Why a long pack goes unread
A steering group pack that runs to dozens of pages, arrives two days before the meeting and covers every work package serves the people who wrote it. The chair reads the first page. Members read the section they own. Most of the pack is never referred to in the meeting. This is a common pattern rather than a measured statistic.
Three pressures produce the long pack. Teams worried about omissions include everything. Teams worried about escalation soften the language and move difficult findings into technical annexes. Teams short of time reuse last month's template and change little. Each produces length, for a different reason.
Shortening the pack carries a political risk. Sponsors expect to see what the controls function has done, auditors expect an evidence trail and package leads want their area covered. The answer is to separate the two jobs. The pack that supports a decision is short. The evidence sits in the main pack and the annexes, and the summary points to it.
This guide sets out a structure for the short pack. It is a design, not a template, and it depends on discipline. The aim is focus. Detail has a place, but it follows the headline.
The one-page summary
Every pack opens with a one-page summary. The limit forces the editorial choices that show what matters. A longer summary holds more information, but the steering group needs only the part it can act on.
The page holds five elements. First, a two-sentence status statement giving overall health and the headline position. Second, five KPIs with current value, trend and status. Third, a short variance narrative for each KPI that is off plan. Fourth, the three risks the steering group should know about this month. Fifth, one to three decision requests. Those counts apply throughout this guide.
The page carries the headline variances only. Package-level variances, with their causes, owners and actions, sit in the main pack. If the state of the project does not fit on the page, the editorial choices are not finished.
The table above shows a populated example. Each row carries a value, a direction, a status and a reason. The two decisions it refers to are: (1) approve the £1.5m contingency drawdown for temporary works; (2) agree the funding route for the £4.0m not covered at P80.
Variance narrative
A variance number tells the reader what moved. A variance narrative tells them why and what is being asked of them. 'Cost forecast £148m against plan £142m' is a number. 'Cost forecast £148m (approved budget £142m, +4.2%): £4.5m from implemented compensation events 34 to 41, £1.5m from unbudgeted temporary works; contingency drawdown recommended, see decision 1' is a narrative. The £4.5m and the £1.5m add to the £6.0m variance, and the baseline is stated.
On an NEC4 contract, say which events are implemented compensation events, which are notified and awaiting a quotation, and which are early warnings that have not yet become events. Approved changes normally move the approved budget. State whether the baseline shown is the original budget or the current approved budget, because the variance means different things in each case.
A good narrative says what changed, why, who owns the response and what happens next. Three sentences per variance is usually enough. Here is an illustrative example. 'The mechanical package forecast has increased by £1.8m because the ventilation system was redesigned to meet Approved Document F. The redesign was completed in February and the supply chain impact is now clear. The mechanical package lead is managing the supplier response and will report the revised delivery dates at the next meeting.' The defensive version reads 'the mechanical package is subject to ongoing commercial review pending finalisation of regulatory requirements'. It names no cause and no owner, and gives the reader nothing to follow up.
Passive constructions often signal defensive writing. 'Costs were incurred due to unforeseen circumstances' hides who did what. 'The ground contractor found harder rock than the site investigation indicated, adding £650k to bulk excavation' says what happened. Check drafts for passive phrasing and rewrite it.
Put difficult variances where they will be seen: the summary and the main narrative, not a sub-section of an annex. The narrative must also be accurate. If the cause of a delay is shared between client and contractor, or is unresolved, say so. A report that blames one party for convenience is read with suspicion from then on.
The five KPIs and where each number comes from
Use five headline KPIs. Fewer than five leaves out part of the picture, and more than five stops the reader holding all of it. Cost, schedule, contingency and risk exposure are common to most major programmes. The fifth is project-specific: safety performance, commissioning readiness or consent approvals. Show each with its current value, trend and status. A forecast 2% over plan and falling is a different message from one 2% over and rising. Show the last six months beside the current value so the steering group sees the direction.
Cost: estimate at completion against the approved budget. If earned value is in use, report CPI (EV divided by AC) and the EAC method, and state the budget at completion used. Schedule: forecast completion against contractual completion, with float on the critical path and the next-most-critical path. Earned schedule gives SPI(t) as ES divided by AT, which behaves better than SPI(cost) late in a project. Contingency: the current unallocated provision, what has been drawn and what is requested.
Risk exposure needs a definition. A P80 is a confidence level on a distribution of cost or completion date. By itself it does not measure exposure. Report the P50 and P80 from the latest QCRA and QSRA against the approved budget and the contractual completion date, with the data date of the analysis. Then give the headroom: the difference between the P80 and the forecast, set against what is held to cover it.
Be exact about what is held. Contingency is the provision approved for risk within the project, whether shown inside the budget line or alongside it; say which. Risk allowance is the time or cost the contractor includes in a programme or price for identified risk, for example the time risk allowance shown on an NEC4 first programme under clause 31.2. Management reserve is held outside the performance measurement baseline and is not allocated to work packages; say who controls it. The steering group should know which of the three each figure refers to.
The HM Treasury Green Book requires optimism bias and risk adjustment in appraisal. It does not mandate P80, so the confidence level used for a KPI should be the one the programme's governance has agreed and recorded.
Status ratings: RAG and the Delivery Confidence Assessment
A rating is useful if it keeps its meaning. Green means on plan and expected to stay so. Amber means a material issue needs attention. Red means a critical issue needs intervention now. Two failures destroy this. In one, amber becomes the default because green feels complacent and red feels like an escalation. A pack in which nothing is ever green reports team anxiety, and a pack in which the same items stay amber for months shows that the rating has lost meaning. In the other, a team rates red items amber because 'we do not want to alarm the steering group', or green because 'we are working on it'.
