IT Project Budget Overruns: Why Software Goes Over Budget and What to Do About It
Large IT projects run 45% over budget on average. When they exceed 50% over, the average overrun becomes 447%. Understanding why helps avoid it.
Why IT Projects Are Uniquely Hard to Estimate
Unlike construction -- where you measure progress in cubic metres of concrete and linear metres of steelwork -- software has no physical progress visible. A construction project manager can stand on site and estimate 40% completion. A software project manager looking at a codebase cannot reliably do the same.
Key factors that make IT estimation uniquely difficult:
- ‣Requirements volatility: Business requirements change faster than estimates update. What was scoped in month 1 looks different by month 3.
- ‣Integration complexity: The interfaces between systems are consistently underestimated. Two systems that work independently fail when connected.
- ‣Technical debt: Accumulated poor decisions from previous development phases create hidden cost. Debt is invisible in estimates but very visible in delivery time.
- ‣Vendor management: Offshore development teams, third-party integrations, and SaaS dependencies add failure points that are hard to cost upfront.
- ‣The 90% syndrome: IT projects often report 90% complete for months while the last 10% takes as long as the first 90%.
IT-Specific Overrun Causes
Requirements volatility
The most common IT overrun cause. Business stakeholders change requirements throughout delivery, often without recognising the cost impact. Each change seems small; accumulated, they double the scope.
Cloud and infrastructure cost growth
Cloud bills consistently exceed estimates as usage scales. AWS, Azure, and GCP costs are hard to predict without production traffic data. Data egress charges, idle resources, and unoptimised workloads add up. Link: platformengineeringcost.com for cloud cost benchmarks.
Underestimated integration complexity
Connecting to existing systems, legacy databases, and third-party APIs is consistently 3-5x harder than estimated. Data quality issues, API rate limits, and authentication complexity all add time.
Shadow IT and scope additions
Business units add features informally, bypassing formal change control. Each feature seems trivial. By go-live, the system is 40% larger than specified.
Offshore and vendor management failures
Poorly specified requirements, time-zone communication friction, and quality gaps in offshore teams add rework. Fixed-price offshore contracts often result in reduced scope rather than cost savings.
Technical debt repayment
Legacy system rewrites frequently uncover far more complexity than expected. The legacy system was built over 20 years; recreating it in 2 years always takes longer than planned.
Famous IT Project Overruns
| Project | Original Budget | Actual Cost | Outcome |
|---|---|---|---|
| NHS National Programme for IT (NPfIT) | £6.4B | £10B+ | Cancelled 2011 -- never delivered |
| FBI Virtual Case File | $170M | $170M spent | Abandoned 2005 -- replaced by newer system |
| Healthcare.gov (US) | $93.7M | $1.7B+ | Failed at launch in 2013, required emergency rebuild |
| Australian Census 2016 | $9.6M (IT portion) | $30M+ | Site crashed on census night due to DDoS and capacity failure |
| Universal Credit (DWP UK) | £2.4B | £6B+ | Ongoing since 2012 -- 10 years late, partial delivery |
Prevention Strategies for IT Projects
- ✓Definition of done: Every sprint story must have clear acceptance criteria before development starts. Prevents endless revision cycles.
- ✓Sprint-level budget tracking: Track cost per story point. If cost per point rises, it is an early warning of scope creep or team performance issues.
- ✓Cloud cost alerts: Set billing alerts at 50%, 75%, 90% of monthly budget. Never discover a $500K cloud bill at month end. Use AWS Cost Explorer or Azure Cost Management.
- ✓Monthly budget reviews with EVM metrics: Even in agile projects, a monthly CPI review using story points as Earned Value is feasible and effective. A CPI below 0.9 for two consecutive months requires escalation.
- ✓Change control in agile: Every item added to the backlog that was not in the original scope requires a documented change order with cost impact assessment. This sounds bureaucratic but prevents death by a thousand cuts.
- ✓Fixed-price contracts where scope is known: For well-defined modules, fixed-price subcontracts give cost certainty. For exploratory or novel work, time-boxed budgets (not open-ended) are the equivalent.
Frequently Asked Questions
What percentage of IT projects go over budget?
Large IT projects (over $15M) run 45% over budget on average, according to McKinsey. The Standish Group CHAOS Report 2020 finds 66% of technology projects either fail outright or are challenged. For the worst cases, McKinsey found IT projects that exceed their budget by more than 50% have an average overrun of 447%.
Why are IT projects harder to estimate than construction projects?
Software has no physical progress visible -- you cannot look at a codebase and estimate completion the way you can measure concrete poured. Requirements in software are elastic and change faster than estimates can be updated. Technical debt accumulates invisibly. Integration complexity between systems is consistently underestimated. Unlike construction, software team productivity varies enormously.
Does agile methodology reduce IT project budget overruns?
The evidence is mixed. Agile delivers faster feedback and reduces the risk of building the wrong thing, reducing wasted investment. However, agile can enable scope creep because there is no fixed end-point. Agile works well for reducing total cost of failure but does not inherently prevent budget overruns. The key is combining agile delivery with firm budget governance and sprint-level cost tracking.