AlphaScrum, the method.
A delivery operating system for organisations that need agility to hold up under audit, procurement, regulation and distributed teams.
Scrum told you how a team should work. It never told you how twenty teams, three vendors, two regulators and one board should work together. That is the gap this method closes.
Why another method
Because the existing ones answer a smaller question.
Scrum is a team-level framework, and an excellent one. It describes how a small group can inspect and adapt its way through uncertain work. It is deliberately silent on almost everything that determines whether a large organisation can actually ship: how teams are bounded, how funding works, how a change reaches production, how audit evidence is produced, how three vendors and two regulators fit into a two-week cadence.
The scaling frameworks answer some of that, at a cost: a great deal of structure, a vocabulary to learn, and a tendency to reproduce the coordination overhead they were meant to remove. Some of them are genuinely the right answer in the right conditions, and we will tell you when we think one is.
AlphaScrum is not a replacement for any of them. It is the diagnostic and sequencing layer that decides which of them you need, in what order, and what has to change structurally before any of it will hold.
Principles
Six positions the method rests on
These are the load-bearing claims. If you disagree with all six, we are probably not the right firm for you, and that is a useful thing to establish in the first meeting.
Outcomes are contracts, velocity is not
Story points are a local planning aid. They are not a commitment, a productivity measure, or anything a board should ever see. Commit to outcomes and to flow, and let estimation stay the cheap internal tool it was designed to be.
Optimise the queue, not the worker
In knowledge work, the overwhelming majority of elapsed time is wait time. Utilisation targets above roughly eighty percent lengthen every queue in the system. Busy teams and fast teams are different things, and the difference is usually work in progress.
Governance is a design problem, not a tax
Audit trails, segregation of duties and change approval are real constraints in regulated industries. Treat them as inputs to pipeline design and they become a by-product of delivery. Treat them as an enemy and they become a quarterly crisis.
Structure beats exhortation
You cannot coach your way out of a team topology that forces six handoffs to ship one change. Change the boundaries, the interfaces and the ownership first; the behaviour follows. Culture is a lagging indicator of structure.
Distributed is a first-class mode, not a degraded one
Colocation is one solution to communication bandwidth, not the only one. Written decision records, overlap windows designed deliberately, and asynchronous-first defaults outperform a colocated team with poor documentation discipline.
The engagement must be designed to end
Every engagement carries a handover date and a named internal successor from day one. Capability that cannot outlive the consultant was never capability — it was rental.
The ALPHA loop
Five phases, run as a loop rather than a plan
Each phase has entry criteria, artifacts and an exit test. You can enter at any point — most organisations that already have delivery teams enter at Locate — but you cannot skip a phase and keep the result.
Align
Agree what "done" means before anyone writes a line of code.
Most agile failure is not a delivery failure — it is an alignment failure discovered late. Align establishes the outcome contract: the measurable business result, the constraints that cannot be negotiated, and the decision rights that say who breaks a tie.
Practices
- Outcome contract — one page, signed by the accountable executive
- Constraint register — regulatory, contractual, architectural, political
- Decision-rights map (RAPID-style) applied to the delivery scope
- Definition of Value, upstream of the Definition of Done
Artifacts produced
The antipattern this phase prevents
Locate
Find the real constraint before you scale the team.
Adding people to a system you have not measured is how a three-team programme becomes a nine-team programme with the same throughput. Locate instruments the value stream, finds the binding constraint, and refuses to staff around it.
Practices
- Value-stream mapping with wait-time versus touch-time split
- Flow baseline: throughput, cycle time, WIP, flow efficiency, failure demand
- Dependency and handoff census across teams and vendors
- Capability heat map against the constraint, not against the org chart
Artifacts produced
The antipattern this phase prevents
Pressurise
Make the system deliver under real conditions, at small size.
A pilot that is protected from the organisation proves nothing. Pressurise runs one genuine slice of work end to end — through the same procurement, security review, change board and release path the whole portfolio will use — and fixes what breaks.
Practices
- Thin vertical slice through every gate, not a sandbox pilot
- Deployment-pipeline hardening and release-path reduction
- Governance rehearsal: audit evidence produced as a by-product of delivery
- Explicit failure budget and a written rollback path
Artifacts produced
The antipattern this phase prevents
Harden
Turn the working slice into a repeatable operating standard.
