Exercise: Prioritize a Customer Backlog

It's week two at Brightline Health, a mid-size clinic network piloting your company's AI document-processing product. You've collected a backlog of requests from five different stakeholders. Everyone says theirs is urgent. You have 10 engineering days before the next steering-committee meeting.

You need a defensible, explainable way to decide what to work on. A simple scoring model is a great FDE tool: it turns a political argument ("my request matters most") into a transparent conversation ("let's agree on the impact score").

The scoring model

Each request is a dictionary:

{
    "id": "BH-3",
    "title": "Auto-extract insurance member IDs",
    "impact": 5,     # 1-5: how much it moves the agreed success metric
    "urgency": 4,    # 1-5: cost of waiting
    "effort": 3,     # engineering days, 1 or more
    "blocked": False # True if we're waiting on the customer (data access, approvals, ...)
}

The priority score is:

score = impact × urgency ÷ effort

rounded to 2 decimal places. A blocked request scores 0, because you can't make progress on it no matter how important it is. (Unblocking it is a separate action item for the customer.)

Your task

Write two functions in the editor.

1. score(request) returns the priority score as a float, following the rules above.

2. plan_sprint(requests, capacity) returns a list of request ids to work on, in priority order:

1. Rank requests by score, highest first. Break ties by lower effort first, then by id alphabetically.
2. Walk down the ranked list. If a request's effort fits in the remaining capacity, include it and subtract its effort. If it doesn't fit, skip it and keep going, since a smaller request further down may still fit.
3. Never include a request with a score of 0.
4. Don't modify the input list.

Example

backlog = [
    {"id": "A", "title": "...", "impact": 4, "urgency": 5, "effort": 4, "blocked": False},  # 5.0
    {"id": "B", "title": "...", "impact": 3, "urgency": 3, "effort": 1, "blocked": False},  # 9.0
    {"id": "C", "title": "...", "impact": 5, "urgency": 5, "effort": 8, "blocked": False},  # 3.12
]
plan_sprint(backlog, capacity=6)  # -> ["B", "A"]  (C needs 8 days; only 1 left)

Press Run to try your code with the sample backlog at the bottom of the file, then Submit to grade it.

FDE tip: In a real engagement, share the scored table with the customer's sponsor before you start the work. Disagreements about scores are cheap to fix on Monday and expensive to fix after you've built the wrong thing.