Backlog Refinement: How to Kill Bad Ideas Early With ROI Math

For most teams, Backlog Refinement has become a “ceremony theater” where items on the Jira board are quickly skimmed over. However, in true agility, Refinement is not just a meeting, but a process of weeding out and refining. This process begins not in the meeting room, but the moment an IDEA is born.

Part 2 of the Agile Series — The Complete Cheat Sheet.

1. Backlog Funnel: Not a Recipe, But an Art of Filtering

When an idea falls into the mouth of the funnel, we face two paths: Either a structure that does everything by the book but drowns in clunkiness (over-engineered approaches like SAFe), or pragmatic, result-oriented agility.

First Filter: “Back-of-the-envelope” Calculation

This is one of the most valuable lessons I learned from a visionary leader: You don’t need detailed reports to understand if a business idea makes sense. If an idea doesn’t say “it has potential” with a quick math calculation (1-2 minutes) that fits on the back of an envelope, never waste time on that job.

Every “junk” idea that isn’t eliminated at this stage turns into a massive cost that steals the team’s time during the analysis and development phases later.

2. ROI Math: Comparing Apples into Apples

If the idea passes the “back-of-the-envelope” test, we need to compare it with other tasks. However, the return of every task doesn’t speak the same language. Here we have three main categories: Legal (Necessity), Revenue (Turnover), and Cost Down (Efficiency).

To speak a common language with the Business unit, it is essential to equate these values on the EBITDA (profit) axis:

• Legal & Cost Down: These tasks prevent money from leaving our pockets, meaning they are written directly to profit.
• Revenue: This task brings turnover into the safe. However, not all turnover is profit.

3. Estimating Effort: T-Shirt Legend

Structures like SAFe can sometimes turn even estimation into a bureaucracy. Our approach is a simple but disciplined T-Shirt Legend model. As the work grows, risk and uncertainty increase, so we increase the duration exponentially, not linearly:

Size	Corresponding Duration (Approx)	Point (For Analysis)
XS	1 Sprint	1
S	2 Sprints	2
M	4 Sprints	4
L	8 Sprints	8
XL	16 Sprints	16
2XL	32 Sprints	32

4. ROI and Time to Market

We now have a normalized Effect and a scored Effort. Our equation is simple: ROI = Effect / Effort.

Here we have a critical strategy: If the ROI of two tasks is close to each other; we always choose the one with the smaller effort. We call this Time to Market. The sooner you go to market, the sooner you start earning that money and the faster you learn the real big risks (like customer reaction).

5. “Doing Scrum” vs “Being Agile”

Many teams get stuck on the question “Are we doing this according to SAFe rules?”. If the process slows you down, it is no longer Agile.

The Over-Engineering Trap: If you are doing everything by the book but not producing value for the customer, you are just playing “process theater”.

Our Approach: Simplicity. Instead of asking our questions while doing the work (during the sprint), you should exhaust them during this simple math and analysis phase. Otherwise, you will drown in a constant back & forth.