Most construction reports are full of numbers and short on signal. Percent complete, dollars spent, number of RFIs, weather days, manpower charts. Executives flip through them looking for one thing: is this project going to finish on time and on budget, and if not, how bad is it? Too often the report doesn't answer.
The fix isn't more data. It's a short list of key performance indicators that each answer a real question, have a clear definition, and come with a threshold that tells you when to act. This guide sets out the ten KPIs we think owners should actually track, grouped by schedule, cost, change, risk, quality and safety, with the definitions and trigger points that make them useful.
What makes a KPI worth tracking
A good owner KPI passes three tests. It's forward-looking, telling you something about where the project is heading, not just where it's been. It's hard to game, because it's calculated from data the owner can check. And it has a threshold, a point at which someone has to explain what's happening and what they'll do about it.
Percent complete fails all three. It's backward-looking, it's usually a judgment by the party being measured, and there's no level at which it triggers action. It belongs in the report, but it isn't a KPI.
View the data
| Area | KPI | Example trigger to act |
|---|---|---|
| Schedule | 1. Forecast completion vs baseline | Slip on a fixed date, or worsening trend |
| Schedule | 2. Baseline Execution Index (BEI) | Below 0.95 |
| Schedule | 3. Critical Path Length Index (CPLI) | Below 0.95 |
| Cost | 4. Forecast final cost vs budget | Overrun above remaining contingency |
| Cost | 5. Cost performance index (CPI) | Below 1.0 for three months running |
| Cost | 6. Contingency vs work remaining | Burn running ahead of % complete |
| Change | 7. Change cost factor, pending changes | Pending backlog growing month on month |
| Risk | 8. Risk exposure vs contingency | Exposure larger than cushion left |
| Quality | 9. Deficiencies, NCRs, RFI turnaround | Aged NCRs; RFIs past contract time |
| Safety | 10. Total recordable incident rate | Any upward trend; any lost-time case |
Schedule KPIs
1. Forecast completion versus baseline
The single most important schedule number: how many days the forecast substantial completion date has moved from the approved baseline, and in which direction. Track it every month and show the trend. A project that slips five days a month is telling you something that a single snapshot won't.
The forecast has to come from an updated, logic-driven CPM schedule, not from the contractor's narrative. Our guide to reviewing a construction schedule covers what makes that forecast credible.
2. Baseline Execution Index (BEI)
BEI compares the number of activities actually completed with the number that the baseline said should be complete by the status date. It's one of the checks in the U.S. Defense Contract Management Agency's 14-point schedule assessment, widely used as a schedule health standard. A BEI of 0.95 or higher is generally treated as passing.
BEI is useful because it counts tasks, not dollars. It flags a project that's quietly falling behind on the many small activities that feed the critical path, often before the completion date moves.
3. Critical Path Length Index (CPLI)
CPLI measures how efficiently the remaining critical work has to be performed to hit the contract date. It's the remaining critical path duration plus its total float, divided by the remaining critical path duration. A value of 1.0 means the team must achieve exactly one day of critical work per day elapsed. Below 1.0, they need to work faster than planned. Analysts commonly flag a CPLI below 0.95 for further review.
CPLI is a good counterweight to optimistic narratives. If it's below 0.95 and the contractor says they'll recover, ask what will change to make the remaining work go faster.
Cost KPIs
4. Forecast final cost versus approved budget
The cost equivalent of forecast completion: the estimate at completion compared with the approved budget, including approved, pending and anticipated changes. Show it as a number and as a trend. Our guide to construction cost forecasting explains how to build that forecast and test it.
5. Cost performance index (CPI)
CPI compares the value of work completed with what it actually cost. Below 1.0 means the project is getting less than a dollar of work for each dollar spent. It's most useful as a cross-check on the bottom-up forecast. If CPI has been 0.92 for six months and the forecast says the project will finish on budget, someone needs to explain why performance will suddenly improve. Our guide to earned value management covers CPI and its companion, SPI.
6. Contingency remaining versus work remaining
Contingency should be drawn down roughly in step with progress and risk retirement. If 40% of the work is done and 60% of contingency is gone, the remaining 60% of the work has less than half the cushion it was planned to have. Plot contingency remaining against percent complete every month. Our guide to construction contingency covers how to set the original amount.
