What it means
Human reasoning is not a neutral weighing of evidence. Once a person holds a view, new information is filtered through it: supporting facts are salient and easily accepted, contradicting facts are scrutinised for flaws or not noticed at all, ambiguous facts are read in the belief's favour, and memory later retrieves the supporting facts more readily.
The result is a belief that feels increasingly well-founded regardless of whether it is true. Psychologists have documented the effect for decades; it appears in every population and every domain, and expertise does not remove it, though expertise does improve the quality of the belief that the bias then protects.
In financial work the bias has particular channels. Forecasting: a sales forecast built by someone who wants the sales to happen weights the pipeline optimistically and treats the lost deals as anomalies.
Investment: an analyst who has published a buy recommendation reads the subsequent results as confirming it, and the sell recommendation comes late. Credit: a relationship manager who has championed a borrower sees the covenant breach as technical.
Budgeting and business cases: the sponsor of a project collects the data that supports it and presents the risks as manageable. Audit and review: a reviewer who expects the figures to be right finds them right; an auditor who trusts management accepts management's explanation for the unusual item.
Strategy: a leadership team that has committed to a course reads the market as validating it until the market makes the point unmistakably. Each of these has produced expensive failures whose post-mortems found that the contradicting evidence had been available and had been discounted.
The bias compounds with related effects. Anchoring fixes the initial estimate and biases adjustment towards it.
Sunk cost commitment makes abandoning a belief costly and so makes evidence against it unwelcome. Group dynamics amplify the bias when everyone in the room holds the same view and dissent is socially expensive.
Incentives sharpen it when the person's bonus, reputation or job depends on the belief being right. Because the bias is unconscious, self-awareness alone does not correct it; people who know about confirmation bias exhibit it just as strongly.
What works is structure. Deliberately seek the evidence that would contradict the view: what would I expect to see if I were wrong, and have I looked?
Assign a devil's advocate or a red team whose job is to make the opposite case. Pre-commit to decision criteria: before the data arrives, write down what result would change the decision, so that the data is judged against a standard set when the mind was open.
Use base rates: how often do projects like this succeed, how often do companies with these numbers recover, regardless of what this one's story says. Separate the person who made the forecast from the person who reviews it.
Keep a decision journal and review it: which beliefs were held, what evidence was weighed, what happened, and where the pattern of discounting appeared. Reward changing one's mind when the evidence warrants it, rather than punishing it as inconsistency.
For finance functions, whose job is to be the organisation's source of objective information, the bias is an occupational hazard and its management a professional duty. A finance team that confirms what the business wants to hear has abandoned its purpose; one that systematically tests it is doing its job.
In practice
Real-world examples.
Example
An equity analyst maintains a buy recommendation through four quarters of missed guidance, describing each miss as one-off, until the company issues a profit warning.
Example
A bank's credit committee approves a loan to a well-liked customer despite a falling interest cover ratio that the relationship manager characterises as temporary.
Example
A retailer's board interprets falling like-for-like sales as weather and competitor promotions for two years before accepting that its format has dated.
Think of it
“Confirmation bias is seeing what you want to see-finding evidence for what you already believe.
Formula
Calculation
Confirmation bias has no formula, but its effect can be measured and countered:
Forecast bias = Sum of (Forecast minus Actual) / Sum of Actual, tracked by forecaster; persistent one-sided error indicates biased information processing
Pre-mortem test: before deciding, list the reasons the decision might fail and the evidence that would show it; check that evidence has been sought
Base rate check: Expected outcome = Base rate for comparable cases, adjusted only by evidence specific to this case that would survive scrutiny by a sceptic
Worked example. A company's investment committee reviews a proposal from its operations director to build a new plant, business case NPV $8,000,000 at a 9% cost of capital, based on a forecast that the new capacity will be 85% utilised within two years.
