Score impact/effort for backlog items
Hundreds of items in the backlog and priority gets re-argued every sprint. The agent scores value against effort on one scale and explains every score.
Medium · 20 min · once every two weeks
What you get
Backlog, 143 items — by quadrant Do now (high value, low effort): 12 PM-318 Autosave drafts why: 3rd most common support ticket, ~2 days of work Big bets (high value, high effort): 9 Small wins (low value, low effort): 47 Skip: 61 Held back: 14 items with nothing in the description to score.
A sample on made-up data — your numbers will be your own.
Who it fits
- Sprint planning where priority is argued from scratch every time and the loudest voice wins.
- You need the reason on record: a score with no rationale does not survive the first argument.
When it won't work
Tracker items are one-line titles with no description — there is nothing there to score.
How the agent does it
Name board and metrics
Say which board to take and what value is measured by. Without that the agent scores general importance and it turns into a matter of taste.
Run the scoring
The agent walks the items on one scale, writes a rationale under every score and holds back the ones with too little description.
Check the rationales
Ask where value was scored with no metric behind it, and where effort is the agent's own guess rather than an estimate from the team.
What you set
Yandex Tracker or Kaiten. By hand: the board or queue, the metrics value is measured by, and the scoring scale.
What you'll need
KaitenStarter prompt
Copy the prompt or open it straight in a chat with the agent.
Read the backlog tasks in [Yandex Tracker / Kaiten], queue [name], and score each one by impact and effort. Count impact by [revenue / churn / support load] and by the number of users affected; effort — by the team's estimate, if it is set on the task. Apply the [1–5] scale identically to all of them and attach a key explaining what each score means. Do not assign a score where the task has no description: collect those in a "nothing to score" list stating which information is missing. Do not pass your own guess about task size off as the team's estimate — mark where the effort came from. Output a table: task, impact, effort, the rationale for each score, quadrant — and the ten tasks worth taking first.
Open in chatCheck the scores. List: — tasks where impact was set without leaning on a metric — on what reasoning; — tasks where effort is your guess rather than the team's estimate; — tasks that duplicate each other or are contained one within another; — tasks that cannot be done separately because of dependencies but ended up in different quadrants; — tasks where you scored the title because there is no description; — which scores you can refine by reading the linked tasks and which require a conversation with the team.
A second prompt — the agent uses it to review its own work and show what's left for you.