The cure is sponsor behaviour. If a red rating brings blame, ratings drift to amber whatever the facts. If it brings an investigation, they stay honest.
Projects reporting to NISTA (formerly the IPA) use a five-point Delivery Confidence Assessment: Red, Amber/Red, Amber, Amber/Green and Green. Use that scale where it is required, and keep your internal ratings consistent with it. The objection to decorative colour is aimed at heat maps and colour-coded cells used as styling, not at a mandated assessment scale.
Three risks and one to three decisions
The risk section covers three risks: the ones the steering group needs to know about this month, not the register. If the risk manager believes all 47 entries are equally important, the prioritisation has not been done. The full register goes in an annex.
For each risk, give four things. State the risk in specific terms: 'objection from the planning authority to the revised facade design, which could delay submission by 6 to 8 weeks' rather than 'stakeholder risk'. Give the current probability and impact. State what is being done. State what the steering group is asked to note, decide or escalate. Three risks at this level take about half a page.
Decision requests must be specific. 'The steering group is asked to note the risks' is a ritual, not a decision. 'The steering group is asked to approve the drawdown of £1.5m from contingency to fund temporary works' is a decision. A pack with no decision requests treats the steering group as an audience, and the controls function shares responsibility for that.
Limit the pack to one to three decision requests a month. A long list tends to get only some of its items decided, and a pack that carries deferred decisions month after month is showing a governance backlog. Bring forward only the decisions that are ready.
Summarising a QSRA or QCRA result for a steering group
A quantitative risk analysis result reaches a steering group as three things: a headline, a confidence statement and the drivers.
The headline is the answer to the question the group asked. For example: 'The QSRA shows a 25% chance of completing by the contractual date of 31 January 2028. The P50 is 28 February 2028 and the P80 is 10 April 2028, which is 10 weeks after contractual completion.' The confidence statement gives the data date, the baseline programme used, the number of iterations and the main assumptions, so the reader knows how far to trust it. The drivers are the three to five risks or uncertainties that move the result most, taken from the sensitivity analysis. Name them and say what could be done about each.
Then connect the result to the other KPIs. Show the QCRA P80 against the budget and the forecast, and state the unfunded amount. If the analysis is more than a few months old or the programme has been rebaselined since, say that the figures are out of date. Reporting from a stale run is a common way for a risk KPI to mislead.
On a QSRA we start the steering group summary with the question the group needs answered, then show the confidence at the contractual date, the P50 and P80, and the drivers. The detailed outputs stay in the report.
Reporting calendar
A pack can only be issued three to five working days before the meeting if the calendar allows for it. Work back from the meeting date. Set a data cut-off for cost, programme and risk inputs. Allow time for the controls team to analyse and write, for package leads to check their narrative, and for the project manager and sponsor to review. Issue the pack with the buffer intact. A pack that arrives the day before is read less carefully than a good pack that arrives on time.
Agree who supplies each input and when. Cost comes from the commercial team's forecast, schedule from the programme update and its status date, and the contingency position from the risk register and change log. The QRA run date and the date of the latest register review should appear in the pack.
On NEC4 contracts the contractor's early warning register under clause 15 and the status of revised programmes under clauses 31 and 32 are natural inputs. Use them as sources for the schedule and risk KPIs. Do not treat them as a substitute for the project's own forecast.
Anti-patterns
Colour as decoration. Heat maps, traffic lights and colour-coded cells everywhere carry less than clear text. Use RAG for the few indicators where it adds meaning.
Charts with no commentary. A line showing the EAC rising over twelve months, with no explanation, is data without information. Give each chart a title that states the fact and at least one line of commentary.
Updates that are not updates. 'Milestone tracking on plan' and 'commercial activities continuing as planned' have the shape of an update and no content. If nothing has changed, write 'no change since last month'.
The buried headline. The most important point goes in the first line of the summary. If the forecast has moved materially or the completion date has slipped, that is the headline.
The late pack. A late pack gets less attention than an on-time one, however good it is. Fix the calendar so the issue date holds.
Calibrating to the audience
This guide uses 'steering group' for the body that reviews project performance each month, and 'project board' for the body that directs the project. PRINCE2 and APM usage differ, and many organisations treat the two as the same group, so check what your governance documents call each forum and how they relate.
A project board or programme board often needs only the one-page summary, with the main pack as an annex. It wants portfolio context, strategic issues and major decisions. An operational working group needs more detail: package-level variances, risk log entries and change control. The summary still helps, because it keeps the context.
A client review may need contractually sensitive material removed, cost information structured differently and the client's priority areas emphasised. This is appropriate tailoring. The underlying data must be the same in every version. Selection and framing change, and the numbers do not.
Formal stage reviews follow a prescribed structure and are separate artefacts. A NISTA Gateway Review (Gates 0 to 5) assesses a project at a point in time against the review's criteria. In the Ministry of Defence, Initial Gate and Main Gate are investment approval points in the CADMID lifecycle. Main Gate comes at the end of the Assessment phase and is supported by a Full Business Case. Each has its own submission requirements. The monthly cadence should produce the data these reviews draw on, but the review material is not the monthly pack.
For each reporting artefact, ask what it is for, who reads it and what decision it supports. Producing one pack for every audience is the more common failure, and these three questions are how to avoid it.