Work packages, milestones and budgets: the project plan funders expect

Ask experienced assessors where weak bids fall down and the answer is rarely the idea — it's deliverability. The project plan is where a promising proposal proves it's an executable one, and it's marked accordingly. Yet most applicants treat the plan as an afterthought: a vague timeline pasted in after the prose is finished. Strong bids do the opposite — they build the plan early and let it discipline everything else.

Think in project months, not dates

Grant projects start when the award and contracting land, not when you'd like. That's why bid plans are written in relative time: M1 is the first project month, a 24-month project runs M1–M24, and milestones fall at "end of M8", not "August". Writing your plan this way from the start avoids a painful rework and matches how funders read.

Work packages: the unit of accountability

A work package (WP) is a coherent block of work with one named lead, a start and end month, defined outputs, and its own budget. For a typical SME R&D project of 12–36 months, four to seven WPs is the normal range. A structure that recurs across funded projects:

• WP1 — Core technical development (often split into two WPs when there are distinct workstreams)
• WP2 — Integration, trials or validation — where the technology meets reality
• WP3 — Evaluation and evidence — measuring what the trial proved
• WP-final — Project management, dissemination and exploitation — running the whole window

Three properties assessors check quickly:

• Dependencies are explicit and honoured. If WP3's trial needs WP2's integration, WP3 shouldn't start before WP2 ends — a plan that violates its own dependencies reads as unexamined.
• Every WP has a named, plausible lead. The same person leading everything is a key-person risk; name deputies for critical packages.
• Effort is stated. Person-months per WP let an assessor sanity-check budget against activity in seconds.

Milestones: few, checkable, consequential

A milestone is a binary, dateable event — "trial system commissioned at site", "gate review passed" — not an activity. Four to eight across a project is plenty. The test for a good one: could a monitoring officer verify it happened in a ten-minute call? Milestones that are really just WP end-dates restated add nothing; milestones tied to go/no-go decisions show a team that manages risk rather than narrating it.

Budgets: by category, reconciled, defensible

UK funders cost projects in standard categories — labour, overheads, materials, capital usage, subcontracting, and travel. Three rules keep a budget credible:

• Cost per work package, then roll up. A single project-level number invites doubt; per-WP costs by category show the money follows the work.
• Reconcile everywhere. The plan's total, the finance table, and every number quoted in prose must agree to the pound. Mismatches are the fastest credibility leak in a bid.
• Respect grant intensity and match funding. The grant covers a percentage of eligible costs that varies with company size and project type; your plan must show you can fund the remainder — and subcontracting a large share of the work to others undermines the case that you should be funded.

People make the plan believable

Named people with stated roles and allocated hours turn a Gantt chart into a commitment. Allocating your team across WPs also surfaces the problems you want to find before assessors do: the engineer allocated 250% of their time in M6, the WP with no senior oversight, the critical skill covered by exactly one person.

Let the plan discipline the prose

A plan built early makes every other section easier to write and harder to doubt: the approach section can point at WPs instead of hand-waving, the risk section can attach mitigations to specific packages and months, and the budget justification writes itself. When prose and plan are made together, the bid reads like one project rather than a story with a spreadsheet stapled on.

GrantSpark's planning workspace is built around exactly this shape: work packages, tasks and milestones on an interactive Gantt in project months, costs by funder category, person-month effort and per-person hour allocation — drafted by AI from your application, refined by hand, and fed back into the AI's drafting and review of your prose.