What AI-generated answers actually look like in an open-end
We read 120,000 rejected open-ends line by line and found six patterns you can automate — plus two you should never use to reject a real person.
Resources
What we have written down
about sample and data.
No listicles. Most of this started as an internal debrief — the mistake we made, the invoice we mis-forecast.
Insights
What we have been
thinking about
AllMethodCase studyIndustryPractical guide
Method
Case study
A concept test in 72 hours: the schedule we actually used
Five packaging routes down to one. Which steps run in parallel, which cannot, and where the extra week usually hides.
Industry
2026 sample pricing: the cost is not the sample, it is the redo
Across 12,400 projects, rework, delay and invalid completes account for 31% of total budget spend.
Method
Stop answering every sample size question with "300 is fine"
Sample size follows from the number of cross-tabs, the effect size and your tolerance for error. Here is the lookup table.
Product
A quota cell is not filling. Four options, four different costs
Relax the screener, swap the source, add recruitment, or admit the gap. A comparison of what each really costs you.
Industry
Can your research data legally leave the country?
Three ways to implement "data stays local, results travel", with a checklist you can hand to legal.
Customer stories
Three weeks to three days
takes more than a better tool
None of these were solved by software alone. The sequence changed.
CPG
Five packaging routes, decided in 72 hours
A 400-complete screen and an 800-complete validation sharing one quota structure.
3 daysBrief to final route
Technology
Finding out why users actually left
120 churned users recruited from the previous 30 days; survey plus 20 interviews.
41%Churn drivers rewritten
Financial services
Pricing a new investment product
Stratified by assets, PSM sensitivity crossed with acceptance, compliance-reviewed.
9 daysIncluding compliance review
Reports & whitepapers
Give us an email.
Take them with you.
All public, no questions about company size or budget. We just want one fewer deck claiming all sample quality is basically the same.
- Current 2026 editions, with methodology appendices and definition of every metric.
- Free to quote in internal meetings — just cite the source.
- One update per quarter after that, one-click unsubscribe.
2026 Sample Quality Report
Rejection structures across 48M completes, with real shares for AI answering, device farms and straightlining.
68 pages · PDF
Sample pricing and fielding benchmarks (H1 2026)
Median CPI and start time by audience type, city tier and market — built for budgeting.
24 pages · PDF
Cross-border research compliance handbook
Three implementation paths for localisation, de-identified delivery and transfer, with flowcharts and checklists.
32 pages · PDF
Start your first project
Brief it today. Work with usable data tomorrow.
$500 in sample credit on your first order. No minimums, no contracts, no platform seat fees. If you would rather have someone run it with you, one of our researchers will field the whole study.
No credit card 14-day refund window Export to SPSS / Excel / CSV