5.3 Managing a Product Backlog: Order and Refine

Managing a Backlog

The backlog answers the question the timeline and the board both dodge: of everything we could do, what comes next — and why that? It is the only project view whose entire job is deciding what not to work on yet.

New to this? This page assumes you know what a task is and how work reaches a board. It is the third of the three project views. The three project views  ·  Kanban boards  ·  Glossary

In short

  • A backlog is ordered, not grouped. Exactly one item is next. “These five are all high priority” is not an order.
  • Detail belongs at the top, not everywhere. Items get smaller and clearer as they rise; specifying the bottom of the list is wasted work.
  • Refinement is a habit, not an event. A backlog only stays usable if something drains it and something shapes it, every week.
  • An item nobody will ever pull should be deleted. A backlog that only grows has stopped being a decision-making tool.

What a backlog actually is

A backlog is a single ordered list of everything the team could work on, most important first. That is the whole definition, and the word doing the work is ordered.

Most teams get this wrong in the same way: they sort items into priority buckets — High, Medium, Low — and end up with thirty items marked High. Buckets feel like prioritisation because they involve making judgements, but they answer the wrong question. The team does not need to know that eight things matter. It needs to know which one to start on Monday morning.

Forcing a strict sequence is uncomfortable precisely because it is where the real decision lives. If two items genuinely cannot be separated, the tie-break is cheap — pick either. What is expensive is never making the call, because then it gets made informally, by whoever is loudest or whichever ticket happens to be open.

Detail belongs at the top

The second property of a healthy backlog is that it is not uniformly detailed. Items near the top are small, well understood and estimated. Items near the bottom are large, vague and unestimated — deliberately.

A backlog showing the refinement gradient and the ready line Seven items in priority order. The top four are small, solid-bordered and estimated in points, and sit above a dashed ready line. The bottom three are progressively larger, dashed and unestimated, marked approximately thirteen points, unknown, and epic. Items are refined as they rise so they cross the ready line already small and understood. TOP OF THE BACKLOG 1 Fix timezone on invoices 3 pts 2 Add CSV export to reports 5 pts 3 Weekly email digest 3 pts 4 SSO for the enterprise plan 8 pts Ready line 5 Bulk edit across projects ~13 pts 6 Reporting improvements ? 7 Mobile experience epic Ready to start Small, understood and estimated. The team could pick any of these up today. Not ready yet Bigger, vaguer, unestimated. Refine them before they reach the line — not after. Order is a strict sequence, not a set of buckets: exactly one item is next.
Items are refined as they rise. “Mobile experience” at position seven does not need requirements yet — it may be cut, reshaped or overtaken before it gets near the top, and any detail written today is likely to be thrown away.

This gradient is what makes a backlog cheap to maintain. Writing acceptance criteria for item forty is not diligence; it is speculative work on something that may never be built. Specify late, and specify only what is about to be pulled.

Refinement: the habit that keeps it usable

Refinement is the recurring work of moving items up the gradient: splitting large items, adding the detail that makes them startable, estimating them, and deleting the ones that no longer matter. Most teams do this in a short session once a week or once a sprint.

A useful test for whether an item has crossed the line — sometimes formalised as a Definition of Ready:

  • Anyone on the team could start it tomorrow without a meeting to find out what it means.
  • It is small enough to finish inside one cycle. If it cannot, it is still an epic and needs splitting.
  • You can say how you would know it worked. Not a full spec — a checkable outcome.
  • Its dependencies are known, and any that are external are already in motion.
How much to keep ready. Roughly one to two cycles' worth. Less and the team stalls waiting for decisions; more and you are writing detail for work that will be reprioritised before anyone reaches it. Refinement is the one place where running out is worse than running over.

How to decide the order

Ordering is a judgement, but it should be an argued judgement. Three lenses that make the argument concrete:

LensThe question it asksBest for
Value vs. effortWhat is the most outcome per unit of work?A first sort, quickly — it surfaces the cheap wins and the expensive nice-to-haves
Cost of delayWhat does it cost us for every week this is not done?Separating genuinely urgent work from work that merely feels urgent
MoSCoWMust, Should, Could, or Won't have this time?Fixed-date releases, where the valuable column is “Won't”

Whichever you use, two things matter more than the method. First, the ordering must be visible — if only one person knows why item three sits above item four, the backlog is not a shared artefact. Second, it must be allowed to change. Reordering in response to new information is the system working; a backlog frozen since kickoff has quietly become a plan.

