Skip to content
Tutorials Center
My learning

Topics

Lesson 22 of 29 ยท Pitfalls and Scaling

Common Mistakes and Scaling Agile

Learn the most common Agile and Scrum mistakes, how Jira gets misused, why velocity fails as a target, and how SAFe, LeSS and Nexus scale Scrum.

Most teams that struggle with Agile do not fail because Scrum or Jira is broken. They fail because they keep the visible parts (the meetings, the board, the vocabulary) and drop the parts that make those things work: honest feedback, a real goal, a team that decides how to do its own work. The result looks like Agile from the outside and feels like bureaucracy on the inside.

This lesson names the most common failure patterns so you can spot them early, shows how Jira gets misused, explains why velocity turns harmful the moment it becomes a target, and gives a short, honest overview of the frameworks that try to scale Scrum to many teams. It builds on Scrum at a Glance, Scrum Events and Velocity, Burndown and Cumulative Flow. The runnable examples are plain JavaScript models: they illustrate the ideas and are not measurements of any real team.

Scrum-but

"Scrum-but" is a phrase for teams that say "we do Scrum, but...": but we skip the retrospective, but we have no Product Owner, but the sprint is eight weeks long, but the manager assigns the tickets. Each "but" removes a piece that Scrum includes on purpose, because each piece exposes a particular kind of problem.

  • The Retrospective exposes how the team works. Without it, the same irritations repeat for years.
  • A Product Owner with real authority exposes what is valuable. Without one, priorities are decided by whoever shouted last.
  • A short Sprint limits how long a wrong assumption survives.
  • Self-managing Developers are the ones closest to the code, so they can best judge how to build it.

Dropping a practice is not forbidden. Scrum is intentionally a minimal framework, and teams add and adapt a lot. The question to ask is: which problem did that practice make visible, and what do we use instead to see it? If nobody can answer, the "but" is hiding a problem rather than solving one.

Zombie Scrum

Zombie Scrum describes a team that has every Scrum event on the calendar and none of the life. The Sprint happens, the Daily Scrum happens, the Review happens, yet nobody learns anything and the product does not get better. Typical signs:

  • The Daily Scrum is a status report to a manager, not planning by the Developers.
  • The Sprint Review is a slide show; no working software is shown and no stakeholder disagrees with anything.
  • Nobody can state the Sprint Goal, because there is none.
  • Retrospectives produce a list that nobody reads again.
  • Releases happen rarely, so "working software" is theoretical.
  • The Product Owner is a relay for requests, not a decision maker.

Signs of a zombie team

Status meetings The Daily Scrum reports to a manager.
Slide-show reviews Nobody sees working software.
No Sprint Goal Work is a list, not a purpose.
Dead retros Actions are never followed up.

The events happen; the learning does not.

1 / 2

The cure is rarely more process. It is usually shorter feedback loops (release to real users sooner), a clearer purpose (a Sprint Goal people actually care about) and a safe space to say that something is not working.

Using velocity as a target

Velocity is the number of story points a team finishes per Sprint. It is a useful forecasting tool for the team itself. The trouble starts when someone turns it into a goal ("we need 40 points per Sprint") or compares teams ("Team A does 60, Team B only 30").

This is a case of Goodhart's law: when a measure becomes a target, it stops being a good measure. Story points are relative guesses by one team. They have no shared unit, so the easiest way to "improve" velocity is to estimate bigger. The number goes up, the product does not, and often quality drops because corners are cut to hit the figure.

The first TryIt is a deliberately simple model. The team's real output never changes, management asks for eight percent more every Sprint, and the team makes sure the number is hit.

JS Goodhart's law in a Sprint table

Press Run (or Ctrl + Enter) to see the console output here.

No output.

Reported velocity climbs from 22 to 38 while the real work stays at 20 units, and the escaped bugs (an invented but plausible side effect in this model) climb. Use velocity to answer "roughly how much can we take next Sprint?", never "who is performing?". Better outcome measures are things users and the business can feel: how often you release, how long a change takes from idea to production, how many defects reach users and whether customers actually use the feature.

How Jira gets misused

Jira is a flexible tool, and flexibility is a trap. Common misuses:

  • A workflow that mirrors an org chart. Fourteen statuses such as "Waiting for Architect Review" turn the board into a handoff diagram. Fewer statuses (for example To Do, In Progress, In Review, Done) make flow visible and problems obvious.
  • Jira as a surveillance tool. Counting tickets closed per person, or time logged per ticket, teaches people to game the tool and to avoid risky, valuable work.
  • Tickets as the specification. A title and nothing else is not a conversation. Stories are reminders to talk; they still need acceptance criteria.
  • A backlog that is a graveyard. Two thousand tickets nobody will ever build hide the ten that matter. Prune it; if something is truly important it will come back.
  • Everything is customized. Dozens of custom fields that no one fills in make every ticket slow to create. Add a field only when someone will use the data to decide something. See Workflows, Permissions and Customization.
  • Using the board as the process. Moving a card to Done does not mean the work meets the Definition of Done.

