feels good. It looks like progress. Having the discipline to delete it is what separates project administrators from Product Economists. You cannot negotiate with the Sunk Cost Fallacy in a boardroom; you have to break it with undeniable data.

To prove the organization wrong and secure our financial mandate, we didn't argue. We ran a "Scream Test."

We identified the lowest-usage features contributing to the highest maintenance costs. Instead of formally deprecating them and triggering a massive change-management panic, we simply toggled them off in the staging and shadow environments, and gracefully hid the UI elements in production for a subset of users.

Then, we waited for the phones to ring.

They didn't. Over a 30-day period, out of thousands of active users, exactly zero support tickets were filed regarding the missing tools.

When we brought that data back to the executive team, the conversation shifted instantly. We weren't "deleting features" anymore. We were eliminating an ongoing maintenance tax that was quietly eroding our gross margins by millions of dollars a year. We deleted the code, reduced the cloud infrastructure footprint, and instantly improved the company's Gross Margin profile without acquiring a single new customer.

The smartest teams understand that their time isn't a factory for code; it is a scarce resource for choosing the right fight. They invest 80 percent of their time deeply exploring user pain points, and only 20 percent actually building the solution.

Whether you are evaluating a new feature for your 2026 roadmap, or deciding whether to maintain a legacy system, force your team to answer these three non-negotiable questions:

1. Is this problem actually worth solving? Not every problem deserves a solution. Does this feature tie back to a clear financial or strategic objective? Will it reduce our Customer Acquisition Cost (CAC)? Will it increase our Net Revenue Retention (NRR)? If you cannot map the feature to a financial outcome, you are just building for vanity. Validate the business impact before you allocate the capital.

2. Who is struggling with this daily? Real problems have real people losing sleep. Are we listening to the verified Voice of the Customer, or are we just reacting to the loudest internal voice on the executive team? If you can't name the specific user persona who will pay for this solution, drop it.

3. How will we know we have actually fixed it? Define your success metrics upfront. "User delight" is not a metric; it is a vibe. If you cannot mathematically measure the impact of the feature, you cannot validate the solution. If a feature fails to hit its target metrics after 90 days, it shouldn't be iterated on indefinitely, it should be deleted.

In today's era of AI-generated code and automated deployment, we can build almost anything we imagine, faster than ever before. But as the cost of writing code drops to zero, the cost of maintaining the wrong code compounds.

The future of product leadership is not about generating more output. It is about business architecture. It is about having the executive stamina to look at your sprawling, complex product and systematically delete whatever no longer serves the P&L.

Innovation isn't just about what you add next. It’s about what you have the courage to take away.

--- Richard Ewing | The Product Economist Founder, Exogram.ai | I stop B2B SaaS AI features from destroying gross margins.