Which bids should receive scarce proposal capacity?

Decision boundary

Freeze a portfolio snapshot before comparing demand

A portfolio allocation needs a cut-off. Record every live pursuit and material pre-bid activity known at that time, including proposals not yet authorized to proceed. For each item, retain the buyer, procedure, lot, stage, controlling document version, next gate, official deadline, internal safe date and accountable owner. The snapshot should also show work that is absent because it falls outside the chosen business unit, geography or horizon. Without those boundaries, an opportunity can enter after the decision and consume capacity without displacing anything on paper.

Choose a horizon that matches the evidence. A twelve-month revenue plan is too coarse for daily specialist assignments, while a one-week board conceals the next wave of bids. Many teams need two linked views: a rolling six- to twelve-week allocation for named work and a longer directional view for hiring or contracted support. The near view grants capacity. The long view signals likely shortages but does not book a person against an unqualified opportunity.

Use one source cut and one decision authority. Opportunity owners submit their current records before the cut-off. Changes after it enter a visible log with the source, time and affected constraint. This protects the comparison from private optimism and creates a reasoned path to reopen the decision. It also lets an agent distinguish an obsolete allocation from the one currently in force.

Capacity supply

Protect delivery before declaring any bid capacity

Start from gross working time only to reconcile the calendar. Deduct signed customer delivery, incident and service coverage, regulatory work, management duties, approved leave, training and any protected improvement work. Then apply the organization’s approved reserve for disruption and forecast error. The remainder is not automatically fungible. A proposal writer, security architect and delegated pricing approver represent different capacity pools even when each has an unbooked day.

GovS 002 version 2.1 describes resource management as balancing supply and demand for the right skills, equipment and facilities when needed. It also requires business, operational and service continuity when critical resources are lost. Applied here, a portfolio plan that succeeds only by stripping cover from contracted work is infeasible. The delivery owner must confirm what is protected and the portfolio authority must record any authorized exception rather than burying it in a percentage.

Classify supply as `confirmed`, `conditional`, `provisional`, `unverified` or `unavailable`. A contractor option is conditional until commercial authority, start date, access and relevant competence are proved. A colleague expected to finish another project is provisional until its owner releases the time. Include access, language, geography, conflict and delegated authority. Preserve only the personal detail required for the decision; an agent normally needs a pseudonymous resource ID and verified capabilities, not performance notes or health information.

Bid demand

Fund the next decision gate rather than the imagined full bid

Each pursuit requests a small set of dated work packages. A work package names its output, required skill or authority, earliest start, need date, effort range, dependencies and completion evidence. The portfolio imports the estimate rather than asking opportunity owners for a rounded staffing wish. Demand for a new solution design remains separate from response writing. One hour of authorized contract review cannot be exchanged for one hour of formatting simply because both are labeled bid effort.

Plan only to the next reversible gate unless later work must be reserved now. An early opportunity might need qualification research and one solution-fit decision, not a submission team. A live tender might need the minimum compliant response through solution freeze, after which the portfolio can reassess. This staged treatment keeps uncertain future work from crowding out approved delivery and current bids. It also prevents a proceed decision from becoming an open-ended claim on the team.

Import rather than recreate the evidence that changes allocation. Eligibility comes from its qualification record. Response effort comes from a work estimate. Delivery fit, contract exposure, contribution, award probability and remaining bid investment retain their owners, dates and ranges. Missing evidence stays missing. A capacity board should not raise a probability or suppress a risk because the preferred portfolio otherwise fails.

Feasibility

Find the constraint that changes the feasible portfolio

An overload report is only the start. Test whether moving work within its allowable window removes the shortage. Then test permitted delegation, preparation by another role, reuse of approved evidence and reduction to the minimum safe response. Preserve dependencies and approval rights while doing so. If the solution architect remains over capacity in week three after all safe moves, that pool is binding. Extra writing capacity will not change the feasible set.

Some work is divisible and some is not. Research can be split by notice. A coherent solution baseline, pricing decision, executive review or final release often needs one owner and a protected block. Ten spare half-hours do not necessarily replace one five-hour reasoning block. Include setup, handoff and context loss where they materially consume the scarce pool. Do not assign a universal productivity penalty. Use the observed work pattern or an explicit assumption with a range.

Show slack as carefully as shortage. If writers have capacity but legal approval is binding, buying more drafting hides the problem. If the portfolio fits after one low-confidence demand estimate is corrected, the correct action may be to improve that estimate. The record should name the binding pool, the time bucket, the amount and range of shortfall, the bids affected and the smallest decision that changes feasibility.

