You join a project that is already moving. People are meeting, work is being assigned and someone has promised a date. But the meetings do not connect, the plan is more like a list of guidelines, and nobody is quite sure who can make which decision.

That is a difficult place to manage from. You are expected to help delivery move while the basic rules are still being worked out.

The answer is usually not to stop everything and write a giant project manual. Start by making the current position visible, agree the next few decisions and put just enough structure around the work to keep people aligned. Then improve the arrangements as you learn more.

This example is illustrative. It combines common project conditions rather than describing a client or a specific project.

First, find out what is already happening

Before you introduce a new plan or meeting, work out what people believe they are doing. Ask the sponsor, delivery leads and people doing the work. Their answers may not match. That difference is useful information; it is not a reason to blame the team.

Make a short list under four headings:

• In progress: work that has started, and who is doing it.
• Promised: dates, outcomes or commitments already given to someone.
• Decided: choices that have an owner and a record you can find.
• Assumed: things people are treating as agreed, but nobody has confirmed.

For each item, add the source and the person who can confirm it. If you cannot confirm a date, call it a forecast or an assumption. A tidy plan with invented certainty is not much help.

For example, a team may be configuring a new process before the approval route is agreed. That does not automatically mean the work must stop. It does mean you should find out what the work depends on, what could be costly to undo, and who is authorised to accept that exposure.

Agree the next useful outcome

When the final plan is not ready, agree a near-term outcome that is useful and reversible. Make it small enough to understand, but meaningful enough to move the project forward.

For a process change, that might be a short walkthrough of the current process and a tested draft of the proposed one. It is more useful than “continue discovery” because people can tell what they are expected to produce and what happens next.

Write down:

• What should be produced or learned?
• Who is responsible for the work and who accepts the result?
• What does “ready” mean?
• What information or decision is needed first?
• When will the team check progress and decide the next step?

Keep a near-term delivery view separate from a longer-range outline. Mark the outline as provisional if it depends on discovery, estimates or decisions that are still open. A plan can be useful before every detail is known, as long as readers can see what is firm and what is not.

Connect meetings to decisions and work

Disjointed meetings are often a symptom of unclear purpose. One group discusses progress, another asks for decisions, and the delivery team leaves with actions that never make it into the plan.

For each recurring meeting, write one sentence explaining why it exists. Then ask:

• Who needs to attend to contribute or decide?
• What should be prepared before the meeting?
• What decisions can this group make?
• Where will decisions, actions and unresolved questions be recorded?
• Who needs the outcome afterwards?

If two meetings have the same purpose and people, consider combining them or making their boundary clearer. If a meeting is only for sharing information, a short written update may do the job. Do not cancel a required assurance or approval meeting just because it seems repetitive; confirm the requirement with the owner first.

At the end of each working meeting, read back the decisions and actions. Give each action one named owner and a due date agreed with that person. Record a decision with its date, decision-maker and any conditions. “We discussed it” is not the same as “we agreed it.”

Set the minimum working rules

You do not need to settle every process question at once. You do need enough agreement that people can coordinate their work without guessing.

Use this one-page working agreement and fill in what is known. Mark the rest as open, with an owner and a date to resolve it.

Topic	Working agreement
Near-term outcome	What are we trying to deliver or learn next?
Work and ownership	Where are actions kept, and who updates them?
Plan and changes	Where is the current forecast? How are impacts reviewed?
Decisions	Who can decide what? How do we record the answer?
Issues and risks	Where do we raise them, who reviews them, and when do we escalate?
Meetings and updates	Which meetings remain, what are they for, and who receives a summary?
Review point	When will we check whether these arrangements are working?

This is a starting agreement, not a substitute for your organisation’s required controls. Ask the sponsor, PMO or relevant control owner which policies, approvals, records and assurance steps apply.

Clarify authority before the next big commitment

A project manager can usually organise the work and recommend a way forward. That does not mean the PM can approve every change to scope, funding, policy, risk appetite or business outcomes. Check the authority you have actually been given.

Put the decision into a short format:

• Decision needed: what must be approved or chosen?
• Options: what are the realistic choices?
• Impact: what changes to outcomes, time, cost, quality, resources or risk?
• Recommendation: what do you suggest, and why?
• Decision owner and deadline: who can decide, and by when?
• If no decision is made: what work can continue safely, and what is affected?

If authority is unclear, ask the sponsor to name the decision-maker and confirm the interim boundary. Avoid presenting silence as approval. If the choice involves safety, legal or regulatory duties, security, material financial exposure or a difficult-to-reverse commitment, follow the relevant control route before proceeding.

When a proposed change affects the agreed outcome or constraints, record it and assess the impact before treating it as part of the plan. The Association for Project Management describes change control as capturing and tracking a proposed change through to a decision; the approval route depends on the authority set for that project. See its guide to controlling project change decisions.

A practical reset meeting

Invite the sponsor and the people who own the work, decisions or dependencies. Keep the meeting focused. You could use this agenda for 45 minutes:

1. 5 minutes: confirm the outcome the project is meant to achieve.
2. 10 minutes: compare what is in progress, promised, decided and assumed.
3. 10 minutes: identify the next useful outcome and any work that depends on an open decision.
4. 10 minutes: agree decision owners, escalation route and minimum ways of working.
5. 5 minutes: assign actions and dates.
6. 5 minutes: read back what was agreed, what remains unknown and when the group will review it.

If the group cannot agree on an important point, record the different views and the person who will resolve it. Do not force consensus just to finish the agenda.

Example: a process change already in motion

Suppose several teams have started changing a customer hand-off process. One meeting has agreed new steps, but the support team has not confirmed its capacity. The reporting team is working to an old date, and nobody knows whether the policy owner has approved the proposed wording.

A useful first reset might produce this:

• Confirmed: three teams are mapping their current hand-offs.
• Not confirmed: support capacity and the policy wording.
• Next outcome: a tested draft process for one team, with exceptions listed.
• Owners: operations lead prepares the draft; support lead confirms capacity; policy owner decides wording.
• Boundary: do not roll the new wording out until the policy owner approves it.
• Review: sponsor checks the evidence and decides whether to extend the pilot.

Work can continue on mapping and a limited test while the decision is pending. A wider rollout waits for the approval that controls it. That is a more useful distinction than either “stop everything” or “keep going and hope.”

Your first-week checklist

1. Talk to the sponsor and delivery leads; compare their understanding of the outcome.
2. List work in progress, existing promises, recorded decisions and assumptions.
3. Find commitments that may be hard to reverse or require formal approval.
4. Agree one useful near-term outcome, its owner and what “ready” means.
5. Map recurring meetings to their purpose, attendees and decision authority.
6. Start a short decision and action record that people can actually find.
7. Agree the minimum working rules and assign owners to unresolved points.
8. Set a review date to adjust the plan and ways of working as information improves.

Keep the structure proportionate

A messy start does not mean the project needs a large governance process. It means you need to understand the work, make authority clear and agree how people will coordinate. Add controls in proportion to the project’s risk, scale and organisational requirements.

The UK Government’s Project Delivery Functional Standard, version 2.1 (2025), is one formal example. It expects a project governance and management framework, including authority, decision roles and accountabilities. It applies to UK government bodies; other organisations should follow their own policies and tailor their arrangements to the work.

Start with the next decision and the next piece of work. Make assumptions visible, give actions clear owners, and check whether the meeting structure is helping anyone move forward. Then adjust as the project becomes clearer.