Skip to content

Agile Project Management

Applied Market & Business Strategy (the “Strategy & Management Game”) - NIT / TUHH, Hamburg · part of my Technology Management MBA · study notes for revision.


The last chapters were about deciding a strategy and pitching it. This final one is about delivering it - turning a plan into a working product. And here the classic, plan-it-all-upfront way of running projects starts to creak. This chapter is my revision of agile project management: why classic planning struggles when the world keeps moving, what agile optimises for instead, and then Scrum - the most common agile method - walked through end to end.

Quick vocabulary note before we start: agile just means “able to change direction quickly and cheaply”. That’s the whole idea. Everything below is machinery for making change cheap.

1 · Why classic project management struggles

Section titled “1 · Why classic project management struggles”

Classic PM plans the full scope up front, then executes the plan. That works beautifully when the target holds still. The problem is that targets increasingly don’t.

Product life cycles keep shrinking. The teaching example is the VW Golf: across its generations the model life cycle roughly halved - from around eight or nine years down to four or five. If your product must be reinvented twice as often, a project method that assumes a long, stable plan is fighting the market.

And a lot of the up-front effort is wasted. A well-known study of software built the classic way found that a large share of features were barely touched:

How often a built feature was actually usedShare
Neverabout 45%
Rarelyabout 19%
Sometimesabout 16%
Frequentlyabout 13%
Alwaysabout 7%
Roughly 45% of features were never used, and only about 20% were used frequently or always. Building everything the plan asked for meant building a lot of things nobody wanted.

Put those together and you get the core complaint against classic PM: it commits early, and late change is expensive. A customer change request that arrives near the end forces rework of everything downstream - cost and delay balloon. Agile is a direct answer to exactly this pain.

Agile chases two things at once: high stakeholder impact and a low cost of change. The trick is when you spend your influence. In classic PM your ability to shape the product is highest at the start - but that’s exactly when you understand the problem least. Agile keeps the cost of changing your mind flat all the way through, so you can steer using what you learn.

The cleanest way to see the difference is the triple constraint - the three levers of any project: scope (what gets built), time (the schedule), and budget (the cost). You can fix some and must let the others flex.

Fixed (the promise)Flexes (the tuning parameter)
Classic PMScopeTime & budget
Agile PMScopeTime & budget
Classic PM fixes what gets built and lets time and money stretch to deliver it. Agile flips it: fix the time-box and budget, and let scope flex - you always ship something on time, and the least-valuable items are the ones that drop off.

That flip is the heart of agile. A time-box is a fixed slice of calendar time that will not move. Because time and budget are locked, the only thing that can give is scope - and since the backlog is priority-ordered, whatever falls out is the stuff that mattered least anyway.

Agile isn’t one method; it’s a mindset written down in 2001 as the Agile Manifesto. It has four values, each phrased as “we value the left over the right” - not “the right is worthless”, just “when they pull against each other, lean left”.

Individuals & interactions over processes & tools
  • People talking beats rigid procedure
Working software over comprehensive documentation
  • A thing that runs beats a thick spec
Customer collaboration over contract negotiation
  • Work with the customer, don’t fence them off
Responding to change over following a plan
  • Adapt when reality shifts
The four values of the Agile Manifesto. “Software” is the original wording; read it as “the working product” for any field.

Underneath the values sit twelve principles. Rather than dump them verbatim, here’s how I group them into a few themes that actually stick:

ThemeWhat the principles are really saying
Deliver value early & oftenShip something useful fast, then keep shipping in short cycles (weeks, not months). Working product is the only real measure of progress.
Welcome changeTreat late requirement changes as a gift, not a threat - they’re how you win the customer a competitive edge.
Collaborate closelyBusiness people and developers work together daily; face-to-face conversation carries information best.
Trust motivated peopleBuild teams around motivated individuals, give them what they need, and trust them to deliver. The best designs emerge from self-organising teams.
Keep a sustainable paceEveryone - sponsors, developers, users - should be able to hold the same pace indefinitely. No heroics, no burnout.
Keep it simple & excellentMaximise the work not done; pair simplicity with continuous attention to technical quality.
Reflect & adjustAt regular intervals the team stops, asks how to get better, and tunes its own behaviour.

