Sources checked 31 October 2026. The acquisition, documents, and figures below are hypothetical. The review exercise is a practical aid, not a validated measure of bias.
The acquisition presentation is nearly finished. Your distribution company has spent four months pursuing a regional competitor. Three customers praise its delivery service. The sales director sees a route into a market your own team has struggled to enter. In the appendix, warehouse staff describe orders that only the founder knows how to rescue. The deal team calls these manageable exceptions.
You chair the investment committee. The customers may be right, and the exceptions may be manageable. But the presentation gives you little basis for deciding whether the target's service will survive a change of ownership. It shows who likes the business. It does not yet explain what makes the business work.
Before discussing the price again, ask the team to write the claim that carries the acquisition case: “We can retain this delivery performance after the founder leaves.” That sentence gives the next round of diligence a job.
Give the preferred case a fair test
Raymond Nickerson's 1998 review describes confirmation bias as seeking or interpreting evidence in ways that favor a belief already held. It covers several forms of selective treatment, rather than one universal mechanism. The review is a basis for examining how a team handles evidence; it cannot diagnose this committee from a presentation. [1]
In this case, inspect the interview list. Did the seller choose all three customers? Did the team ask for examples of failed deliveries as well as successful ones? When a warehouse manager raised a problem, did someone check the underlying orders, or dismiss the account because the manager opposes the sale?
Those questions can expose uneven scrutiny. They can also reveal sensible choices. A recent operational record may deserve more weight than a secondhand complaint. A committee member can reject your interpretation without being biased. Record the reason for accepting or discounting each finding, then apply that reason to evidence on either side.
Choose a test that separates explanations
Peter Wason's 1960 experiment asked 29 psychology undergraduates to discover a rule behind number triples, starting with 2, 4, 6. Only six stated the correct rule at their first announcement. The task was deliberately structured so that repeatedly testing examples fitting a narrower guessed rule could miss the broader rule. This was a small laboratory task, not a study of acquisition teams. [2]
The practical question is what another favorable interview would establish. Both a transferable operating system and a founder who repeatedly fixes problems could produce satisfied customers. Praise alone does little to separate those explanations.
Joshua Klayman and Young-Won Ha's 1987 analysis adds a caution: testing cases where you expect a property to appear can be informative. This “positive test strategy” is not automatically confirmation bias. Its usefulness depends on the task and which outcomes the test could reveal. [3]
For the acquisition, consider two working explanations. Under A, documented routines and staff capability sustain the service. Under B, founder intervention sustains it. They may both contribute, and other explanations remain possible. Still, comparing them suggests a better inquiry: examine orders during the founder's absence, including the difficult orders, and check who intervened remotely.
Agree on the interpretation before seeing the records. Performance holding up without hidden intervention would count toward transferability. Repeated failures resolved only by founder calls would challenge it. A quiet week with easy orders would leave much unanswered. The team should also specify how incomplete logs would limit either conclusion.
One customer statement, two explanations
“They always sort out our urgent orders” fits both a repeatable process and repeated founder intervention. Compare records that could separate these explanations, then check how those records were produced.
Hypothetical acquisition evidence. These explanations can overlap; the diagram does not assign probabilities.
Keep the source attached to the finding
Imagine the data room contains a delivery dashboard, a customer interview, an exception log, and a report from a week when the founder was away. Each can answer a different question. The dashboard describes recorded performance. The interview describes one customer's experience. Neither automatically explains how that performance was achieved.
Suppose the exception log contains twelve founder interventions in one month. Twelve could be operationally serious or routine. You need the number and type of orders, what counted as an intervention, and whether staff recorded every case. A photograph of a spreadsheet without dates or a named owner is a lead to investigate, even if its contents support your concern.
Use the same discipline on encouraging evidence. An independently checked dispatch record can be reliable within its scope while saying nothing about returns excluded from the file. “Independent” does not mean complete. Two reports copied from that record provide one underlying observation, not two independent confirmations.
The exercise below lets you classify five fictional documents against the two explanations. You decide whether each source is usable for this claim and whether its finding is expected, unexpected, or unclear under each explanation. The comparison sheet keeps the source concerns visible. It does not convert documents into votes or estimate the probability that the acquisition will succeed.
Review the acquisition file
Open each document, inspect its scope, and compare both explanations. Change a source judgment or an expectation to see which findings still separate the explanations.
Decide what would change the decision
A review can become another ritual if nobody specifies what follows from it. For this hypothetical deal, the committee might require a supervised handover trial before approving the integration plan. If the trial exposes founder dependence, the response could be a longer transition, a different price, or withdrawal. Write those possibilities down before the next meeting.
Someone should seek the strongest available evidence against the preferred case. Give that person access to the same records and enough time to check them. Also ask what would weaken the alternative explanation. Otherwise the exercise can become a contest in which the most alarming account wins.
These steps do not guarantee impartial judgment. They make the questions, exclusions, and changes of mind inspectable. New evidence may justify revising a threshold; record why, rather than quietly moving it after an inconvenient result. Time and access may leave a material uncertainty unresolved, and the committee must decide whether it can accept that uncertainty.
At the next review, spend ten minutes on one page. Write the central claim, one plausible alternative, and two observations that would weaken the preferred explanation. Beside each observation, name the record needed, its owner, and the decision it could change. Assign the first check before anyone adds another favorable quotation to the presentation.
Sources
[1] Raymond S. Nickerson, “Confirmation Bias: A Ubiquitous Phenomenon in Many Guises” (1998), Review of General Psychology, 2(2), 175–220. Review; supports the definition and the distinction between evidence search and interpretation.
[2] P. C. Wason, “On the failure to eliminate hypotheses in a conceptual task” (1960), Quarterly Journal of Experimental Psychology, 12(3), 129–140. Original laboratory experiment; its task structure limits direct workplace generalization.
[3] Joshua Klayman and Young-Won Ha, “Confirmation, Disconfirmation, and Information in Hypothesis Testing” (1987), Psychological Review, 94(2), 211–228. Theoretical analysis and review; explains why a positive test can be useful or misleading depending on the task.
One email when we publish. Research, product decisions, and what teams report back.
