When a project budget blows up, the story is rarely a single bad decision. It is a hundred reasonable ones that nobody added up. A contractor extended a week here, a software license renewed there, a vendor invoice came in higher than the estimate. Each was approved by someone with the authority to approve it. None of them, individually, looked like a problem. Together they put the project 18 percent over, and the PMO found out at quarter end.
Controlling portfolio spend is mostly about closing that visibility gap. The schedule gets watched obsessively. The money, and specifically the money that has been committed but not yet invoiced, often does not. Cost is one of the three constraints project portfolio management is supposed to steer, and it is the one most portfolios watch last.
Key takeaways
- Track committed cost, not just actuals. A signed purchase order is real money even before the invoice arrives.
- Tie every commitment back to a project and a budget line so overruns surface early.
- Match approval thresholds to risk, so small spend moves fast and large spend gets real scrutiny.
Committed cost is the number that hides
Most budget tracking watches actuals: money that has been invoiced and paid. But by the time a cost is an actual, it is history. The number that lets you steer is committed cost, the spend you have already agreed to through purchase orders, contracts, and approved work, even though the invoice has not landed. A project can look on budget on actuals while carrying a stack of commitments that will push it well over once they clear. Most of those commitments are born in paperwork, which is why turning project documents into portfolio data matters as much as the tracking itself.
The fix is to track commitments the moment they are made, not when they are paid. Every purchase order against a project should reduce the remaining budget immediately, so the picture reflects what you have actually obligated.
Tie every commitment to a project and a line
Spend control breaks down when purchases float free of the plan. A budget overrun is easy to catch when every purchase order is coded to a specific project and budget line and someone reconciles commitments against the plan on a regular cadence. It is nearly impossible to catch when procurement happens in a separate system that the PMO only sees at month end. Once commitments are coded properly they become a reportable number, and committed-versus-approved spend earns a place on the portfolio dashboard alongside the schedule view. Getting that coding to hold without manual reconciliation is usually a tooling question, and the financial tracking built into PPM tools is what keeps commitments tied to the plan as they are raised.
This is also where approval discipline matters. Purchase orders need a clear approval path, with thresholds matched to risk: small, routine spend should move quickly, while anything large enough to dent a budget should get real scrutiny before it becomes a commitment. Teams managing high purchase-order volume across many projects often run this through dedicated purchase order management software so that every commitment is captured, routed for approval, and visible against the budget rather than living in a chain of email approvals. The principle holds regardless of tooling: a commitment you cannot see is a commitment you cannot control. The same discipline extends to the vendors and contractors those purchase orders pay for, where tracking insurance and vendor and contractor compliance keeps committed spend from turning into an unmanaged liability.
Reconcile on a cadence, inside governance
Budget control is not a year-end audit. It belongs in the same rhythm as the rest of portfolio steering. Pulling committed cost against plan into the regular portfolio governance review means overruns surface while there is still time to do something: renegotiate, reprioritize, or pause downstream spend. Wait for the formal financial close and the only option left is to explain the variance. Where the money should go instead is a prioritization question, which is why budget reviews work best alongside a live view of how the portfolio is prioritized.
Budget and capacity are the same conversation
Money and people are two views of the same constraint. A project that needs more budget usually needs more of a scarce skill too, and both signals should land in the same place. When you review whether a project still deserves funding, review whether it still deserves capacity, using the same lens described in resource and capacity planning. Steering a portfolio means steering both, together, rather than discovering the budget problem and the staffing problem in separate meetings a month apart.