Harden converts what worked into defaults: team topologies, interface contracts, a shared Definition of Done, platform services and the engineering practices without which agility degrades into fast chaos.
Practices
- Team topology design — stream-aligned, platform, enabling, complicated-subsystem
- Interface contracts and team APIs between every pair that must interact
- Trunk-based development, test strategy and continuous delivery baseline
- Internal platform and paved-road services for the highest-friction steps
Artifacts produced
The antipattern this phase prevents
Amplify
Scale the standard, and keep the feedback loop shorter than the change.
Amplify rolls the standard outward, wave by wave, with an internal coaching cadre rather than an external dependency. The measure of success is our removability: an engagement that cannot end is an engagement that failed.
Practices
- Wave planning with per-wave entry and exit criteria
- Internal coach academy and certification against the method
- Portfolio-level flow metrics and quarterly business review cadence
- Explicit exit plan with a handover date agreed at the start
Artifacts produced
The antipattern this phase prevents
Delivery at AI-native speed
Speed is a consequence of economics, not of effort
We build and ship rapidly across AI, blockchain, mobile and platform work. That is not a work-harder claim. It is what happens when one term in the batch-size equation collapses and you respond to it correctly.
The honest caveat
None of this repeals the rest of the library. Little’s Law still holds, dependencies between teams are untouched, and a team that generates ten times the code and still waits three weeks for an environment has gained nothing at all.
- 01Batch size is set by transaction costOrganisations ship in large batches because shipping is expensive — the review, the test cycle, the approvals, the coordination. Reduce the cost of a single change moving through the system and small batches stop being a discipline you impose and start being the cheapest option available.
- 02AI collapses the cost of producing a changeGenerating an implementation, a migration, a test suite or a piece of documentation has become dramatically cheaper. That is a real and large shift in one specific term of the equation — and only that term.
- 03So the constraint moves to verificationProduction is no longer the bottleneck. Review capacity, test confidence and the ability to tell whether a change is correct become the binding constraint. Teams that generate more and verify the same amount have built a bigger queue, which is the oldest mistake in delivery wearing a new costume.
- 04Which is why we invest thereTrusted automated tests, evaluation harnesses for probabilistic systems, small independently releasable changes, fast pipelines, and review aimed at intent and invariants rather than syntax. Verification is the thing we engineer hardest, because it is what actually governs the pace.
- 05Then improvement compoundsWhen each cycle is cheap and each cycle produces trustworthy signal, learning accumulates instead of being averaged away by uncertainty. That is the exponential part — and it is conditional. It only holds while verification keeps pace with generation.
Maturity model
Five levels, and the honest signal for each
Maturity models are usually marketing instruments designed so that every client is level two and every solution is the vendor. This one is written so you can place yourself without us.
The events happen. The outcomes do not.
Stand-ups, sprints and retros are on the calendar. Releases are still quarterly, still manual, still tense. Agile vocabulary has been adopted without any change to how work actually flows.
Individual teams are genuinely good. The system around them is not.
One or two teams have real engineering discipline and predictable delivery. Their throughput is capped by dependencies, shared environments and approval queues they do not control.
The organisation measures the right things and acts on them.
Flow metrics exist at portfolio level and influence funding. Dependencies are actively managed down rather than coordinated around. Deployment is automated for most of the estate.
Boundaries, platforms and funding are designed for change.
Team topologies are deliberate and revisited. A platform team runs a paved road. Funding follows products and durable teams rather than projects. Compliance evidence is generated automatically.
The operating model changes itself faster than the market changes.
Improvement is a standing capability with its own budget and cadence. Experiments at the operating-model level are run, measured and reversed without drama. External help is episodic and narrow.
In practice
Where each phase shows up in an engagement
| Phase | Diagnostic | Transformation | Pods | Academy |
|---|---|---|---|---|
| Align | Core | Core | Quarterly | Module 1 |
| Locate | Core | Core | Onboarding | Module 2 |
| Pressurise | — | Core | First slice | Module 3 |
| Harden | — | Core | Ongoing | Module 4 |
| Amplify | — | Core | — | Lead track |
The method is published, not licensed.
Every practice above is described in the knowledge library in enough detail to run yourself. We charge for doing it with you, not for permission to know it.