The manifesto is the philosophy; in practice a handful of concrete features carry it:

FeatureWhat it meansWhy it matters
Autonomous cross-functional teamOne team holds all the skills to build the thing, and decides for itself howBalanced decisions, fast reaction to change
Continuous improvementThe team keeps learning and adjusting as it goesThe team gets better during the project, not after
Time-boxesWork happens in fixed, unmovable slices of timeForces tough, timely prioritisation
Return on investment (ROI)Value-for-money is the number-one thing to prioritise byBuild the high-value items first
Product qualityQuality is baked in, not inspected in at the endEnables shipping in increments safely
Pull principleThe team pulls the next item when it has capacity, rather than being pushed workOptimal use of the team’s real capacity

Scrum is the best-known agile method. It’s not a big process - it’s a short, repeating loop plus a few roles and rules. The whole thing runs on a fixed rhythm called a sprint (a time-box, usually one to four weeks). Here’s the cycle:

Product Backlogeverything we might build, prioritised
→
Sprint Planningpick & break down top items
→
Sprint Backlogthis sprint’s task list
→
The Sprintbuild it · Daily Scrum each day
→
Incrementa shippable product slice
Sprint Reviewdemo · honest customer feedback
→
Retrospectivehow do we improve as a team?
↻
…back to the Backlogrefine, re-prioritise, next sprint
One turn of the Scrum loop. Plan a small batch, build it in a fixed time-box, show it, reflect - then loop straight back and do it again. The backlog is a living list that keeps getting re-prioritised as you learn.

The rest of the chapter just unpacks each piece: the roles who run it, the artefacts they work on, and the events where they meet.

Scrum has exactly three roles, and the boundaries between them matter.

Product Ownerthe voice of the customer
Scrum Masterthe guardian of the process
Development Teamthe people who build it
Three roles, one team. Notice nobody here is a traditional “boss” telling the team what to do day to day.
RoleWhat they ownThe mindset
Product OwnerThe Product Backlog - its content, its wording and above all its priority order. One per team.Customer advocate. Constantly asks “what’s the most valuable thing to build next?” and defends the customer’s interest.
Scrum MasterThe process. Removes “road blocks” for the team and coaches them toward self-organisation. One per team.Servant leader, not a manager. Serves the team, protects the rules, has no authority to hand out tasks.
Development TeamThe increment - they build the product, fully self-organised, guided only by the backlog specification.Self-managing. Decides how to do the work and does not take directives from outside the project.

Artefacts are the physical (or digital) things the team works on. Scrum has three: the Product Backlog, the Sprint Backlog, and the Increment.

6.1 The Product Backlog - and the DEEP rule

Section titled “6.1 The Product Backlog - and the DEEP rule”

The Product Backlog is the central specification artefact: one prioritised list of everything the product might need. The Product Owner manages it. Items can be polished user stories or just rough sketches and ideas - deliberately uneven in detail. A good backlog follows the DEEP rule:

  1. Detailed appropriately - top items are specified in detail; far-off items stay rough. Don’t over-specify what you might never build.
  2. Emergent - the backlog is alive. Items appear, change and get deleted as the project teaches you things.
  3. Estimated - every item carries a rough work-load estimate, more accurate near the top.
  4. Prioritised - items sit in strict priority order, most valuable first.

The payoff of “detailed appropriately” is lean, just-in-time specification: you spend your specifying effort only on what you’re about to build, not on a giant document written before you understand the problem.

Backlog items are usually written as user stories - a short description of a requirement from the user’s point of view, small enough to fit in one sprint, and valuable in itself. The classic template:

As a [type of user], I want [goal], so that [reason].

There’s a neat variant that puts the reason first:

For [reason], I as [type of user] want [goal].

Why bother reversing it? Two reasons. Leading with the reason keeps the why central instead of letting it get dropped, and it nudges you to name a real user type rather than accidentally writing the story from the Product Owner’s perspective. Small change, better stories.

However you phrase them, good stories follow the CCC guideline:

CMeaning
CardWrite the story on a small physical card. The limited space forces focus - you can’t over-specify.
ConversationThe real detail lives in a face-to-face discussion, not on the card.
ConfirmationAgree the acceptance criteria with the Product Owner up front - written on the back of the card.

