Purpose Alignment Model

Goal

The Purpose Alignment Model classifies by purpose, along two binary dimensions. The Agile Extension gives it the aim of assessing ideas on customer value and business value, and the resulting placement reads as a recommended action for each item.

The problem it addresses is how much to invest. A ranked list says what the team takes next; it does not say whether the item deserves custom development, an off-the-shelf package, an outsourcing contract or removal. Without that distinction, an organisation spends on excellence in payroll, where meeting the standard would do.

The deliverable is the completed grid. The group writes in one line what the organisation has to do better than anyone else, and every later request is tested against that line without reconvening the group. The model shows whether a subject deserves strategic attention; it does not say which strategy or which decision would be right, a limitation the Agile Extension states.

Usage

When to use it

• A backlog where everything is declared a priority: classifying by purpose isolates the one to three activities that deserve custom development.
• A build, buy or outsource decision: the Parity and Partner quadrants point to the standard and to outsourcing.
• A portfolio of initiatives to frame: the same grid sorts programmes as well as features.
• A budget cut to prepare: the Who Cares? quadrant names the activities whose removal costs nothing.
• A team refinement session: the model takes minutes to teach and fits into an existing meeting.

When not to use it

• Setting the order of work: the quadrant gives a type of investment without giving a rank; use ranking.
• Choosing between two costed options: both fall in the same quadrant; use cost-benefit analysis.
• The effect of a feature on customer satisfaction: the grid ignores perceived satisfaction; use Kano analysis.

Description

Two binary dimensions

The first dimension is market differentiation: does this activity set the organisation apart from its competitors? The Agile Extension counts as differentiating whatever is a selling point, whatever a competitor struggles to match or whatever carries strategic value. A differentiating activity calls for custom development.

The second dimension is criticality to operations: would the organisation come to a halt without this activity, or would it breach a legal or contractual obligation? Payroll in a manufacturing company answers yes; its quarterly newsletter answers no.

The two dimensions are independent, which is what produces the four cells. The model was designed for companies competing in a market. The Agile Extension notes that public administrations and non-profit organisations replace the differentiation axis with stakeholder value, alignment with the mission or the production of a public good; the reasoning stays the same.

The four quadrants

Differentiating gathers what sets the organisation apart in its market and what its operation depends on. Custom development is justified here. Nickolaisen limits this quadrant to one to three activities. A Differentiating quadrant with twelve entries signals a failed placement.

Parity gathers what the organisation cannot run without and what gives it no advantage over its competitors: finance, human resources, payroll, regulatory compliance. The Agile Extension holds that adopting standard market practice is usually enough here. The decision that follows is the standard: a configured off-the-shelf package rather than custom development, with the process taken as the package offers it.

Partner gathers what has value for the customer without the organisation's survival depending on it. The Agile Extension draws the conclusion: since these activities come after critical operations, the organisation will never give them the means to excel, whereas a partner whose trade this is will run them better. The decision that follows is a contract.

Who Cares? gathers what brings neither customer value nor continuity. These are candidates for removal; the resources they tie up go back to the portfolio. The quadrant fills mostly with legacy: a report whose recipient has changed job, an option kept for two customers who have left.

Running the session

The model takes a few minutes to teach: a short facilitated session with the people who decide. The Agile Extension makes the result conditional on the right stakeholders being present; the preparation is what costs most.

The session opens with the two questions written on the board. The group then names the one to three things the organisation has to do better than anyone else and writes them in one line, against which later requests are tested without reopening the debate.

The group discusses each item and places it in a quadrant, on the two questions alone. The order matters: the group answers both questions before naming the cell, otherwise the cell they want dictates the answer.

The placement reads as decisions: Who Cares? leads to removal, Partner to outsourcing, Parity to adopting the standard and Differentiating to building. These decisions commit budgets and contracts, so they go up to the committee that owns them; the session produces the recommendation and the case for it.

What makes the technique fail

The Differentiating quadrant that overflows

Every team argues its own work is differentiating, because the label carries the budget and the standing. The placement turns into a distribution of prestige, and the model no longer sorts anything. The limit of one to three entries is announced before the placement. Announced afterwards, it looks like a ruling aimed at the team it excludes. The session keeps repeating that Parity work is indispensable without conferring an advantage: a salary paid late stops the company; payroll wins the company no customer.

Purpose taken for priority

The model says what kind of value an item carries; the order of work is settled elsewhere. An organisation that reads the four quadrants as four ranks will deliver its three differentiating features before the compliance work that keeps it in business. The placement feeds prioritization; the rank comes second, on expected value and on dependencies.

