How to avoid letting the first number control your estimate

Reduce the influence of a first number by gathering estimates and assumptions before showing a preferred budget or forecast. Compare the evidence behind the differences, then build an appropriately qualified range. Independence is a preparation method; it does not guarantee that any estimate is accurate.

Anchoring bias is the influence of an initial reference value on a later estimate beyond what the relevant evidence warrants. Tversky and Kahneman’s 1974 paper describes adjustment from initial values and shows how estimates can remain biased toward those starting points. The operating process below applies that research to executive estimates without claiming the paper tested this exact procedure.

Find the first number in the room

Composite scene: You are reviewing a proposed budget at home. The departmental submissions cluster around a figure you mentioned casually during an early discussion. You cannot locate the evidence behind your original figure, and the later estimates cite one another rather than an independent basis.

Do not begin by accusing the team of copying you. Ask where each estimate came from and when the contributor first encountered the initial number. A common value may reflect shared evidence, a common constraint, or influence from the reference. Similarity alone does not establish the cause.

Locate the decision the estimate supports. A preliminary planning range, an authorized spending limit, and a forecast of actual cost are different quantities. If contributors treated them as interchangeable, clarify the task before trying to remove a bias.

Separate evidence from a reference point

A reference can be useful when it summarizes relevant evidence. A prior project with comparable scope may inform a new estimate. The problem is giving an initial figure more influence than its evidential relevance deserves.

Ask what makes the reference comparable. Identify differences in scope, timing, operating conditions, and information quality. Have qualified finance or technical owners assess those differences where the estimate has consequential implications.

Avoid treating a number as sound because it came from a senior executive. Equally, do not discard a well-supported reference merely because it arrived first. The task is to examine its basis and limitations.

Tversky and Kahneman discuss judgment under uncertainty, including anchoring and adjustment. Their paper does not establish that every forecast near an initial number is biased or that replacing it with a more distant estimate improves accuracy.

The independent-estimate method

1. Define the quantity and scope

State exactly what contributors are estimating. Specify the work included, the relevant time period, and known conditions. Give everyone the same confirmed information needed for that task.

For a hypothetical implementation, distinguish total project cost from the next authorized commitment. Explain whether the estimate includes internal capacity, transition work, and ongoing obligations. Ask the appropriate owners to determine which categories belong in the actual estimate rather than inventing accounting treatment in the workshop.

2. Collect separate estimates before revealing a preference

Ask each relevant contributor for an estimate, an appropriately supported range, and the assumptions driving it. Do this before displaying a proposed executive answer or a colleague’s submission where practical.

Record what information the contributors already saw. You cannot restore independence by saying “ignore the first number” after everyone has used it. If an important estimate has been exposed to a weak reference, consider an authorized fresh review by someone with suitable expertise and access to the relevant evidence.

3. Compare reasoning before combining values

Group differences by their cause. One estimate may include work another excludes. One may rely on a current supplier quotation while another relies on an old project. One may assume a resource becomes available that management has not approved.

Resolve definition differences before interpreting numerical disagreement. Do not average unlike quantities and call the result balanced. Where expert judgments remain different, preserve the reason and obtain qualified analytical support as needed.

4. Test the original reference explicitly

Identify its source, purpose, and date. Ask what evidence would justify retaining it and what evidence would require a change. Check whether the initial value was a target, a negotiation proposal, a constraint, or a genuine estimate.

If it was only a planning placeholder, label it accordingly in the record. If it reflects a real obligation or capacity limit, preserve that consequence. Avoid replacing one unsupported center with another merely to demonstrate independence from the CEO.

5. Approve an estimate with visible uncertainty

The authorized owner should state the estimate’s purpose, basis, assumptions, and limits. Identify the evidence that will update it. A range should reflect defensible judgment or analysis rather than an arbitrary spread around a preferred answer.

For material financial or technical use, have qualified owners review the method and inputs. Preserve the original versions so later changes can be traced to new information instead of remembered as what management always expected.