The four numbers every project budget line needs
Most budget reports carry two numbers, approved and spent, and those two cannot tell you whether a project is in trouble. You need four, and the useful information lives in the gaps between them rather than in any single figure.
| Number | What it means | What the gap tells you |
|---|---|---|
| Approved budget | The amount governance signed off, at the budget line level, not just the project total | Against committed: how much room is left to obligate |
| Committed | Purchase orders, contracts, and approved work you are legally or practically obligated to pay | Against actual: the size of your blind spot, the money owed but not yet billed |
| Actual | Invoiced and posted | Against approved: the only number most reports show, and the least useful for steering |
| Forecast at completion | Actual, plus open commitments, plus estimated remaining uncommitted work | Against approved: the variance you will actually have to explain |
Swipe to see more →
The pair that matters most is committed against actual. A project showing 40 percent of budget spent, with another 45 percent already committed on open purchase orders, is not at 40 percent. It is at 85 percent with almost no room left, and every report that hides the middle two rows will tell the steering committee it is fine. Forecast at completion is where this connects to earned value management, which formalizes the same idea with schedule performance folded in.
Commitment lag decides whether your budget view is real
Here is the measurement worth adding, because it explains why budget reports are wrong more often than they are late. Commitment lag is the average number of days between a commitment being made and that commitment appearing in the portfolio budget view. Not when it is invoiced. When it becomes visible.
| Commitment lag | What it means in practice |
|---|---|
| 0 to 2 days | The budget view is close to real time. Overruns surface while you can still act on them. |
| 3 to 10 days | Workable. You will catch most problems inside the monthly cycle. |
| 11 to 30 days | You are steering on a picture that is one reporting cycle stale. Expect surprises at close. |
| Over 30 days | You are not managing the budget, you are reporting history. Fix this before buying anything. |
Swipe to see more →
Measure it by sampling: take twenty commitments raised last quarter, find the date each was approved, find the date each first appeared against a project in the portfolio view, and take the average. Treat these bands as a starting calibration against your own history rather than an industry constant. Most organizations that run the sample for the first time are surprised, because the lag is not caused by the tooling. It is caused by nobody owning the coding step between procurement and the portfolio.
What a reconcilable commitment record actually contains
A commitment you cannot reconcile is a number you will end up arguing about. These are the fields that make the difference between a purchase order log and a commitment ledger you can steer with.
| Field | Why it is needed |
|---|---|
| Purchase order or contract reference | The join key back to finance. Without it, reconciliation is manual forever. |
| Project and budget line code | Project alone is not enough. Line level coding is what makes an overrun attributable. |
| Vendor and legal entity | Entity mismatches are a common cause of invoices that cannot be matched to a commitment. |
| Committed amount and currency | The obligation, excluding tax handling stated explicitly one way or the other. |
| Invoiced to date | Lets you compute the open balance, which is the number that belongs on the dashboard. |
| Open balance | Committed minus invoiced. The actual exposure still to land. |
| Expected clear date | Turns a flat number into a cash and period forecast. |
| Fiscal period split | Multi year commitments booked entirely in year one are a top source of phantom overruns. |
| Named owner | Somebody must be answerable for closing or releasing it. Unowned commitments never get released. |
Swipe to see more →
Approval thresholds that do not get routed around
Thresholds are usually set by asking how much risk finance is willing to tolerate. That is the wrong question, and it produces the classic failure: thresholds set so low that the approval path becomes an obstacle, and people route around it. The right question is what level of scrutiny actually changes a decision.
| Commitment size | Approval that adds value | What goes wrong if you over control it |
|---|---|---|
| Routine, well below a budget line's tolerance | Project manager, recorded not reviewed | A queue forms, work stalls, and people buy on expense cards instead |
| Material to one budget line | Budget line owner, same week | Purchases get split to stay under the line |
| Material to the project total | Project sponsor, with the revised forecast attached | Approvals become rubber stamps because there are too many |
| Material to the portfolio, or multi year | The governance forum that approved the budget | Nothing. This tier should be slow. |
Swipe to see more →
Purchase order splitting is the signal to watch
When a threshold is too low, spend does not shrink. It fragments. The same vendor receives three purchase orders of 4,800 dollars in the same month against a 5,000 dollar approval limit, and each one is individually compliant. Splitting is easy to detect once you look: group commitments by vendor, project, and month, and flag any group whose total crosses a threshold that none of its members crossed. A pattern of near threshold values is the tell. Treat it as a design problem with the threshold rather than misconduct by the person, because that is almost always what it is.