A short rule of thumb: if a Jira setting does not help the team see work, decide something or learn something, remove it.

Spotting the signs automatically

Team health is not something a script can really measure, but writing the warning signs down as rules is a good way to make them concrete. This stand-in checks a profile of a made-up team against the patterns above and prints advice. Change the values and run it again.

JS Anti-pattern detector

Press Run (or Ctrl + Enter) to see the console output here.

No output.

The thresholds (15 minutes, 20 percent carry-over, 7 statuses) are rules of thumb chosen for this exercise, not official limits, except the 15-minute Daily Scrum time-box, which comes from the Scrum Guide. The point is the habit: turn a vague feeling ("our Scrum is off") into specific, checkable statements, then pick one to fix in the next Retrospective.

Scaling Agile: a short overview

One Scrum team is typically around ten people or fewer. When a product needs dozens or hundreds of people, several teams must work on one product and coordinate. Frameworks for this exist; none is "the" answer, and all of them add structure that you should only adopt when you feel the pain they address.

Three well-known options

SAFe Big, prescriptive, enterprise-wide.
LeSS Few rules, one backlog, one Product Owner.
Nexus Scrum teams plus an integration team.

All three scale Scrum; they differ in how much they prescribe.

1 / 2
  • SAFe (Scaled Agile Framework) is the most widely adopted and the most prescriptive. It organizes many teams into an Agile Release Train that plans together every roughly ten weeks in a Program Increment (PI) planning event, and it adds roles and layers for portfolio, large solution and team levels. Large organizations like it because it gives a map; critics say it can bring back heavy planning and hierarchy.
  • LeSS (Large-Scale Scrum) takes the opposite approach: keep Scrum as it is, with one Product Owner and one Product Backlog for the whole product, and many teams working from it. It adds few new roles and removes layers rather than adding them.
  • Nexus is by Scrum.org. It connects three to nine Scrum teams working on one product with a Nexus Integration Team that helps resolve cross-team dependencies and keeps one integrated Increment.

Whatever you choose, scaling mostly means managing dependencies. The best first move is often to reduce them: organize teams around customer features rather than technical layers, share one Definition of Done and integrate continuously. A scaling framework cannot fix an architecture that forces every change through five teams.

Remote and distributed teams

Agile works with remote teams, but it needs deliberate habits because you lose the hallway:

  • Write more down. Decisions, acceptance criteria and working agreements belong in tickets or a shared page, not in someone's memory.
  • Overlap hours. Teams spread across many time zones should agree on a few core hours for the Daily Scrum and collaboration, and shift Planning and Review into that window.
  • Make the board the shared room. A clean, current Jira board replaces looking over a colleague's shoulder. Update it as you work, not at the end of the week.
  • Prefer short, frequent video sessions for hard conversations, and chat for quick questions. Record Reviews for people who cannot attend.
  • Rotate the inconvenient meeting time so that the same region does not always stay up late.
  • Build trust on purpose. Start Retrospectives with a check-in, and keep cameras-on expectations reasonable rather than mandatory.

Check your understanding

Quick check ยท 5 questions

of 5 answered You scored / 5 ยท perfect!

  1. 1. What does the term Scrum-but describe?

    Scrum-but means 'we do Scrum, but...'. Each omitted practice made a specific problem visible, so dropping it without a replacement hides that problem.
  2. 2. Why does velocity usually fail as a performance target?

    Goodhart's law: a measure that becomes a target stops being a good measure. Teams can raise points by estimating bigger while real output stays flat.
  3. 3. Which situation best matches zombie Scrum?

    Zombie Scrum has the form of Scrum without the feedback and learning. Events happen, but no one inspects a real Increment or adapts.
  4. 4. Which framework keeps one Product Owner and one Product Backlog for many teams and adds few roles?

    LeSS (Large-Scale Scrum) scales by keeping Scrum simple with one backlog and one PO. SAFe is more prescriptive and Nexus adds an integration team.
  5. 5. What is a sound rule for adding a custom field or status to a Jira workflow?

    Every extra field or status adds friction. Keep only those that help the team make a decision or see how work flows.

Summary

  • Scrum-but and zombie Scrum keep the form of Scrum while losing the feedback; ask which problem each dropped practice used to expose.
  • Velocity is for the team's own forecasting. As a target it invites inflated estimates, lower quality and cross-team comparisons that mean nothing.
  • Jira misuse is usually over-complication: too many statuses and fields, tickets as specifications, a backlog nobody prunes and metrics used for surveillance.
  • SAFe, LeSS and Nexus scale Scrum with different amounts of structure; reducing dependencies helps more than any framework.
  • Remote teams need more written decisions, a shared up-to-date board, overlapping hours and deliberate trust building.

Next, put the ideas into practice with Project: Sprint Planner.