The finance director applies a structured challenge:
- Base rate: the company's last six capacity expansions reached their forecast utilisation on average 18 months late and at 70% of the forecast level in year two. Applying the base rate to this project: utilisation 60% in year two rather than 85%, reaching 85% in year four. NPV falls to $1,500,000.
- Disconfirming evidence sought: the proposal cites three customers who have "indicated" they would take the additional capacity. The finance director asks for written commitments. One customer confirms a contract; one declines to commit; one has already placed the volume with a competitor. The forecast's customer base is a third of what the proposal assumed from those sources.
- Pre-committed criteria: the committee had agreed, before seeing the proposal, that plant investments must show positive NPV at 60% of forecast volume. At 60%, this proposal's NPV is minus $2,000,000.
- Pre-mortem: asked to imagine the project has failed in three years and explain why, the team lists: a competitor's expansion (already announced, and not in the proposal), a customer concentration risk (the confirmed customer is 40% of the forecast), and the historical pattern of slow ramp-up.
The operations director's response is that the base rate reflects past management, that the declining customer will come round, and that the competitor's expansion is in a different segment. Each point may be true; each is also exactly what confirmation bias would produce. The committee defers the full plant, approves a $3,000,000 first phase sized to the confirmed contract, and sets a trigger for the second phase: signed commitments covering 70% of its capacity. Two years later, the first phase is at 90% utilisation, the second phase has been triggered by two new contracts, and the total investment has produced a positive NPV that the original single-stage proposal, built on the unconfirmed customers, would have missed by two years of under-utilisation.
The forecast bias record: the operations director's last five volume forecasts averaged 28% above actual. The finance director's proposals now show each sponsor's forecast bias alongside the forecast, which has reduced the average optimism to 12% in a year, not because sponsors have become less biased but because they know the number will be shown.Case study
Seen in the real world.
A technology company's founder believed that its main product would succeed in the American market, and the company invested $15,000,000 over three years in a US sales operation. Each quarter's results were read through that belief: slow sales were the sales team's fault, then the pricing, then the timing; a competitor's success proved the market existed; a large lost deal was "close". The board received the same interpretation because the founder presented the results.
A new non-executive director, an experienced operator, asked for three things: the base rate of European software companies entering the US (most failed to reach profitability within five years), the specific evidence from lost deals (customers cited two product gaps the company had dismissed as unimportant), and a written statement from the founder of what results in the next two quarters would cause him to withdraw. The founder, to his credit, wrote down a threshold. The next two quarters missed it.
The company withdrew, wrote off $15,000,000, and refocused on its home market, where it doubled in three years. The founder later said that the pre-commitment had been the only thing that could have changed his mind, because every quarter's data on its own had been explainable.
Watch out
Common mistakes.
- Believing that awareness of confirmation bias protects against it. It does not; only process does.
- Gathering more evidence without changing how it is evaluated, which produces a larger body of evidence filtered the same way.
- Treating a change of mind as weakness. A decision process that punishes revision guarantees that biased beliefs persist.
Questions
People also ask.
How do I know if confirmation bias is affecting a decision?
Ask what evidence would change the decision and whether it has been sought; check the forecaster's track record for one-sided error; and have someone independent argue the opposite case.
What is the difference between confirmation bias and optimism bias?
Optimism bias is the tendency to expect better outcomes than the base rate justifies. Confirmation bias is the tendency to process evidence in favour of an existing belief, whatever its direction. They often combine in business cases.
What is the single most effective countermeasure?
Pre-committing to the criteria for a decision, and to what evidence would reverse it, before the evidence arrives.
From the founder's library
Take it further with the book.
Build your financial confidence beyond this definition. Shihan's full-length guide, Accounting Fundamentals, takes the same plain-English approach and turns it into a complete, practical playbook for non-finance managers, business owners and students - with chapter-end quiz answers and presentation slides included.
25% off with code MMHQ25, applied at checkout. Priced in USD - checkout may show the equivalent in your local currency.
View the book and save 25%