The placement treated as settled

A differentiating advantage erodes: the competitor copies it, the market catches up and the item slides from Differentiating to Parity with nobody having moved it. The organisation goes on funding custom development where an off-the-shelf package would do. The placement is redone at the pace of the planning cycle, and leaving the Differentiating quadrant is as weighty a decision as entering it.

The shortcut of simplicity

The Agile Extension warns that two closed questions can erase a nuance that decides the case for a given feature. The common case is the composite item: a customer portal whose authentication belongs to Parity and whose configurator belongs to Differentiating. The remedy is to split it, down to the grain at which each piece gets a straight answer to both questions.

The model asked to stand in for a strategy

The Agile Extension states that the model assumes sound intent in the business strategy. An organisation whose strategy is wrong will get a placement consistent with it, presented with the authority of a grid. The check therefore sits outside the model: the placement is read back against the organisation's stated objectives, and a Differentiating quadrant that matches none of them calls either the strategy or the placement into question.

AI considerations

Before the session, a language model prepares the list: from a backlog export it groups the items by subject and flags those whose grain is too coarse to take an answer to the two questions, which keeps the session from being spent on splitting. Afterwards it checks consistency: given the line the group wrote and the placement obtained, it lists the items whose position contradicts that of a neighbouring item.

It also helps spot erosion. Revisiting a placement means knowing what competitors offer today, and a model with search access gathers the published offerings and the feature pages. The analyst verifies that material before using it to challenge a Differentiating placement.

The placement itself cannot be delegated. The answer to the differentiation question encodes the organisation's strategy, which the model does not know and which it will replace with whatever usually sets a company in that sector apart. A model told to fill in the grid puts too many items in Differentiating, because feature descriptions are written to sell. The criticality question is settled on internal facts, contracts, regulatory obligations and operational dependencies, which an outside model cannot weigh. A backlog run through the grid is a map of the organisation's strategy and of its weaknesses: it does not go out to a publicly hosted service before confidentiality has been settled.

Examples

A sports equipment company in the canton of Vaud, around 120 employees, sells online and in two shops. Its product team runs five items from its roadmap through the grid.

The group first writes its line on the board: "We invest in what helps a customer choose the equipment that fits their body and their sport."

Moving a single item carries the whole budget difference. Returns handling was in the plan as a custom development at CHF 220'000; placed in Parity, it comes down to configuring the standard ERP module, CHF 40'000, and the CHF 180'000 freed take the configurator's budget from CHF 300'000 to CHF 480'000. No item changed rank: only the type of investment changed.

Visualisations

The grid is drawn as four cells of equal size, differentiation on the horizontal axis and criticality on the vertical axis, with the low values at the bottom left. Each cell carries its name, the reading of the two answers that lead to it and the verb of the decision. The imbalance has to be visible at a glance: a crowded Differentiating quadrant signals a placement that needs redoing; an empty Who Cares? quadrant, a group that dared remove nothing.

The completed grid then reads as an inventory, one label per item placed in its cell. The decisions and their budget consequences fit in a table beside it, one row per item, the document the session produces for the committee.

Cost

Tooling

The whiteboard remains the tool of the session: two axes drawn, one sticky note per item, moved by hand as the discussion goes. The final grid is photographed; transferring it into a tool comes afterwards.

Collaborative whiteboards (Miro, Mural and their equivalents) reproduce that surface for a team spread over several sites. A grid template saves losing the first few minutes to drawing the axes.

Backlog management tools (Jira, Azure DevOps Boards and their equivalents) take in the result of the session. A custom field holding the quadrant on each item keeps the placement attached to the item and makes the backlog filterable by quadrant during refinement. Without that field, the grid stays a photograph in a meeting record.

The spreadsheet holds the register: one row per item, its quadrant, its decision, the amount and the date of the placement. It is the form that survives the session and makes two successive reviews comparable.

The differentiation axis is settled on what competitors do; that material comes from benchmarking and market analysis. A Differentiating placement resting on internal conviction alone remains a hypothesis still to be tested.

Sources

• IIBA, Agile Extension to the BABOK Guide, §7.11 Purpose Alignment Model: the purpose of the technique, the two dimensions, the definition of the four quadrants, the variant used by public administrations and non-profit organisations, the strengths stated, among them the application at any horizon, as well as the three limitations.
• Pollyanna Pixton, Niel Nickolaisen, Todd Little and Kent J. McDonald, Stand Back and Deliver: Accelerating Business Agility, Addison-Wesley, 2009: the original publication of the model, due to Niel Nickolaisen, with the limit of one to three differentiating activities.