Some items are too big to be a single story - those are epics. An epic still carries clear business value and is described briefly, but it breaks down into several user stories during backlog refinement, and doesn’t have to be finished in one sprint. Each story then breaks into tasks during sprint planning.

Epicbig, valuable, multi-sprint
↓ breaks down (refinement)
User story
User story
User story
↓ breaks down (sprint planning)
task
task
task
The zoom levels: an epic (“organise a conference dinner”) splits into stories (“book a restaurant”, “book a return bus”), each of which splits into concrete tasks.

Story mapping is a technique for laying epics and stories out in chronological order - say, building a house: preparation → foundations → structure → interior. It gives everyone the big-picture overview, doubles as a sanity check that the logic and sequence make sense, and feeds the backlog board.

6.4 The backlog board and “ready for sprint”

Section titled “6.4 The backlog board and “ready for sprint””

The backlog board organises the Product Backlog visually. Items flow toward a Ready for sprint column - but only once they pass a readiness check. Typical readiness criteria:

  • Product Owner and team share a common understanding of the item and of why it matters to the customer.
  • Acceptance criteria are agreed between them.
  • The item is the right size - no single item should exceed about a third of the team’s sprint capacity.

Items become “ready” during backlog refinement (more on that below).

Once a sprint starts, the selected items move into the Sprint Backlog - the realisation plan of tasks for this one sprint. It usually lives on a physical task board, the central tool for the team to manage itself. Tasks flow left to right through columns:

Backlogall tasks for the sprint
→
To-Dopicked for this iteration
→
In Processsomeone is on it now
→
Donecompleted
The task board - a dead-simple visual of project status. Team members pull a task into “In Process” and move it to “Done” when finished; a colour code shows who owns what.

The increment is the deliverable of a sprint: an individually marketable slice of product - for software, a release. Every sprint produces a new one, and each should raise customer value. Ideally an increment improves both the observable side (features the user sees) and the hidden side (technical quality under the hood) at once.

7 · Estimation - story points and planning poker

Section titled “7 · Estimation - story points and planning poker”

Before a sprint, the team has to size the work. Agile does this with relative estimation, not hours.

A story point is an abstract, relative measure of work-load - not hours, not days. You pick a small reference item, give it some arbitrary number of points, and then size everything else relative to that. The scale is non-linear, roughly Fibonacci-style - 1, 2, 3, 5, 8, 13, 20, 40… - and that’s deliberate:

The scale1 · 2 · 3 · 5 · 8 · 13 · 20 · 40 · …

Why the gaps grow the bigger the item, the fuzzier the estimate - so the buckets get wider

Nobody can honestly tell “85 points” from “86” - so the scale doesn’t even offer that choice. It forces honest, coarse estimates for big items and fine ones only for small items.

The clever part: because points are relative and abstract, they’re independent of team speed. As a team gets faster over a project, an item’s point estimate stays stable - only how many points they clear per sprint (their velocity) goes up. And the shared reference item gives the team an objective anchor to argue from, so agreeing on numbers gets easier.

Planning poker is how the team turns individual guesses into a shared estimate - in two rounds:

  1. Round one. Each member privately picks a card for a task’s work-load, then everyone reveals at once. The highest and lowest estimators explain their reasoning - the outliers are where the useful disagreement hides.
  2. Round two. After the discussion, everyone estimates again. If the numbers are now close, take the average and move on.

The point isn’t the number - it’s the conversation the disagreement triggers. A wide spread means people are picturing different work, and that’s exactly what you want to surface before you commit. (Special cards exist too - for “let’s split this, it’s too big”, “merge it, too small”, or just “can we have a break?”.)

Scrum’s meetings are all time-boxed and each has one job. Here’s the set, in order around the loop.

EventWhenWhoThe one job
Backlog refinementOngoingProduct Owner, team, Scrum MasterKeep the backlog fresh - add/remove items, re-prioritise, get top items “ready”
Sprint planningStart of sprintTeam, Scrum Master, (PO)Agree the sprint target; pick and break down items into tasks
Daily ScrumEvery dayTeam (SM, PO optional)Re-coordinate toward the sprint target
Sprint reviewEnd of sprintEveryone + customerDemo the increment; gather honest feedback
RetrospectiveEnd of sprintTeam, Scrum Master, POImprove how the team works

Refinement keeps the backlog current with “fresh” customer input. The team adds new items and drops obsolete ones, folds in feedback from the last review, updates estimates, and specifies the top items to sprint-readiness. Advice from the slides: it can take up to half a day, but shouldn’t eat more than about 10% of the team’s capacity - and stick to just-in-time specification, don’t over-polish.

