Leverage Maps #11: Stop Managing Your Backlog, Start Designing It

Your backlog isn't just a pile of work. It's the output of your organizational system. And like any system, it responds to incentives, constraints, and feedback loops that shape what gets built and what gets ignored.

The systems diagram below shows how infrastructure backlogs actually form. Not from random ticket creation, but from systematic flows driven by how people are allocated, what gets prioritized, and how emergent work consumes capacity. Some of this comes from what I learned at Goodfellas Pizzeria in Cincinnati, Ohio. The systems framework builds on Will Larson's work on engineering systems.

The system diagram reveals multiple sources feeding your backlog:

Feature requests from users and stakeholders Bugs discovered by engineers or customers
Maintenance work driven by version drift and platform stability
Strategic investments in architecture and platform improvements
Exploratory work when facing uncertainty or technical risk
Tooling improvements to boost developer efficiency Emergencies that disrupt planned execution

Each flows in at different rates, shaped by company culture, financial constraints, personnel decisions, and strategic priorities.

The shift from management to design:

Instead of constantly triaging what flows in, you can intentionally shape these streams. Adjust allocation rates. Create dedicated capacity for maintenance. Design feedback loops that prevent emergency spikes. Build space for strategic exploration.

You can't optimize delivery without understanding what's flowing into your system. You can't reduce technical debt without designing capacity for maintenance work. You can't scale sustainably if emergencies constantly hijack your plans.

This helped me see backlog management less as endless firefighting and more as intentional system architecture. The most successful engineering teams don't just react to their backlogs. They consciously design the flows that create them.

What would happen if you designed your backlog flows instead of just managing the output?