Start the Project

A new project starts with drawing, because drawing is the thing everyone knows how to do. Six weeks later it turns out nobody established who signs off, the survey has not been commissioned, and the budget was never tested against the brief.

The decisions that cost the most are all available in the first week and almost none of them require a drawing.

Workflow

Step 1. Establish what is actually agreed, and with whom

Ask, in one message:

1. What is the project, in one sentence, and what has the client actually asked for?
2. Is there a signed agreement? If not, what is being relied on, and is work proceeding without one?
3. Who decides? Not who the contact is. Who can approve a design, approve a cost, and sign. Is it one person, a committee, a board, or someone not in the room yet?
4. What has been said or promised so far, including verbally, and by whom?
5. What is the client measuring this project on: a date, a budget, an area, a permission, an occupancy, a sale?
6. What do they have already: a site, a survey, a brief, a previous scheme, a consultant they intend to use?

Question 3 is the one that predicts the project. A client who cannot say who decides will produce a project that reverses decisions, and knowing that in week one changes how the work is structured. Point at stakeholder-map.

Question 5 is the second. A project measured on a date and a project measured on a budget are run differently, and clients rarely say which unless asked.

Step 2. Sort the unknowns into blocking and not blocking

Most of what is unknown at the start does not matter yet. A few things stop everything, and they are usually the ones nobody commissioned:

Unknown	Blocks what	Who answers	How long	Commissioned?

Work through the ones that are usually blocking and usually late:

• The survey. Nothing can be drawn accurately without it. Point at survey-review and existing-conditions-survey.
• The geotechnical investigation. Blocks the foundation and often the levels. Point at geotech-report-review.
• Hazardous materials, on any existing building, before the scope is fixed.
• The zoning position, and whether the scheme needs discretionary approval. Point at zoning-summary and entitlement-tracker.
• The code path, where the building type or the existing building makes it non-obvious. Point at senior-code-consultant.
• The budget, and whether it has ever been tested against the brief. Point at fee-estimator and budget-reconciliation.
• Utility capacity on infill and rural sites both.

Order by lead time, not by importance. A four-week survey commissioned in week one and a two-day zoning check commissioned in week six is the right order even though the zoning matters more.

Step 3. List the decisions that have to be made before drawing

Distinct from unknowns: these are choices, not facts.

Decision	Who makes it	By when	What it blocks

Typically: the delivery method, whether the client holds the consultants or the architect does, what standard of documentation the project gets, whether there is a phasing constraint, what the client's approval process is and how long each round takes, and who on the practice's side is doing what.

Include the practice's own decisions: who runs it, who draws it, what it is being priced at, and what it displaces. Point at utilization-tracker and go-no-go-review if the pursuit record exists.

Step 4. Say plainly what can wait

The list nobody writes, and it is what protects the first month:

Materials, finishes, the elevation, the structural system in detail, the mechanical strategy in detail, anything about the interior beyond area, furniture, landscape, signage.

Say why each can wait: it depends on something above, or it is reversible, or it is cheap to change later. A project that starts with a materials conversation has spent its first month on the most reversible decision available.

Step 5. Set up the record, once

The things that are cheap now and expensive to reconstruct later:

• Where project files live and how they are named. Point at file-organizer.
• Which trackers this project will actually keep, and who updates each. project-tracker, permit-tracker, rfi-tracker, drawing-log — pick the ones that will be maintained and skip the rest. A tracker nobody updates is worse than none.
• How decisions get recorded, and by whom. Point at meeting-notes-cleaner.
• What gets written down about verbal agreements. Point at /draft-email.

Step 6. Report

• Who decides, and the answer to what the project is measured on.
• The blocking unknowns, ordered by lead time, with what is commissioned and what is not.
• The decisions due before drawing, with owners and dates.
• What can wait, and why.
• What is being relied on that is not in writing.
• Anything proceeding without a signed agreement, said plainly.

Then ask where to write it. This is read back at every phase change and it is the document that says whether the project started well.

Rules

• Ask who decides, and refuse to accept the name of the contact as the answer.
• Ask what the project is measured on.
• Order the unknowns by lead time, not by importance.
• Say what can wait, explicitly, and why.
• Name every tracker the project will actually keep and skip the rest.
• Say plainly when work is proceeding without a signed agreement.
• Point everything at a skill or a party. Do none of it here.

Anti-patterns

• Starting with a sketch.
• Treating the client contact as the decision-maker.
• Commissioning the survey after the scheme is drawn.
• Fixing the scope before the hazardous materials survey.
• Never testing the budget against the brief.
• Setting up six trackers nobody will update.
• Spending the first month on materials.
• Working without an agreement and not saying so out loud.

Resources

None. This skill is one file. Output is written directly at the path you choose.