Back to Blog & Insights

Insight

Get Independent Estimates Before the Group Discussion

When the first estimate comes from the most senior person, teams may adjust around it instead of sharing what they know. Collect numbers and confidence independently, then use the discussion to understand differences.

By WWA Editorial Team 3 min read
A business team reviewing separately prepared demand forecasts for a specialty manufacturing order

A senior leader says the seasonal order will reach 40,000 units. Within minutes, sales sees 42,000, operations suggests 38,000, and finance calls 40,000 reasonable. The apparent agreement may be comforting, but it does not reveal what each person believed before hearing the first number. The opening estimate has become the room’s anchor.

That is a practical problem for teams estimating demand, project duration, or cost. Discussion can spread useful information, but it can also make errors more alike and give persuasive or senior voices disproportionate influence. Knowledge at Wharton describes research in which independent estimates were collected before discussion; the effectiveness of the subsequent conversation was connected to how well participants’ confidence reflected their accuracy.

Preserve the information before comparing it

Before anyone presents a view, ask each relevant person to submit an estimate privately, along with a plain-language confidence judgment. For example: How likely is this estimate to fall within an acceptable range? What assumptions would change it materially? The purpose is not secrecy for its own sake. It is to preserve distinct information before social influence changes the numbers.

Consider a hypothetical specialty manufacturer preparing a limited production run of decorative hardware for a seasonal retail order. Sales expects 36,000 units based on customer conversations. Operations estimates 29,000 because a finishing step may constrain output. Finance submits 32,000 after reviewing comparable orders, while the account manager estimates 44,000 but reports low confidence because the retailer has not confirmed store participation.

If the managing director had opened with 40,000, some of those differences might never have surfaced. Collected independently, the estimates reveal more than a possible total. They expose uncertainty about customer demand, production capacity, and the reliability of early account signals.

Treat the spread as business information

The next decision is not simply whether to average the four numbers. First, examine why they differ. Are people estimating the same outcome and time period? Does one person have task-specific information? Is a high or low estimate driven by a constraint others missed? Is confidence based on repeated experience, or merely expressed strongly?

Research reported in Scientific Reports in 2024 found that independently eliciting two kinds of estimates produced better-quality aggregated estimates than eliciting them jointly in the experiment studied. A 2022 Scientific Reports reanalysis also found that the results of social influence depended on both how influence was centralized and how initial estimates were distributed. No influence structure performed best in every context. These findings support a careful process, not a universal claim that independence or averaging will always deliver the right answer.

For the manufacturer, the team might learn that operations is estimating feasible output while sales is estimating retailer demand. Those numbers should not be blended as though they answer an identical question. The practical choice may be to separate the demand forecast from the capacity forecast, resolve the finishing constraint, and then select a production commitment that reflects both.

Use confidence, context, and outcomes

Confidence should inform the conversation without becoming a substitute for evidence. A confident contributor may possess relevant experience, or may simply communicate with certainty. Ask: What would have to be true for this estimate to hold? Which assumption explains most of the spread? Who has made similar estimates before? What evidence could reduce uncertainty before the commitment is final?

Historical accuracy matters as well. Harvard Business School Working Knowledge recommends drawing probability forecasts from diverse sources and checking forecasts against actual outcomes. Teams can retain each original estimate, confidence assessment, assumptions, final decision, and result. Over time, that record shows whose judgment is useful for particular tasks and whether confidence is calibrated.

The central choice is straightforward: capture individual judgments before the meeting shapes them, then discuss the differences rather than rushing toward consensus. Combine estimates only after understanding what they represent, what each contributor knows, and how similar judgments performed previously. The goal is not quieter meetings. It is a better-informed commitment.

Sources