A short daily sync - about 15 minutes, same time and place, standing up (standing keeps it short). Each member answers three questions:

Yesterdaywhat did I do toward the sprint target?
→
Todaywhat will I do toward it?
→
Blockersis anything in the team’s way?
The three questions of the Daily Scrum. Best held in front of the task board so everyone can see the flow. Note it’s the team coordinating itself - not a status report to a manager.

8.3 Sprint review - feedback on the product

Section titled “8.3 Sprint review - feedback on the product”

At the sprint’s end the team demonstrates the increment to the Product Owner, customer, users and other stakeholders. They inspect what was built, and the feedback is categorised (must / should / could / won’t) back into the backlog. Crucial advice from the slides: don’t optimise for nice feedback. Go looking for the honest, possibly uncomfortable insight - that’s the feedback that actually improves the product.

The retrospective is the other feedback loop: not “is the product good?” but “are we working well together?”. The team gathers data, digs for root causes (the “5 whys”), and commits to a few concrete, SMART improvements for next sprint. It’s the direct agile equivalent of “lessons learned” in classic PM - except it happens every sprint, not once at the very end.

8.5 The heartbeat: the Deming / PDCA cycle

Section titled “8.5 The heartbeat: the Deming / PDCA cycle”

Zoom out and every Scrum loop is really the Deming cycle, also called PDCA - Plan, Do, Check, Act:

Plandecide what to do
→
Dodo it
→
Checkdid it work?
→
Actdecide what to change
↻
Plan again…
Plan → Do → Check → Act, forever. Sprint planning is “Plan”, the sprint is “Do”, the review is “Check”, the retrospective is “Act”. The whole method is just this cycle running on a fast loop.

How do you see whether a sprint is on track? The burndown chart plots remaining work (in story points) against sprint time (in days). It starts at the total planned work and should “burn down” to zero as tasks complete.

Startall story points remaining
→
Ideal linestraight line down to zero
vs
Actual linethe real remaining work
→
Endzero - all tasks done
Compare actual against ideal at a glance: if the actual line sits below the ideal, you’re ahead of schedule; if it sits above, you’re behind. One picture tells the whole team where the sprint stands.

Pure agile is rare in big regulated industries. More often agile lives inside a plan-based frame - and the slides gave three real patterns worth knowing:

PatternThe setupWhy it’s arranged this way
Agile inside a V-modelA medical device is developed under a certified, waterfall-style V-model (specify fully, then verify level by level), but the software sub-project runs agile inside it.The regulator demands a design-freeze and 100% specification per level - no incremental releases allowed - yet each prototype still gives real customer feedback.
Agile synced with a waterfall roll-outAn HR tool is built agile, timed to land alongside the first wave of a classic waterfall regional roll-out (EMEA, then others).The build can flex and learn; the roll-out is repetitive and predictable, so it stays classic. Each runs in the mode that suits it.
Agile then classic, consecutivelyAgile development of the concept first, then a classic verification-and-approval phase governed by the regulatory regime.Explore freely while the design is fluid; switch to disciplined, documented process once it must be certified.

The lesson: agile and classic aren’t enemies. You match the method to the part of the work - flex where you’re learning, plan where you’re repeating or must comply.

Finally, agile has grown beyond projects into a way of running whole companies. The driver is VUCA markets - Volatile, Uncertain, Complex, Ambiguous. (Think of the automotive industry being reshaped by new mobility players.) If your market can flip in months, an org that decides its roadmap once a year can’t keep up.

One lens on this is Laloux’s evolution of organisational forms - a sweep from tiny tribes, to command-and-control chiefdoms, to rule-bound hierarchies (church/military), to the innovation-and-meritocracy of modern corporations, to value-driven pluralistic firms - and, as the next stage, self-managing “teal” organisations built on self-management, wholeness and an evolving sense of purpose. It’s the same agile idea - trust self-organising teams, respond to change - scaled up from a Scrum team to an entire company.

That completes the course. Back to the course overview - from deciding a strategy, to pitching it, to delivering it.