Product practice
The difference matters more than it sounds. A wish list grows without limit and nobody's job is to keep it honest. A queue has an order, the order is disputable, and something is supposed to come out the top of it on a schedule. Most of what goes wrong with backlogs is really a queue that quietly turned back into a wish list.
Anatomy
A backlog is only as useful as the items sitting in it. An item that's just a feature name — "add dark mode," "improve onboarding" — can't really be prioritized, because there's nothing to compare it against except a gut feeling about how nice it sounds.
A well-formed item carries four things: the problem it addresses, why it matters now rather than later, a rough size so it can be weighed against other items, and a way to tell afterward whether it actually worked. Missing any one of these doesn't disqualify an idea — it just means the idea isn't ready to be ranked yet.
Four lenses
None of these is the correct method. Each answers a slightly different question, and a team that only ever reaches for one tends to systematically under-rank whatever that method is blind to.
Scores each item on how many people it reaches in a period, how much it affects them, and how confident the team is in those first two numbers — then divides by the effort required. The result is one comparable number per item.
Sorts items into Must, Should, Could, and Won't-have for a fixed scope — a release, a quarter, a compliance deadline. It answers what's in versus out, which makes it well suited to negotiating scope with stakeholders.
Weighted shortest job first divides an estimate of cost of delay — what waiting on this item actually costs — by an estimate of job size. It favors work that unlocks the most value for the least effort, and it's the lens most sensitive to urgency.
Classifies items by the shape of their effect on satisfaction: basic needs whose absence causes anger but whose presence goes unnoticed, performance needs where more is proportionally better, and delight needs that create disproportionate satisfaction precisely because nobody expected them.
Signature tool
One sample item — "password reset failures" — sits in a small backlog of four. Adjust its Reach, Impact, Confidence, and Effort and watch its rank move against the other three, whose scores stay fixed. Nothing else in the queue changed; only the estimate did.
How many users this touches in a typical quarter.
0.25 = minimal, 1 = medium, 3 = massive.
How much evidence backs Reach and Impact, versus guesswork.
Rough size of the work, in person-weeks.
Getting it wrong
A backlog nobody has fully reviewed in a quarter isn't a queue anymore — its bottom half functions as an idea archive, and treating the two the same way weakens both.
If the top ten items are the same ten items they were three planning cycles ago, either the ranking method has stopped being applied or something is silently blocking work from actually leaving the queue.
An item nobody is accountable for revisiting doesn't get re-evaluated — it just ages in place until someone rediscovers it and re-argues its priority from scratch.
Cutting refinement first feels efficient in the moment. It's usually the reason the next planning session runs long, arguing about items that were never made ready.
Questions
Four things: the problem stated plainly, why it matters now, a rough size for comparison, and a way to tell afterward whether it worked. An item missing one of these isn't disqualified — it's just not ready to be ranked yet.
A method that scores each item on Reach, Impact, and Confidence, then divides by Effort, producing one comparable number. The catch is that all the real judgment lives inside the Impact and Confidence estimates, which makes RICE the easiest of the four lenses to game without anyone noticing.
MoSCoW sorts items into must/should/could/won't for a fixed scope and says nothing about order within a bucket. WSJF produces a single ranked number by dividing cost of delay by job size, favoring the most value per unit of effort. MoSCoW answers what's in scope; WSJF answers what order to do it in.
Short enough that the team can meaningfully re-examine the top of it every planning cycle. A common warning sign is a backlog nobody has fully reviewed in a quarter — the bottom of the list starts functioning as an idea archive rather than a working queue.
The recurring practice of reviewing, re-ordering, splitting, and sometimes removing items so the top of the backlog stays ready for planning. Skip it and the backlog accumulates stale items whose original context nobody remembers, turning re-prioritization into a debate about assumptions instead of a comparison of current evidence.