Construction Cost Forecasting: A Practical Guide

What construction cost forecasting actually answers

Four questions get asked about project cost, and they are not the same question.

• Estimating asks: what should this project cost before the work begins?
• Budgeting asks: what amount has been approved, and how is it allocated?
• Cost reporting asks: what has been committed and spent so far?
• Forecasting asks: where is the project likely to finish based on what we know now?

Only the last one is forward-looking. An estimate is a pre-construction position. A budget is an authorization. A cost report is a record of what already happened. A forecast is a judgment about the end of the job, made with today's information.

That distinction matters because a forecast is supposed to move. If scope changed, a subcontract was bought above budget, a change is pending, or remaining work is now understood differently, the forecast should reflect it. A static budget with month-end actuals pasted next to it is not a forecast. It is a history report with an unchanged target beside it.

A forecast that never moves is not stable. It is unmaintained.

The practical forecast equation

Most working forecasts, regardless of the format they are reported in, reduce to one operating equation:

Projected Final Cost = Actual Cost to Date + Remaining Committed Cost + Forecast Remaining Uncommitted Cost + Expected Change / Risk Exposure

Teams structure this differently — by cost code, by phase, by trade, by contract — and some carry exposure in a separate column rather than inside the projection. The logic is the same in every version: account for what has already happened, what is contractually committed, what still has to be bought or performed, and what known exposure has not yet become an approved cost.

Three terms are worth defining precisely, because they get used interchangeably and should not be.

• EAC / projected final cost — the current best view of what the project will cost at completion.
• ETC / cost-to-complete — the expected remaining cost from today through completion.
• Variance at completion — the current approved or revised budget minus the projected final cost.

EAC is the number leadership reacts to. ETC is the number the project team can actually defend line by line. Variance at completion is the number that decides whether anyone has to do something about it.

The six inputs that make or break the forecast

1. Revised budget

The revised budget is the current approved baseline: original budget plus executed changes, with any approved transfers or contingency draws reflected. It is the reference point for every variance you report. The common failure mode is reporting against an original budget that stopped being true months ago, which makes every variance conversation start with an argument about which number is the baseline.

2. Actual costs to date

Actuals are invoices, payments, labor, and direct costs posted against budget lines. They are the most trusted input and the most misleading one used alone, because they lag the work. A trade can be 70% complete with 30% of its invoices processed. Reading actuals as progress systematically understates where the job is.

3. Remaining commitments

Committed cost is money that is contractually owed whether or not an invoice has arrived — subcontracts, purchase orders, executed change orders against those contracts. The failure mode is waiting for the invoice. A subcontract awarded above the budget line has already created a variance; if the forecast only recognizes it when the first payment application is processed, the team loses the window where the exposure could still be managed.

4. Pending and approved changes

Approved changes belong in the revised budget and the commitment position. Pending changes are the harder case: priced but unapproved, submitted but unresolved, identified but unpriced. Excluding them until approval produces a forecast that is technically clean and operationally useless, because everyone on the project already knows the exposure exists. Carry pending change exposure explicitly, labeled as pending, with a stated assumption about probable value.

5. Cost-to-complete for uncommitted scope

Any scope not yet bought still has to be forecast. Late-bid packages, allowances, owner-furnished items, general conditions running to a later completion date, and unawarded trades all carry real remaining cost. Treating an uncommitted line as zero because nothing has been spent or committed is the single most common way a forecast quietly understates final cost.

6. Risk and contingency assumptions

Contingency should correspond to remaining uncertainty: unresolved design issues, unpriced changes, procurement risk, schedule-driven cost, unfinished scope. The failure mode is contingency as a plug — a balance that makes the forecast land where the project wants it to land. If contingency is being drawn down while the risk picture is unchanged, the forecast is absorbing overruns instead of reporting them.

Four common forecasting methods

1. Bottom-up cost-to-complete

The project team reforecasts remaining cost line by line — by cost code, trade, or contract — using current buy information, remaining scope, and known changes. This is the most defensible method for an active project under real project controls, because every number in the summary traces to someone's stated assumption. It costs the most time, and it depends on discipline across the whole team rather than one person at the top of the report.