Use an assumption comparison that reveals the disagreement

A working table can show contributor, quantity estimated, evidence used, scope exclusions, major dependency, range, and information still missing. Restrict access where the inputs include confidential commercial information.

For the hypothetical implementation, one contributor may assume an internal team supplies a critical capability. Another may price external support because that team is committed elsewhere. The most useful discussion is whether the capacity assumption holds, not which estimate demonstrates greater optimism.

If the assumption cannot be verified before a decision is required, say so. Ask what limited commitment is permissible under the actual authority and risk arrangements. The table helps identify the question; it does not authorize spending or establish a financial recommendation.

Handle a first number that comes from a counterpart

An initial supplier proposal can become a reference for an internal estimate. Separate the quoted terms from your own requirements and alternatives. Ask the appropriate commercial and technical owners to examine scope before treating movement from the opening proposal as value gained.

A proposal below an earlier quote may still be unsuitable. A higher proposal may include different obligations. Compare the whole arrangement using authorized evidence and qualified review. Do not infer a reservation point or market value merely from the sequence of offers.

For negotiation-specific preparation, use the first-offer decision guide. Keep the present estimate review focused on what management needs to know about the work and its consequences.

Make a challenge possible without turning it into disloyalty

Edmondson’s 1999 study examines interpersonal risk within work teams. Applied to an estimate review, it gives you a reason to inspect how management responds when someone challenges the CEO’s starting figure.

Use proposed wording such as:

The initial figure is a reference we need to test, not an answer you must defend. Show your estimate, its scope, and the assumption driving the difference. If the current evidence points elsewhere, explain that basis. We will distinguish the approved limit from the estimate of what the work requires.

The study does not prove this wording removes anchoring. Use it to establish the process honestly, then examine the evidence contributors actually supply. A person may still have seen the same reference or shared the same incomplete data.

Keep the estimate history intact

Fischhoff’s 1975 experiments examine how outcome knowledge changes judgments about what could have been known earlier. That research supports preserving the information available when an estimate was made, while keeping the proposed recordkeeping application distinct from the experimental finding.

When an estimate changes, record the new information and its date. Do not overwrite the old assumption and later judge the original contributor as though the revised evidence had always been available.

Industrial/organizational psychology, the psychology of work and organizations, helps connect individual judgment with the group process used to produce a forecast. It does not supply a universal correction factor for an estimate that may have been anchored.

Ask what would move the estimate

Before approving the summary, ask each contributor which uncertain input has the greatest influence on their answer. Have the appropriate analyst examine a defensible alternative for that input. The purpose is to show the dependency, not to manufacture a precise probability for an unknown condition.

For the implementation example, a delivery estimate may depend on when the internal specialist becomes available. Show how the plan changes if that availability is later than assumed. The authorized decision owner can then examine the actual dependency instead of debating whether the central estimate feels conservative enough.

Review accuracy without inventing certainty

When the outcome becomes observable, compare it with the estimate and its assumptions. Ask what was learned after the original decision, which dependency changed, and which relevant information could reasonably have been obtained earlier.

A forecast that happened to be close is not proof that its evidence was sound. A forecast that missed is not proof of anchoring. Inspect the process and information rather than attributing a cognitive mechanism from the outcome alone.

If similar estimates recur, ask qualified analysts how to review them consistently across an appropriate set. Avoid reporting a calibration rate or improvement percentage from an unsuitable or incomplete sample.

Correct a contaminated estimate without humiliating its author

If a contributor says they used your figure because they believed it was an instruction, clarify the instruction. Accept responsibility for presenting a preference ambiguously. Then ask what evidence would support a fresh estimate under the clarified scope.

If the figure was an authorized spending ceiling, do not confuse respecting the ceiling with estimating the underlying work. Management may need to change scope when a defensible estimate exceeds the approved commitment. That is an operating choice requiring the proper authority.

The useful result is an estimate whose reasoning can be examined without relying on the first number displayed. Keep its scope, assumptions, and update conditions visible so the next decision follows current evidence rather than the momentum of an early suggestion.