How large programmes are governed, measured and run.
Notes from our founders on strategy execution: what goes wrong in the room and what good looks like. No definitions, no listicles.
AI will polish a watermelon unless it reads the evidence
A watermelon project is green outside and red inside. An AI summary written from owners' self-assessments makes the skin shine. Point the model at the evidence instead, keep the claim and the proof apart, and leave the rating with an approver who signs it.
Approved against people who are already booked
Much of what is called change fatigue is a capacity problem: portfolios are approved against money while the same teams and experts are booked past what they can carry. Approve against capacity by team and month, sequence instead of stacking, and treat each start as a decision.
Benefits the CFO can book: define value the way finance counts it
A benefit that finance does not recognise never reaches the accounts, however carefully the programme tracks it. Write each benefit as the line it will move, name a finance owner, agree the baseline with finance before approval, and report planned against realised after go-live.
Board packs that help directors govern
A board pack built by hand buries the decisions, disagrees with the dashboards and cannot say who stands behind each number, and AI drafting makes that harder to see. Lead with the decisions owed, show what moved and why, and produce every figure from the approved record.
Dependencies are the programme: manage the joins between plans
Most programme slips start at a join between two plans that nobody owns. Declare dependencies at every level, make both sides agree the date, compute the critical path across projects, and price every slip before you accept it.
Key results and the work meant to move them
Delivery progress says whether the work was done. Key-result attainment says whether anything changed. Show the two side by side and never blend them, link each key initiative to the key result it should move, score against a phased plan, and decide when they diverge.
OKRs or KPIs is the wrong question
KPIs keep the business healthy. OKRs change it. An organisation needs both, in one chain from its vision to the smallest task, each measure doing the job it was chosen for, under level names the organisation already uses.
OKRs that pull the whole organisation one way
OKRs pull an organisation one way only while every team key result can be traced to a company key result, and every piece of work to the key result it should move. Keep the objectives few, measure outcomes, and draw the links before the quarter starts.
Proving the PMO is worth keeping when the cost review comes
A programme office that describes itself by reports produced and maturity scores is first in line when costs are cut. Record what it changed instead: decisions taken differently or sooner, money moved, slips caught early and hours given back, and show the sponsor that page every quarter.
The AI portfolio needs a baseline before go-live
AI pilots are funded as experiments, so nobody measures the work before the pilot starts and nobody owns the benefit. Govern them like any other investment: one register, a baseline taken before go-live, a named benefit owner, gates that ask for evidence, and a decision to stop.
The project nobody will stop: write the exit at the entry
A portfolio that only ever adds work starves its best projects. Write the conditions for stopping on the day a project is approved, judge each review by how much money and how many people move, and make a stop a decision with a route, an approver and a record.
The time an agent gives back, and who owns the call
An agent can collect, assemble and check much of a PMO's week. Set its autonomy task by task: draft, propose, or apply once a person approves. Let it assemble the evidence, keep the rating with a named owner, and spend the hours on decisions and people.
What the PMO is for when AI writes the status report
Once the status report drafts itself, a PMO that only compiled it has nothing left to defend. Four judgements remain, and none can be drafted: which trade-off to put to the committee, what to stop, who answers for a miss, and what counts as proof.
A slipped milestone is a KPI forecast change
A milestone slip usually reaches the KPI it was meant to move months later, as a missed reading. Link each KPI to the work that delivers it, revise the forecast when that work slips, and review delivery and outcomes in the same meeting.
Green, and wrong: why programme status needs evidence
A status that the owner types in will drift towards green, however honest the owner. The fix is a rule: nothing counts until the evidence behind it has been approved by someone else, and the record of that approval cannot be changed.
Governance is a workflow: give every change a route
A governance document describes committees. The decision is made in an email thread and ratified a month later. Governance exists only where each kind of change has a route: who approves it, above what value, by when, who stands in, and what record is kept.
Agents that act, and the person who signs
An agent that only chats changes nothing, and one that changes records unseen cannot be trusted. The pattern that works: the agent reads as the person, proposes one change with before and after, and a named person approves it through the usual route.
Why strategy stalls between the board pack and the work
A strategy reaches the work through three translations, each done once, by hand, in a different tool. Within months nobody can say which pillar a project serves. Keep one model, make every project name what it serves, and review the work that has no link.