2. Trend-based forecasting

Trend methods project remaining cost from observed patterns: production rates, unit costs, burn rate, or the variance behavior of the job so far. Their real value is as a challenge check. If the bottom-up forecast says the remaining work will be delivered at budget while the first half of the job ran consistently over, the trend is telling you the assumptions deserve scrutiny. It is not a substitute for known scope information — a trend cannot know about a subcontract signed yesterday.

3. Percent-complete / top-down forecasting

Apply a completion percentage to the budget and extrapolate. This is fast, works reasonably on simple or repetitive scopes, and is fine for an early directional read. Its weakness is that it averages away exposure: a project 60% complete overall can be carrying one trade at double its budget and the top-down number will not show it.

4. Earned value / performance-index forecasting

Earned value methods compare the budgeted value of work performed against actual cost, then use the resulting performance indices to project final cost. In formal project-controls environments — federal work, large capital programs, anywhere reporting requirements demand it — this is rigorous and well understood. It requires disciplined progress measurement to mean anything, and it is one legitimate way to forecast rather than the only one. Teams without a reliable earned-value baseline are usually better served by a disciplined bottom-up cost-to-complete.

A worked example

Take a fictional project at a mid-construction reporting date.

• Revised budget: $10.0M
• Actual cost to date: $4.2M
• Remaining committed cost: $4.1M
• Forecast remaining uncommitted scope: $1.4M
• Expected pending change / risk exposure: $0.5M

Projected final cost: $4.2M + $4.1M + $1.4M + $0.5M = $10.2M. Variance at completion: $10.0M − $10.2M = $0.2M over budget.

Compare that to the reading available from actuals alone: the project has spent $4.2M of a $10.0M budget, so it appears to have $5.8M of room. That statement is true and useless. It ignores $4.1M already contractually committed, $1.4M of scope still to buy, and $0.5M of exposure the team is already tracking. The forecast says the project is over; the spend report says the project is comfortable.

One more point on this example: the $10.2M is only credible if each of its components is traceable by line item. A summary cell containing a single judgment cannot be reviewed, challenged, or explained when it moves. The forecast should be built from assumptions someone owns, not arrived at at the top.

The monthly forecasting process

Forecasting works when it is a cadence rather than an event. A workable monthly sequence:

1. Freeze the reporting cut-off date. Every input — actuals, commitments, changes — reflects the same date across the project.
2. Reconcile actuals and newly approved costs against the correct budget lines.
3. Update remaining commitments, including new awards and executed changes to existing contracts.
4. Review every open and pending change, with a stated position on probable value and timing.
5. Reforecast uncommitted scope and cost-to-complete by line item.
6. Challenge risk and contingency assumptions against the remaining risk picture, not the remaining balance.
7. Explain movement from the prior forecast — what moved, by how much, and why.

Step seven is the one most often skipped and the one that carries the most weight. Leadership should not only see the new number; they should see why the number moved. A forecast that arrives without an explanation of its movement invites the wrong reaction — either unwarranted alarm or unwarranted confidence.

Use the construction progress report template to carry that explanation into the weekly or monthly report.

The same principle holds across the rest of the reporting stack. If you are building this cadence from the ground up, the broader framework sits in our guide to construction project controls.

What a useful forecast review should show

Whatever format the report takes, a forecast review is complete when it shows:

• Revised budget
• Actual cost
• Committed cost
• Pending change exposure
• ETC / cost-to-complete
• Projected final cost / EAC
• Variance to budget
• Contingency remaining
• Forecast movement since the prior period
• Major assumptions and risks driving the movement

If a review cannot show the last two, it is a status report, not a forecast review.

Early warning signs the forecast is going stale

Stale forecasts rarely announce themselves. They show these symptoms first:

• The forecast has not changed for multiple periods while the project clearly has.
• Large uncommitted balances remain late in the project.
• Pending changes are excluded from the forecast until they are approved.
• Commitments are not reconciled to the budget lines they were bought against.
• ETC is copied forward period after period without being challenged.
• Contingency is being burned with no corresponding change in the stated risk exposure.
• The leadership report cannot explain movement by cost code or by cause.

Any one of these is worth a conversation. Three or more together usually means the forecast has become a reporting artifact that the project team no longer uses to make decisions.

