Audit a research summary for evidence cherry picking

Inspect qualitative and quantitative research decks to detect selective reporting, omitted counter-evidence, and unvalidated generalizations.

Prompt

Prompt
Act as an objective principal research auditor. Analyze the research summary pasted below for evidence bias, selective quoting, and unsubstantiated conclusions. Identify where positive findings are highlighted while negative or conflicting feedback is minimized, buried, or absent. For each identified instance of bias, point out the specific claim, explain why the evidence appears weak or skewed, and formulate the exact challenge question stakeholders should ask. Output format: A markdown table with columns Claim Made, Suspected Bias or Gap, and Stakeholder Challenge Question. Follow the table with a section titled Missing Context Checklist listing 3 to 4 specific data points needed to validate the findings before acting. Here is the research text:

Example output

Example for: an internal product summary advocating for an immediate mobile redesign

Claim MadeSuspected Bias or GapStakeholder Challenge Question
"Users unanimously praised the new navigation layout during prototype walkthroughs."Absolute language ("unanimously") based on an unstated sample size; likely omits hesitation or task failure rates.What was the exact completion rate for key tasks on the new layout, and how many total participants were tested?
"Customer feedback indicates the current desktop experience is obsolete and needs to be de-prioritized."Selective synthesis using dramatic qualitative quotes to justify cutting engineering investments without usage metrics.What percentage of total monthly recurring revenue and daily active sessions currently originates from desktop users?
"Early beta testing shows an immediate 40% lift in feature discovery among new accounts."Survivorship and novelty bias; measures day-one exploratory clicks without reporting 30-day retention or cohort drop-off.Did the 40% discovery lift persist into week two, and was this tested against a matched control group?
Missing Context Checklist
  • Sample Size and Composition: Breakdown of participant job titles, account tiers, and whether they were existing power users or net-new prospects.
  • Negative Sentiment Distribution: The complete list of reported usability friction points and negative sentiment tags, not just the selected quotes.
  • Baseline Comparison Metrics: Benchmark performance metrics comparing the existing production interface against the prototype using identical task goals.
Written by AI from this prompt. Your results will vary.

Paste your executive summary, presentation narrative, or research readout into Claude, ChatGPT, or Gemini after the prompt. Follow up by asking the model to rewrite the findings into balanced, risk-aware prose.

Make it repeatable

Keep using this prompt without pasting it in every time.

Keep exploring

More in Research