Stage gates that cannot be waved through
A gate becomes a ceremony when its criteria are written late, its conditions are tracked by nobody, and the date outranks it. State the evidence before the stage starts, name who decides, track every condition, and treat rewind and closure as decisions.
KPI cascades that survive the first bad quarter
A cascade is drawn once and tested the first time a number disappoints. It holds when every KPI has a written definition, when plan, actual and forecast are kept apart, and when a change to a target is a decision with an approver and a history.
Portfolio reviews that decide something
A portfolio review that gives every project equal time informs the room and decides nothing. Put every project on one scale, take the exceptions first, state the decision wanted before the meeting, and record each decision with an owner and a date.
Delivered, and no benefit: who owns the outcome after go-live
A project ends at go-live and its benefit starts there, after the team has gone. The benefit is measured only if it has an owner outside the project, a baseline taken before the work, and a KPI that someone already reports.
What a PMO should stop doing by hand, and what it should keep
Most of a PMO's week goes on carrying facts from one place to another. Hand collection, reminders, escalation, assembly and first-pass checks to the system. Keep challenge, judgement, coaching and the decision with people, and put a name on every change.
The weekly PMO meeting should not start with a status round
A weekly meeting that opens with a status round spends its hour on what everyone could have read. Collect the updates before the meeting, put only the exceptions on the agenda, end each item with a decision or an owned action, and fix the record at the close.
Business reviews that change something
A business review that walks every KPI in order spends its time on the ones that are fine. Close the numbers before the review, lead with what moved, what was restated and what is missing, and give every off-track KPI a cause, an action, an owner and a date.
Seven tests for strategy execution software
Every demo shows the happy path, so every demo looks the same. Products differ in what they refuse, who they send a change to and what they keep. Seven tests, each run on your own data, show that before you sign.
Running a national programme: what the centre needs to see
A national programme runs through dozens of entities, and each reports in its own format with its own meaning of green. The centre can leave every entity its method, provided all of them share one model, one definition of done backed by evidence, and a record of delegated authority.
After Project Online: keep the scheduler, replace the portfolio layer
Project Online was the layer above the schedules: the central store, the roll-up, the resource pool, the reports and the approvals. Planners can keep the desktop tool they know. The PMO has to replace the portfolio layer, and should first decide what that layer is for.
A platform nobody opens: why rollouts stall after go-live
A programme platform stalls after go-live when it is a second place to type, opens on an empty screen and is read by nobody senior. Start with one portfolio and its real data, build the committee's pack only from it, and withdraw the old template on a named date.
A business case for programme software, without invented savings
A business case built on a vendor's percentage will not survive a finance director. Build it from what the organisation can observe itself: hours spent collecting status, the wait before a decision, the cost of one wrong green. Say what will not be saved, and who confirms each number.
From PMO to enterprise PMO: what changes at organisation scale
A PMO that runs IT projects well is asked to run every department's portfolio. An enterprise PMO needs what a project office can do without: portfolio choices, common standards across methods, governance by delegated authority, and roll-ups the board will trust.
SPM, PPM and programme management: what each one is for
Strategic portfolio management, project portfolio management, programme management and KPI tools each answer a different question. Choose by the question your office is accountable for, and make the layers share one model, or the strategy office ends up reconciling four tools by hand.
What a strategy management office does, and the software it needs
A strategy management office sits between the board and the enterprise PMO. It owns the strategy cascade, the KPI targets, the portfolio's alignment and the review rhythm. It fails when it turns into a reporting team, and its software should make that harder.
What a transformation partner should leave behind
On the day a partner leaves, the client should be able to run the next reporting cycle alone. That takes four things left behind: the plan in a system the client owns, the rules written into it, a team that has already run the rhythm, and a record of decisions.
The best Planview alternatives in 2026: 10 options compared
10 options for teams moving from Planview, each described from its own public pages with what it does well and when to choose it.
The best Project Online alternatives in 2026: 10 options compared
10 options for teams moving from Project Online, each described from its own public pages with what it does well and when to choose it.
Strategic portfolio management (SPM) software in 2026: 10 platforms compared
Ten platforms for strategic portfolio management, each described from its own public pages with what it does well and when to choose it.
No posts match. Clear the search or choose another topic.