Backlog and prioritisation
The highest-scoring item in their backlog was the one they never built.
Import wizard v2 was a textbook quick win: high value, low effort, top of the ranked list. The effort estimate was sound. The value was a guess nobody had checked.
Prioritisation ranks what you believe. It can't tell you whether you're right.
Steps 01–02
A backlog becomes a plan when the items become comparable
FlowDesk's Q4 backlog held six epics and about two hundred smaller items.
Backlogs turn into arguments because the items aren't comparable. Two people can hold opposite views about the same epic all quarter without ever discovering what they disagree about.
Scoring value and effort relocates that argument. Once ROI is value divided by effort and both numbers are visible, the disagreement becomes specific: you think this is a 40 and I think it's an 80, and now we can talk about why.
The relocation is the whole benefit. Nobody at FlowDesk changed their mind because of a formula. They changed their mind because the formula made them say a number out loud.
Estimating effort is a skill. Estimating value is a claim about the future wearing a number's clothes.
Steps 03–04
The cutline is a policy, and it should be visible
Four quadrants, one line, and an argument that happens once instead of weekly.
Plotted rather than listed, the same six epics stop being a queue and become a shape: three quick wins, one strategic bet, one fill-in, one thing nobody should touch. Two of those three quick wins rest on the same belief.
Then somebody drags the cutline. That's the moment worth designing for, and it works because the software leaves the threshold to a person. A team that has physically placed its line has agreed what "not this quarter" means, and can point at it the next time someone senior asks for a favour.
Custom shift templates sat below the line for three consecutive quarters at FlowDesk. Nobody had to re-litigate it, and the customers who asked for it got an honest answer.
A backlog with no line in it is a list of things you feel bad about.
Step 05 · the honest part
The matrix ranked it first. The matrix was wrong.
Working exactly as designed, on an input nobody had checked.
Import wizard v2 scored value 80 against effort 20. An ROI of 4.0, comfortably above the cutline, first in the ranked list, and cheap. Every prioritisation framework in existence would have told FlowDesk to build it in October.
The effort estimate was good — engineers are usually decent at effort. The value estimate encoded a belief: that setup difficulty was what lost those accounts. Divide a solid number by a rumour and you get a confident rumour.
This is why prioritisation works better as the second-to-last gate before a sprint. Rank the backlog, then look at what first place assumes. When that assumption is important and unproven, the cheapest next move is a two-week test.
A ranked backlog tells you the order of your beliefs. It is silent on whether any of them are true.
Step 06
Epic, features, stories, and the reason still attached
Thirteen points. Two features. Three stories. One objective above all of it.
Only the validated epic got broken down. EPIC-31 became two features and three user stories, estimated and ready, with the objective and the experiment still one click above them.
The stories are written the way stories should be — as a dispatcher I can open a shared schedule without an account — which is only possible because somebody had described the dispatcher in the first place.
Status then lives in whatever tracker the engineers already use. Nobody is asked to move tools, and nobody is asked to explain in a stand-up why this story exists.
One backlog, two views
Prioritise in the matrix. Work in the list. Same data.
Drag an item on the matrix and its score changes in the list. Edit a value in the list and the dot moves. There is no export step between deciding and doing, because there is no second copy.
Fair questions
What people ask about backlog prioritisation
- Isn't value ÷ effort too crude?
Yes, deliberately. A simple formula whose inputs everyone can see beats an elaborate one nobody audits. The precision worth having is in knowing which input is a guess.
- Does this replace our sprint tool?
No. Epics, features and stories sync to the tracker your engineers already use. Prioritisation and intent live here; day-to-day status lives there.
- Who decides where the cutline goes?
You drag it. It records a decision your team made and can be pointed at later. Move it when capacity changes.
- What stops the highest-paid opinion from winning anyway?
Nothing, entirely. But a senior override becomes a visible act — a number changed, by a named person, on a specific date — and can be reconstructed afterwards. In the demo, the epic carrying the CEO's belief was parked with that belief on display, and the record was what settled the question in November.
- Do we need three levels of hierarchy?
Use as many as the work needs. Small teams often stop at features. The level that matters most sits above the hierarchy: the objective the epic reports to.
- How long before this is worth anything?
It depends on how fast you can connect what you already have. For a ten-person team that has typically taken about five days.
What does the top item in your backlog assume?
Score the backlog, draw the line, then check what first place is resting on. It's a two-week question with a quarter-sized answer.
FlowDesk is the Shorter Loop demo workspace. The scoring, the draggable cutline and the objective still attached to the story is live in it — open it and check any number here.