Stratify with purpose
Define the population and meaningful groups first. Set quotas or allocation rules against the question, then document where coverage may be incomplete.
Consumer research / Chicago, ILBetter questions.
Better questions.
Clearer signals.
Stratified digital consumer panels for understanding what people notice, choose, and value. We bring a methodological lens to e-commerce, social audiences, and the decisions behind them.
Audience crosstabILLUSTRATIVE
A simple view of response by audience segment
Example values for display only—not panel findings.
01 / Research planning toolPlan for the people
Plan for the people
you need to reach.
Estimate the invitations needed to reach a net sample after eligibility and attention-check attrition.
Planning estimate
Gross invitations
—
—Net completed responses
—Net margin of error
Assumes each invitation has the stated chance of producing an eligible response, then the stated failure rate applies. If incidence excludes nonresponse, include an expected response rate in it before using this tool.
This is a simple random-sample approximation for the overall population. Stratification, weighting, clustering, subgroup targets, nonresponse and other bias can change precision; calculate each stratum separately and account for design effects in a full plan.
02 / MethodologyFrom panel design
From panel design
to defensible insight.
Good research makes the path from recruitment to interpretation visible.
Review response quality
Use attention checks, response patterns and appropriate bot screening with clear, consistent exclusion rules. Inspect borderline cases before removal.
Report the context
Show sample sizes, question wording, field dates and uncertainty. Describe weighting and design effects before comparing subgroups.
03 / Quality ledgerWhat belongs in a
What belongs in a
panel readout.
A framework for evaluating evidence, not a claim about a specific completed study.
| Quality dimension | Measure to report | Why it matters |
|---|---|---|
| Coverage | Eligible n / stratum | Shows who is represented |
| Attention | Failures / reviewed n | Makes exclusions visible |
| Consistency | Cronbach’s α, if applicable | Checks multi-item scale reliability |
| Precision | ±% and confidence level | Frames sampling uncertainty |
- Sample sizes
- n= reported
- Quality rules
- documented
- Uncertainty
- contextualized
Cronbach’s alpha is appropriate for certain multi-item scales; it is not a general measure of panel quality.
A note from methodology
Director of Survey Methodology
“A useful number comes with its denominator, its question, and an honest account of who was able to answer.”Nadia Kowalski
Director of Survey Methodology
04 / Research notesQuestions worth
Questions worth
looking into.
Practical guides to audience behavior, buying decisions, and research design.
Instagram discovery: measuring the path from view to follow ↗
A framework for separating discovery, attention, profile interest, and follow-through.
How shoppers respond to online prices and vouchers ↗
A practical panel study design for discount behavior, cart friction, and price sensitivity.
Public profiles, private circles: studying audience boundaries ↗
How public figures and teams can think about separate audiences without assuming perfect privacy.
A measurement plan for marketing a business ↗
A survey-led guide to recall, advocacy, pricing, and creative decisions.
Digital board games: what a player panel can reveal ↗
A study framework for session length, asynchronous play, and virtual tabletop adoption.
How small businesses evaluate referral partners ↗
A survey framework for contractor referrals, professional services, and local trust.