Menu

Product

How you competeConcept testsBox studiesThe workspace

Report

Example reportReading a verdictMethodShoppersThe appSign inRequest access

Method

The rules, in the same words as the report.

You set the bars. Dough fixes confidence at 95%. A study can fail. Here is exactly how — no soft landings.

See the example report Request access

Result rule

Four outcomes. One geometry.

A bar clears only when the whole 95% interval sits on the right side of the line you set. Under ten answers is not a result.

Cleared
Whole interval at or above the bar.
Not cleared
Whole interval below the bar.
Too close to call
Interval straddles the bar.
Not enough responses
Fewer than ten answers. Not a result.

The contract

You set the bars. Dough fixes the rest.

You

  • Which bars are on (head to head, liking, price)
  • Where each bar sits
  • The benchmark and the price anchor
  • Moving bars — until the first shopper starts

Dough

  • Confidence fixed at 95%
  • Wilson intervals on the counts that matter
  • Bars freeze after the first shopper
  • Open answers scrubbed of contact noise

You never set the confidence. That is intentional — so a later reader can trust what was asked.

Overall verdict

One missed bar sinks the study.

The overall verdict takes the hardest result among the bars you turned on. A bar you left off is not tested. “Cleared” means every bar you set has cleared.

SimulatedNot cleared

Keel Lemon Water

New label vs current pack · 120 New York shoppers

Head to head
Cleared · 65 of 100
Price · $2.49
Not cleared

Winning the pick is not enough. Price missed — the study does not clear.

Read the worked example →

What a bar measures

Three ways shoppers can clear — or not.

Head to head

Direct choice

Each shopper picks your design or the benchmark. Interval is Wilson. “Neither” is shown and is not a win. Timing-flagged choices are excluded.

Liking

Love it / like it

Share who answer “love it” or “like it,” asked before any battle. Against current pack it is the paired difference on the same shoppers.

Price

Willingness to pay

Stated willingness at or above your price. Never a sales forecast. Because the answer is a band, the share is a range — the bar clears only on the conservative count.

Frozen bars

Set the line before anyone starts. Then it locks.

  1. 01EditMove bars, turn tests on or off
  2. 02First shopperBars freeze. No quiet moves mid-field
  3. 03ReportYour bar next to Dough’s default — so a later reader sees what was asked

Two different objects

A standing is not a study. A study is not a standing.

Standing

Ordinary picks in the app. Appears when enough people have chosen. Pattern — not a commissioned verdict.

Study

A question you commission. Bars frozen. Deadline. Four verdicts — including not cleared.

Same New York shoppers. Same catalog. Different objects — on purpose.Who answers →

Open answers

Words, not a contact list.

Open answers are scrubbed of emails, links, handles, and phone numbers. Names are not detected. Treat the words as words.

Ready to set bars that can fail?

Tell us the brand and the question. A person reviews every request and writes back.

Request access Or email hello@godough.coAlready have access? Sign in