The three routes to portfolio spend visibility, and what each costs
This is where most advice on the subject stops being useful, because the people writing it sell one of the three routes. The honest version is that the cheapest route wins more often than anyone selling the other two will tell you.
| Route | Works when | Where it breaks |
|---|---|---|
| A commitment spreadsheet with a named owner and a weekly update | Under roughly 50 open commitments across the portfolio, one currency, one approval chain. Costs nothing and consistently beats an unconfigured tool. | Volume, multi entity structures, or the moment the owner goes on leave |
| The finance or ERP system you already own | Finance already raises every purchase order and will add a project code field. This is the most commonly overlooked option. | Project level reporting is often poor, and getting a portfolio view out means an export and a rebuild every month |
| Dedicated purchase order or PPM financial tooling | High commitment volume, several approval chains, or a genuine need for real time visibility across many projects | Buying before you have defined budget line coding just automates an undefined process faster |
Swipe to see more →
The sequence matters more than the choice. Define the budget line coding, name the owner, and agree the approval thresholds first. Only then does the routing and approval workflow that dedicated purchase order software automates have something coherent to automate. Teams that buy first almost always end up configuring the tool twice.
Internal labor is the largest cost line with no purchase order
On most internal portfolios, the biggest cost on the project is the salaried people working on it, and none of it generates a purchase order. So a spend control process built entirely around purchase orders can be operating perfectly while missing the majority of what the project costs.
You do not need full time tracking to close this. Take the planned allocation from the resource plan, apply a blended internal rate agreed once with finance, and post it as a committed cost line at the start of each period. It will be approximate. It will also be far closer to the truth than zero, and it makes the trade off visible when a sponsor asks whether to add two more engineers or buy a service. The allocation numbers come straight from resource allocation, which is why budget and capacity belong in the same review.
The monthly commitment reconciliation runbook
Seven steps, roughly ninety minutes once the coding is clean, and it is the single highest value recurring finance task the PMO owns.
- Pull all commitments raised since the last reconciliation and confirm each carries a project and budget line code. Anything uncoded is the exception list, and it gets resolved first.
- Update invoiced to date against each open commitment and recompute the open balance.
- Close commitments that are fully invoiced, and release the residue on commitments that are finished but underspent. Unreleased residue is why portfolios appear to have less headroom than they do.
- Recompute forecast at completion per budget line as actual plus open balance plus estimated remaining uncommitted work.
- Flag every line where forecast exceeds approved, and every line where the committed plus actual figure has crossed 85 percent while significant scope remains.
- Run the splitting check: group by vendor, project, and month, and flag totals that cross a threshold no member crossed.
- Take the flagged lines, and only those, to the governance forum with a recommendation attached, not just a variance.
Step seven is the one that gets dropped, and dropping it is what turns reconciliation into bookkeeping. A variance with no recommendation moves the problem to a room full of people with less context than you have.
Six places portfolio budgets actually leak
| Leak | How it shows up | The control |
|---|---|---|
| Change orders never re-baselined | Scope grew and was approved, but the budget line still shows the original number, so everything looks over | Approving a change order and revising the budget line are one action, never two |
| Multi year commitments booked in year one | A three year license lands entirely in the current period and destroys the forecast | Fiscal period split as a required field on the commitment record |
| Tax, shipping, and currency excluded | Invoices arrive consistently above the commitment and nobody knows why | State the tax and currency treatment on the purchase order template itself |
| Underspent commitments never released | The portfolio looks fully committed while carrying dead headroom nobody can find | Step three of the monthly reconciliation, enforced |
| Internal labor uncosted | Build versus buy decisions get made on a comparison where one side is free | A blended internal rate posted as a committed line each period |
| Spend against cancelled work | The project stopped, the subscriptions and contractors did not | A commitment close out step in the cancellation process, covered in the project cancellation checklist |
Swipe to see more →
Where this fits alongside the rest of portfolio spend control
| Question | Where it is answered |
|---|---|
| How do I build a project budget in the first place | Project budgeting |
| How do I combine cost and schedule performance into one forecast | Earned value management |
| How do I justify the spend before it is approved | The project business case |
| How do I decide which projects deserve the money at all | Portfolio prioritization |
| How do I split the portfolio between run, grow, and transform spend | Run grow transform |
| How do I keep the vendors behind these commitments compliant | Vendor and contractor compliance |
Swipe to see more →
Frequently asked questions
What is the difference between committed cost and actual cost?