• Test timing moves only inside verified windows.
• Move preparation without moving decision authority.
• Keep indivisible work in coherent protected blocks.
• Record slack so the wrong capacity is not purchased.
• Name the smallest intervention that removes the shortfall.

Portfolio choice

Compare feasible combinations, not isolated bid scores

A bid can rank first on its own and still produce a weaker portfolio. It may consume the only specialist needed by two smaller pursuits whose combined contribution, learning or strategic coverage is stronger. The reverse can also be true: splitting across the two can leave both below their credible response floor. Construct complete scenarios that state which gates are funded, which work remains protected and which bids wait or release capacity. Infeasible scenarios stay visible with the violated constraint.

Use current evidence on eligibility, contribution, award probability, strategic outcome, contractual exposure and timing, but keep unlike measures separate. The 2026 Green Book supports comparing options, opportunity cost, uncertainty, sensitivity and switching values. It is public-sector appraisal guidance, not a bidder ranking formula. The transferable discipline is to expose alternatives and the assumption that changes the preference. Do not turn every consideration into points and call the resulting total objective.

Run at least a current case and the credible adverse corner that matters: lower supply, higher effort, a delayed approval or a delivery incident. Matched scenarios prevent a portfolio from using high contribution, low effort and high capacity assumptions that cannot coexist. The authorized decision can prefer resilience over the highest central estimate. Record the reason, contrary evidence and the switching event that would justify another mix.

Control

An allocation works only when lower-priority bookings are released

Issue a decision for every included pursuit. `fund_to_next_gate` grants named capacity through a bounded output. `maintain_minimum_safe_capacity` preserves a credible floor but defers enhancements. `conditional_reserve` holds a limited slot until specified proof arrives. `waitlist` grants no current booking and names the release condition. `pause_and_release` and `stop_and_release` require owners to cancel meetings, tasks, contractor requests and provisional holds. `indeterminate` records the missing fact. `authority_missing` blocks implementation.

Each granted block needs a pool, amount, period, work package, accountable bid owner and capacity owner. State whether unused capacity returns automatically and whether one pursuit may borrow from another. A waitlist must not become shadow work carried through favors. A pause should specify what record remains current and what must be revalidated before restart. If a stop is required because facts changed after work began, use the dedicated live-bid stop process rather than hiding closure inside the capacity plan.

GovS 002 treats portfolio management as recurring work and calls for current component records, defined planning horizons, allocation rules and approved changes. The Orange Book adds risk appetite, clear responsibility, current reporting and escalation triggers. Those principles support a short-lived allocation record. It expires when the horizon ends, a bid passes its funded gate or a named material event occurs. The old record remains available for explanation and calibration, but it no longer grants capacity.

Agent handoff

Give agents a decision record they can verify without exposing people

Consider an illustrative six-week portfolio with four live bids. After protected client work, the team has twenty-four proposal days, five security-review days and three delegated pricing slots. Bid A needs eight proposal days and two security days to reach solution freeze in week two. Bid B needs ten proposal days and two pricing slots by week five. Bid C requests six proposal days, but its mandatory certification remains unresolved. Bid D needs nine proposal days and three security days for a later submission. These numbers are an example, not a productivity benchmark.

The record does not rank all four and fill from the top. It withholds C from the feasible set until the certification gate is resolved. It tests A plus B, A plus D and B plus D against time buckets and minimum plans. If A plus D exceeds security supply in week two even though the six-week total fits, the scenario is infeasible unless an approved timing or staffing change removes that specific conflict. The human authority may fund A and B, place D on a dated waitlist and assign C an eligibility action with no response booking.

An agent receives pseudonymous capacity pools, source references, ranges, constraints, decisions, contrary evidence and expiry. It can answer why D is waiting, what additional security capacity would change and which source made C conditional. It should not receive private salary, health or performance data. If proposal management, research, writing or submission coordination remains the proved bottleneck after these choices, the owner can assess a bounded Zelius engagement. If the blocker is certification, delivery feasibility, product fit or internal approval, more bid delivery capacity will not fix it.

• Expose source identifiers, observation times and confidence for every imported input.
• Use pseudonymous resource pools unless a named identity is required for authority.
• Return violated constraints with every infeasible scenario.
• Distinguish recommendation, authorization and completed reassignment.
• Require separate permission before staffing, spending, communication or submission.