There is no universal contingency percentage for a services project. Start with a complete base estimate, identify residual risks, estimate their effort or cost effects and review plausible scenarios. A probability-weighted allowance can support discussion, but it is not a promise that the project has enough budget for every adverse outcome.
Terms used in this guide. fee: the amount charged to the client; cost: the expense included in the stated project model; margin: revenue left after the included costs; margin percentage = that amount / revenue × 100; baseline: the approved original plan kept for comparison.
In plain words
Contingency is money or time set aside for risks you can see but cannot be sure about. There is no standard percentage. A good contingency is built from named risks: what could happen, how likely it is and what it would cost. It should be shown separately, not hidden inside the tasks.
Who this helps: Project managers, presales consultants and finance staff who price risk.
Questions this guide answers:
- How much contingency should you add to a project estimate?
- What is a typical contingency percentage?
- How do you calculate project contingency?
How do you calculate contingency from risks?
Event expected cost = probability of occurrence × incremental cost impact
Simple expected total = sum of the included event expected costs
This is an expected-value calculation, not a confidence reserve. Probabilities, impacts, overlaps and correlations need explicit review. Expected costs add up even when events are related; overlap means you must define the events so that no cost is counted twice, and correlation changes how likely a very bad total is, not the expected total.
What belongs in contingency rather than in the base estimate?
Contingency covers uncertainty inside the assumed scope. It is not a place to hide work you know about. Sort every item into one of three buckets before adding any reserve.
| Bucket | What it holds | Example |
|---|---|---|
| Base work | Everything required to deliver the known scope, including the easy-to-forget items | Testing, coordination, documentation, handover |
| Risk-response work | Planned actions that reduce a risk; they are certain, so they are base cost | A two-day API spike that will definitely be done |
| Residual contingency | The exposure left after the above | API rework, extra cleanup, an extra release rehearsal |
- A new client-requested feature is a scope change, not a reason to spend the allowance.
- Do not add a cost percentage on top of role hours that already cover the same risk: that counts it twice.
- Keep reserve and profit apart. A USD 4,000 allowance is money expected to cover risk-related cost; the target margin is what the firm wants after included costs. Using margin as hidden contingency makes the approval look stronger than the difficult case.
Example: sizing contingency for an IT project
Fatima, a presales lead, lists the residual risks after reviewing the base scope. For each she needs a probability and a cost if it happens.
| Risk | Assumed probability | Cost if it occurs | Weighted cost |
|---|---|---|---|
| API rework | 30% | USD 8,000 | USD 2,400 |
| Extra migration cleanup | 25% | USD 4,000 | USD 1,000 |
| Additional release rehearsal | 20% | USD 3,000 | USD 600 |
| Simple sum | USD 4,000 |
Reading the result
- On a USD 40,000 base cost, USD 4,000 is 10%. That percentage is an output of these assumptions, not a recommended default.
- It is an expected value across the events. One project does not suffer 30% of an API failure; it has the event, a partial version or nothing. USD 4,000 does not show that the reserve covers any single bad outcome.
- If all three events happen the cost is USD 15,000, taking the total to USD 55,000. A USD 44,000 budget would not cover it.
- Check the relationships: poor source data may cause both the API rework and the cleanup, so they are related. The expected costs still add, but the chance of several happening together is higher than independence would suggest.
Check probability and impact evidence separately. A supplier quote may support the USD 8,000 impact while the 30% remains expert judgment; record that difference. A low-probability, high-impact issue may need escalation even when its weighted cost is small.
How do you choose an allowance?
Fatima does not pick a number from the table. She works through four questions with the delivery lead and finance.
- What decision is this for? Expected cost suits a portfolio discussion. A single fixed-fee commitment may need a more conservative scenario.
- Can the risk be reduced instead of priced? A USD 1,500 investigation that materially reduces an USD 8,000 rework risk may beat carrying it. It needs a concrete test and an expected effect on the uncertain package.
- How big is the exposure against what the firm can absorb? Check whether risks can occur together and whether the contract adjusts for changed client dependencies.
- What does the price look like with and without it?
| Case | Cost | Fee at 30% | Margin if actual cost is USD 55,000 |
|---|---|---|---|
| Base plus the USD 4,000 allowance | USD 44,000 | USD 62,857.14 | 12.5% |
Showing that sensitivity (a USD 55,000 outturn on a USD 62,857.14 fee leaves 12.5%) makes the exposure clearer than a contingency line. Where a work package has a credible effort range, use three-point estimating and avoid adding a second reserve for variation already in that range. Write down which risk treatment covers which uncertainty.
How do you manage contingency during delivery?
Contingency is a reserve for risks you listed. Managing it means using it on purpose, recording why, and learning from what happened.
During delivery
- Keep the base estimate and the reserve in separate lines. If they are mixed, nobody can tell whether the reserve was used.
- Tie every draw to a named event and an owner. “API rework, owner: engineering manager”, not “extra labor”.
- Update the remaining risks when evidence changes or a risk happens.
- Send client scope changes through change control, not through the reserve.
Scenario: the API rework risk happens
- The project has a USD 4,000 reserve from the earlier example. API rework (cost USD 8,000) happens.
- The project manager records a draw of USD 4,000 against the event, so the reserve is now zero.
- The remaining USD 4,000 is real cost with no reserve behind it. She reports it: as a client change request if the client caused it, otherwise as a margin hit.
- She also reviews the two other risks, because the same poor data may now affect migration cleanup.