Spreadsheet vs connected forecast

Spreadsheets forecast well under specific conditions: the scope is contained, the cost structure is stable, and one person controls the inputs and understands every assumption in the model. Plenty of good projects are forecast this way, and there is nothing wrong with it.

The conditions change as the project grows. Budgets live in one place, commitments in another, invoices in accounting, changes in a project-management system, approvals in email, and the forecast in a file that one person rebuilds every month. At that point the spreadsheet stops being the forecast and becomes a reconciliation layer — a monthly effort to reassemble a picture that already exists, in pieces, across four systems.

The issue is not that Excel is bad. It is whether the forecast inputs stay continuously connected to the records that create the cost. When a commitment is awarded, a change is priced, or an invoice is posted, the question is whether the forecast reflects it that day or at the next reconciliation. That connection is the practical job of construction cost management.

Forecasting at portfolio level

Project forecasting and portfolio forecasting need different things from the same records.

A project team needs line-item detail: which cost code moved, which subcontract was bought above budget, which change is pending, and what assumption sits behind every cost-to-complete number. Precision is the point.

Leadership needs something else: comparable definitions across projects so that “committed” and “pending” mean the same thing everywhere, consistent cut-off dates so the numbers can be read side by side, explanations of movement, and a reliable way to identify which projects need attention this month. That is the reporting problem behind capital project management and the reason portfolio reporting has to be built on the same records the project teams maintain rather than a separate roll-up.

When definitions drift between projects, portfolio forecasting degrades into comparing numbers that were produced by different rules — which is worse than no comparison, because it looks authoritative.

A 10-question forecasting checklist

1. Is every actual cost posted to the correct budget line?
2. Are remaining commitments current?
3. Are pending changes visible before approval?
4. Is uncommitted scope explicitly forecast?
5. Can ETC be explained by line item?
6. Are contingency assumptions tied to remaining risk?
7. Can we explain why EAC moved from last period?
8. Are schedule changes reflected where they create cost impact?
9. Is the forecast using the same cut-off date across the project?
10. Can leadership trace the summary number back to the underlying records?

For a wider review of the controls behind the forecast — approvals, commitments, changes, schedule, reporting and closeout — work through the construction project controls checklist.

How Jet.Build approaches forecasting

Jet.Build keeps budget, commitments, actual costs, change orders, pending exposure, and cost-to-complete on the same project record. Projected final cost and remaining cost-to-complete are calculated from those records — by line item, phase, trade, and cost code — so a team preparing a monthly forecast is reviewing and adjusting assumptions rather than reassembling the inputs from disconnected systems. The detail behind that sits on cost management, and the surrounding financial operation — accounting connectivity, invoices, contracts, and cash position — sits on financial management.

None of that removes the judgment. Someone still has to decide what the remaining scope will cost and what the pending changes are worth. What it removes is the reconciliation work that happens before the judgment can be applied.

Frequently asked questions

What is construction cost forecasting?

Construction cost forecasting is the process of projecting where a project's total cost is likely to finish based on actual costs, commitments, remaining work, changes, and known risk. Unlike the original estimate or approved budget, the forecast should change as the project changes.

What is the difference between EAC and ETC in construction?

Estimate at Completion (EAC) is the current forecast of total project cost at completion. Estimate to Complete (ETC), or cost-to-complete, is the expected remaining cost from the current reporting date through completion.

How often should construction cost forecasts be updated?

Many teams update formal forecasts monthly, with more frequent reviews on fast-moving or high-risk projects. The right cadence depends on project complexity, but the forecast should be updated whenever material changes in commitments, scope, schedule, or risk make the prior forecast stale.

What should be included in a construction cost forecast?

A practical forecast should account for the revised budget, actual costs, remaining commitments, pending and approved changes, uncommitted remaining scope, cost-to-complete, and risk or contingency assumptions.

Can construction cost forecasting be done in Excel?

Yes. Excel can work well for smaller or tightly controlled projects. The risk grows when forecast inputs are maintained by different people across accounting, project-management, change-management, and reporting systems, because the spreadsheet can become a separate reconciliation layer rather than a live reflection of project activity.