Volume is not evidence

In most product reviews I have sat in, the direction gets set by whoever speaks most and speaks with least hesitation. A senior engineer, a sales lead with a live deal, an executive with a strong instinct. None of them are acting badly. They are confident and they are present, and those two things together move decisions.

Meanwhile the person who has spent three months in the support queue says nothing, because their observation is anecdotal and the room is moving fast.

That dynamic costs you products.

The organisational research

Work on employee silence documents this: information, concerns, and alternatives stay unspoken when the environment does not invite them. Research on group decision-making describes a spiral of silence, where people read the dominant opinion as consensus because the dissent stayed private.

Both effects concentrate in exactly the meetings where product bets get made. Small group, senior participants, time pressure, a decision expected by the end.

Practices that change it

Five things, all cheap:

Collect written input before discussion. Ask for positions in writing beforehand. Written input escapes the ordering effect where the first confident statement anchors everyone.

Separate the idea from the title. Present competing options without attribution where the format allows. The proposal stands or falls on the argument.

Require the data behind the assertion. When someone says customers want this, ask which customers, how many, and how you know. Not to challenge them, to convert a claim into something the group can evaluate.

Ask for dissent by name. General invitations to disagree produce silence. Asking a specific person what would make this the wrong call produces answers.

Make disagreement survivable. People stay quiet when they have watched dissent get punished. This one takes longer than a meeting to fix, and nothing on the list works without it.

The standard

Our job is to find the best evidence available, not the strongest-held opinion in the room. Every relevant voice gets heard, and no voice gets weight from its title alone.

I have watched features get built because someone influential wanted them while the data said otherwise. In each case the data was in the building. Nobody had made room for it.