At project close
| What happened | What it teaches |
|---|---|
| The risk did not occur | The estimate was cautious, or luck; check the probability you used |
| A mitigation worked | The spike or early test paid off; keep doing it |
| The estimated event was implausible | Remove or lower that risk next time |
An unused reserve is not proof the estimate was wrong. Review the causes before cutting the next reserve, and keep scope additions out of the history so a client’s new requirements do not distort future judgment.
Can you start with a standard contingency percentage?
You can use an internal policy as an initial review rule, but test it against the actual scope and risk register. State that it is your policy, not an industry guarantee. Escalate projects whose risks differ materially from the work on which the policy was based.
Real example: a project manager sizes a risk reserve
An illustrative story with invented figures, to show how the idea works in a real office.
Fatima, a project manager, has a base cost of USD 40,000. She lists three risks: API rework (30%, USD 8,000 if it happens), extra migration cleanup (25%, USD 4,000) and an extra release rehearsal (20%, USD 3,000).
Each risk’s expected cost is probability times impact: 2,400, 1,000 and 600. The total is USD 4,000, which is 10% of the base cost.
She shows it as a separate line and explains that it is an average, not a promise: if the API rework happens alone it costs USD 8,000, twice the whole reserve. When it does happen she records a draw of USD 4,000 and tells the delivery head that USD 4,000 more is needed. If the cause was a client dependency, she raises the remainder as a change request; if it was the firm’s own estimating gap, it is reported as a margin hit.
When do you add contingency, and what comes before and after?
| Stage | Owner | Checkpoint |
|---|---|---|
| Build the base estimate | Presales consultant | Known work is complete before any reserve is considered |
| List residual risks | Presales lead with a delivery lead | Each risk has a probability, an impact and its evidence |
| Set the allowance | Risk owner and final approver | The choice and rationale are recorded; double counting is checked |
| Spend it | Project manager, with the approver | Use is authorized against a named risk and logged |
| Review | Project manager and finance | Unused reserve and overruns feed the next estimate |
Contingency, a range or a change order: how do you handle uncertainty?
There are several ways to handle uncertainty in an estimate.
| Approach | Use it when | Watch for |
|---|---|---|
| Risk-linked contingency (this guide) | Risks can be named | Needs risk discipline |
| Range estimate (low to high) | Early stage | Read as a quote |
| Three-point estimates per task | A few tasks carry the uncertainty | Averages read as promises |
| Change-order mechanism | Scope may legitimately change | Disputes about what changed |
Use named risks to inform both effort ranges and any separate reserve. A range shows alternative outcomes; contingency is an approved allowance against a stated decision basis. Use the change process for agreed scope changes and do not assume any one technique covers every uncertainty.
Project contingency: frequently asked questions
Is the sum of EMVs a 90% confidence reserve?
No. A probability-weighted expectation does not establish a percentile of total project cost. A confidence-based reserve needs a supported model of outcomes and their relationships. State the decision basis rather than attaching a confidence label to a simple sum.
Can contingency compensate for missing QA and release work?
Required activities belong in the base estimate. Adding a buffer without identifying them hides recurring omissions and weakens later calibration. Estimate known work first, then review the residual uncertainty.
Should the client see the contingency?
It depends on the commercial model. In T&M the client usually sees ranges and caps; in fixed price the reserve is often internal. Be consistent and honest about what is included.
How do I manage it during delivery?
Record each draw with the risk event and approver, report the remaining reserve in margin reviews and release it when the risk passes.
What is a typical contingency percentage for a project estimate?
There is no universal figure. Size it from the residual risks in your own project: in the worked example three risks add up to USD 4,000, which is 10% of a USD 40,000 base, but that is an output of the stated probabilities, not a rule.
How do you show contingency in Managed Margin?
First include known QA, coordination and release work in the strategy's role loading. Keep the residual risk calculation outside those base activities until its owner approves the treatment. In the illustration, base cost USD 40,000 plus a separately modeled USD 4,000 allowance gives USD 44,000. At 30% target and no revenue charges, the corresponding fee is approximately USD 62,857.14.
Represent the allowance through an appropriate configured cost rule, checking that it is not also embedded in inflated role hours. Compare a difficult effort or cost strategy separately if the USD 4,000 expected value does not describe the exposure being accepted. Managed Margin prices the configured assumptions; it does not derive event probabilities or a confidence reserve automatically.
Explore the pricing and estimation workflow or review the delivery margin workflow.
Background reading: three-point estimating: the PERT method and protecting margin on a fixed-price project.
Continue the sequence: guides to read next
- Why Extra Hours Reduce Margin on a Fixed-Price Project: why every extra hour comes straight off fixed-price margin
- Three-Point Estimating and the PERT Formula, With an Example: quantifying uncertainty on individual tasks
- Assumptions and Exclusions in an Estimate: 40 IT Examples: assumptions that become risks
- How to Manage Fixed-Fee Project Margins in Consulting: where contingency protects a fixed price
- From ROM to Fixed Price: What Evidence You Need First: what evidence you need before a range becomes a fixed price
Sources and methodology
Outside facts in this guide, and where they come from
- UK HM Treasury Green Book: optimism bias guidance: the finding that early cost and duration estimates are systematically optimistic, and that appraisers should adjust for it
The examples, templates and calculations are original editorial illustrations, not market benchmarks or customer results. Monetary amounts are in USD, exclude tax unless stated, and use the cost boundaries defined in the article. Validate assumptions against your own delivery history before quoting.
Further background: PMI: risk contingency reserve. The worked arithmetic and scenarios above are original; this reference provides context and is not a source for market rates or the stated example amounts.
The application walkthrough was checked against the current stage, pricing, cost-rule and delivery calculation code on 13 October 2026. Forecasts describe recorded actuals plus the remaining plan, not predicted customer outcomes.