Keeping a backlog healthy

The graveyard

Hundreds of items nobody will ever pull. Every one costs attention at each review. If an item has sat untouched for a year, deleting it loses nothing — if it matters, it will come back.

Only one kind of work

A backlog of nothing but features means maintenance, bugs and technical work are being decided somewhere invisible — usually by whoever is on call that week.

The parallel queue

If urgent work arrives by direct message and skips the list entirely, the backlog stops describing reality and the team's real priorities become undiscussable.

One number is worth watching: the age of the oldest item above the ready line. If something has been ready for months and never pulled, it was never actually that important — and the order is lying to you.


Who lives in this view

The backlog is the home screen for whoever owns what gets built.

RoleWhat the backlog tells them
Product owner / requester (home view)Whether their intent is actually reflected in what happens next — and what they are implicitly saying no to
Project manager (home view)Whether there is enough ready work to keep the team moving, and where scope can be cut if the date is at risk
Team membersWhat is coming, so today's decisions do not paint tomorrow into a corner. They pull from the top; they do not need to read the tail
Sponsor / stakeholderThe trade-offs being made on their behalf. This is the view to bring to a steering meeting — not the board

The failure worth naming: when stakeholders are shown the board instead of the backlog, they see activity and assume alignment. The backlog is where disagreement about priorities actually surfaces, which is exactly why it is the uncomfortable one to share — and the useful one.

How this looks in AB

In AB Project Management the Backlog is one view onto the same tasks that appear on the Kanban board and the Gantt view. Ordering is explicit rather than implied by due date, so a re-order is a real decision recorded against real records. Because subtasks live under their parents, a large item can be split in place as it rises — the epic becomes a parent with children rather than being retyped as new tickets — and its progress rolls up automatically once work starts. Items pulled into the board keep their history, so nothing is lost in the hand-off between deciding and doing.


Terms used on this page

Every term below is defined at more length in the full project management glossary.

Backlog
A single ordered list of everything the team could work on, most important first. A prioritisation device, not a storage cupboard. More →
Refinement
The recurring work of splitting, clarifying, estimating and deleting backlog items so that what reaches the top is ready to start. Sometimes called grooming.
Definition of Ready
The test an item must pass before the team will start it — understood, small enough, checkable, dependencies known. The entry gate, where Definition of Done is the exit gate.
Definition of Done
The team-wide checklist every item must satisfy before it counts as finished. More →
Epic
An item too large to finish in one cycle. Perfectly fine at the bottom of a backlog; a problem the moment it reaches the top unsplit.
Story point
A relative unit of size combining effort, complexity and uncertainty. Deliberately not hours — the point is comparison, not conversion. More →
MoSCoW
Sorting requirements into Must, Should, Could and Won't have. The “Won't” column is the one that does the work. More →
Cost of delay
What it costs, per week, for a piece of work not to exist yet. Turns “this feels urgent” into a number two people can argue about.
Velocity
How much the team completes per cycle. Used with the backlog to forecast how far down the list you will realistically get. More →
Scope creep
Gradual expansion through small additions. A visible, ordered backlog is the cheapest defence: every addition has to displace something. More →

Related reading

Where the items go next

Pulled off the top of the backlog, work enters the flow — and the limits that govern it.

5.2 Kanban Boards →

Where the items come from

Scope definition decides what is even eligible to enter the list.

3.1 Defining Scope →

When priorities change

Reordering is healthy; unrecorded change is not. Where the line sits.

6.2 Issue & Change Management →

Published on: 2026-07-26 Last updated on: 2026-07-26

Questions & answers

Have a question about this topic? Ask below — no sign-up needed. The team reviews and answers questions here.

No questions yet — be the first to ask.

Ask a question

We’ll send a one-time email to confirm your address. Questions appear after a quick review.