View the data
| Work complete | Planned contingency left | Actual contingency left |
|---|---|---|
| 0% | 100% | 100% |
| 20% | 88% | 74% |
| 40% | 70% | 45% |
Change KPIs
7. Change cost factor and pending changes
The Construction Industry Institute's benchmarking program defines a change cost factor as the total cost of changes divided by the actual total project cost. Tracked during the job, it tells you how much of the project is being bought through change orders rather than the original competitive price.
Pair it with two working measures: the value of pending changes not yet agreed, and how long they've been outstanding. Pending changes are where forecast surprises hide. A growing backlog of unpriced change requests usually means the cost forecast is understated. Our guide to managing change orders covers how to keep that backlog under control.
Risk KPI
8. Quantified risk exposure versus contingency
Take the open risks in the risk register, estimate the likely cost and schedule effect of each, and compare the total with the contingency and float remaining. If the exposure is larger than the cushion, the forecast is optimistic. This turns the risk register from a list of worries into a number the board can act on.
Keep it simple. The top ten risks, each with a probability, a cost range and an owner, are worth far more than a register of two hundred items that nobody reads.
Quality KPI
9. Open deficiencies, non-conformances and RFI turnaround
Quality is hard to capture in one number, so track a small set together: open non-conformance reports and their age, deficiencies identified at each inspection stage, and the average turnaround on RFIs and submittals. Slow RFI answers are often the owner's or design team's problem, and they turn into delay claims later.
The threshold is about trend rather than an absolute number. Deficiency counts that keep rising late in the job, or NCRs that stay open for months, are a warning about closeout and turnover.
Safety KPI
10. Total recordable incident rate (TRIR)
TRIR expresses recordable injuries and illnesses per 100 full-time workers. The U.S. National Safety Council gives the formula as the number of recordable cases multiplied by 200,000, divided by total hours worked. The 200,000 represents 100 workers at 40 hours a week for 50 weeks. Ask for the project TRIR monthly, along with hours worked so you can check the arithmetic.
Owners should also track leading indicators such as inspections completed and near-miss reports. TRIR tells you how safe the site was. Leading indicators tell you how safe it's likely to be.
Setting thresholds that mean something
A KPI without a trigger is just another number. For each one, agree in advance what green, amber and red mean, and what each level requires. A simple pattern:
- Green: within tolerance, report only.
- Amber: outside tolerance, the project manager explains the cause and the recovery plan in the monthly report.
- Red: escalated to the executive sponsor with options and a decision needed.
Use published thresholds where they exist, such as BEI and CPLI at 0.95, and set your own for the rest based on your risk tolerance. On programs we've supported, a forecast slip of more than two weeks on a fixed opening date, or a forecast overrun larger than the remaining contingency, were common red triggers. What matters most is that the thresholds are written down before the project starts, not negotiated after a bad month.
Where KPIs go wrong
- Too many. Thirty indicators means none of them gets attention. Ten is plenty at the project level.
- Reported by the party being measured with no independent check. The owner or owner's rep should be able to recalculate every KPI from source data.
- No trend. A single month's value means little. Show at least the last six months.
- Thresholds that move. If the baseline is quietly reset every time it's missed, the KPI tells you nothing.
What owners should do
- Pick your ten. Use the list above as a starting point and adapt it to the project, but keep it short.
- Write definitions and thresholds into the reporting requirements of the contract or the project execution plan, so they're agreed before work starts.
- Require the source data, including native schedule files, cost reports and hours worked, so KPIs can be checked independently.
- Report trends, not snapshots, and put red items at the top of the report.
- Tie each red KPI to a decision the executive team is being asked to make.
The bottom line
Owners don't need more construction data. They need a small set of well-defined indicators that look ahead, can be checked, and trigger action at agreed points. Ten good KPIs, tracked consistently and reported without spin, will tell an executive more about a project than a fifty-page monthly report. Our monthly report template shows where they fit.
Want help setting up KPIs and thresholds for your next project or program? Book a 30-minute call with Benjamin R. Consulting.
Sources: Deltek Acumen, DCMA 14-Point Assessment Metric Descriptions; Ten Six Consulting, Understanding the DCMA 14-Point Assessment: Baseline Execution Index; Ten Six Consulting, DCMA 14-Point Assessment and the Critical Path Length Index; Sanders, S.T., Change Management Best Practice Use in NAVFAC and Other Public Projects (University of Texas at Austin, 2000), citing CII Benchmarking and Metrics definitions; National Safety Council, Injury Facts: TRIR definition. Thresholds other than BEI and CPLI are illustrative. This article is general information, not legal advice.