Committed cost is spend you have already agreed to through purchase orders, contracts, and approved work, even though the invoice has not arrived yet. Actual cost is spend that has been invoiced and paid. The distinction matters because a project can look on budget on actuals while carrying a stack of commitments that will push it over once they clear. Tracking commitments the moment they are made is what lets you steer before the overrun lands.
How do you control project budget in a portfolio?
Control project budget by tracking committed cost rather than only actuals, coding every purchase order to a specific project and budget line, and reconciling commitments against the plan on a regular cadence inside portfolio governance. Match approval thresholds to risk so small spend moves fast and large spend gets real scrutiny. The core rule is simple: a commitment you cannot see is a commitment you cannot control.
What is a purchase order in project budget management?
A purchase order is an approved commitment to spend against a project, issued to a vendor before the invoice arrives. In budget management it should reduce the project's remaining budget the instant it is approved, not when it is paid, so the budget picture reflects what you have actually obligated. Purchase orders coded to a project and budget line are what make overruns surface early rather than at quarter close.
How often should you review project budgets?
Review project budgets on the same cadence as the rest of portfolio steering, typically monthly, by pulling committed cost against plan into the regular portfolio governance forum. Reviewing this often means overruns surface while there is still time to renegotiate, reprioritize, or pause downstream spend. Waiting for the formal financial close leaves you explaining a variance instead of preventing it.
Should a purchase order reduce the budget when it is raised or when it is invoiced?
When it is raised. The moment a purchase order is approved you are obligated, so the remaining budget should drop immediately even though no invoice exists. Reducing the budget only on invoice means the budget view lags the obligation by however long the vendor takes to bill, which is commonly 30 to 60 days, and that lag is where overruns hide until they are unfixable.
What is the difference between a purchase order and a contract commitment?
A purchase order is a specific approved order for defined goods or services, usually a fixed amount against one budget line. A contract commitment is the broader obligation created by a signed agreement, which may cover multiple years, several purchase orders, or a rate card with no fixed total. Both belong in the commitment ledger, but contracts need a fiscal period split so a multi year obligation does not land entirely in the current period.
How do you calculate remaining budget on a project?
Remaining budget is the approved amount minus actual invoiced cost minus the open balance on all commitments. Subtracting only actuals is the common error, and it overstates available budget by the whole value of open purchase orders. Compute it at budget line level rather than project level, because a project can be within its total while a specific line is already exhausted.
Who should approve purchase orders on a project?
Match the approver to what the commitment threatens. Routine spend needs a project manager recording it, spend material to one budget line needs that line's owner, spend material to the project total needs the sponsor with the revised forecast attached, and anything material to the portfolio or spanning multiple years goes back to the governance forum that approved the budget. Setting every threshold low does not increase control, it causes purchase splitting.
What is committed cost in project accounting?
Committed cost is the value of obligations you have entered into but not yet been invoiced for, including open purchase orders, signed subcontracts, and approved change orders. It sits between the approved budget and actual cost, and it is the number that turns a budget report from a record of the past into something you can steer with. The gap between committed and actual is the portion of your spend that is real but invisible on an actuals only report.