Optimism Bias on the Bid
A consultancy is bidding for a fixed-price, 18-month digital transformation. The bid team has produced a bottom-up estimate of £4.2m with a stated confidence of '80%'. The estimate was built by summing most-likely durations from a detailed plan, then adding a 7% contingency 'because that is what we usually win with'. The commercial lead has identified that the client's incumbent supplier is being replaced and that the client has a history of changing scope mid-programme; neither point appears in the estimate basis. The bid director wants to submit at £4.2m to be competitive, arguing that adding contingency 'signals weakness'. A rival bid is rumoured to be around £4.5m. The organisation's last three fixed-price programmes overran by 22%, 35% and 18%.
1. Critique the estimate's stated 80% confidence figure, identifying the specific analytical errors.
Read the worked answer
The '80% confidence' figure is not supported by the method used. Four errors:
- Summing most-likely values is not a confidence level. Adding mode durations across a network produces a point estimate near the least likely outcome for the whole project, because the probability that everything goes to plan simultaneously is very low. It is closer to a P20–P30 than a P80.
- No distributional analysis. Confidence requires a range and a probability distribution (Monte Carlo, or at minimum three-point estimates aggregated). A single summed figure cannot yield a percentile.
- Contingency is not derived. 7% is a negotiating convention, not an output of analysis. It happens to be roughly a third of the smallest historical overrun (18%), so it is demonstrably insufficient against the organisation's own record.
- Known risks are excluded from the basis. The incumbent-supplier transition and the client's scope-change history are known drivers of cost. Excluding them from the estimate basis understates both the base and the contingency, and means the risk allowance is not traceable to identified risks.
A defensible P80 would require an aggregated range estimate plus an explicit risk allowance for the identified commercial and scope risks.
2. The bid director argues that adding contingency signals weakness. How should the bid team respond, and what should they recommend submitting?
Read the worked answer
Reframe the argument. A price is a commitment; contingency is the price of being able to keep that commitment. Submitting £4.2m against a true P80 materially above that is not competitive - it is a transfer of the overrun to the consultancy's P&L. The organisation's own record (22%, 35%, 18% overruns) is the strongest evidence: at those rates, £4.2m becomes £5.1m–£5.7m in reality.
Separate the two decisions. What the client sees (price) and what the consultancy holds (risk allowance) are different things. The team can submit a competitive headline price while holding an internal management reserve, provided governance approves the exposure.
Recommendation
- Re-estimate with an aggregated range to produce a genuine P50 and P80.
- Add an explicit risk allowance for the two known drivers (supplier transition, scope-change history), traceable to named risks.
- Present the bid director with the P50/P80 spread and the expected value of the overrun at £4.2m, using the historical record as the prior.
- If commercial pressure requires a lower headline, escalate the residual exposure to the level with authority to accept it - do not let the bid team absorb it silently.
The rumoured £4.5m rival bid is consistent with a properly risk-adjusted price, not evidence that £4.2m is right.