Standards Framework
HYMBS standards are not certifications — no vendor pays to achieve them. They are evaluation criteria that buyers can apply independently when assessing research proposals and suppliers.
How to Use This Framework
These standards are designed as RFP criteria and post-fieldwork audit checklists. They can be applied to any B2B research vendor without prior relationship. No licensing, registration, or fee is required. Copy, adapt, and apply them in your procurement process.
Five Domains
Vendors must provide a written IR estimate with confidence interval before fieldwork. If actual IR falls below estimate by more than 20%, the buyer should not bear additional cost without prior agreement.
No more than 40% of any delivered sample should have completed a survey in the same category within the prior 90 days. Ask vendors for their panel rotation policy in writing.
Vendors must disclose whether sample is sourced from owned panel, exchange/aggregator, or river sampling. Each source type carries distinct quality profiles; mixed-source samples require split reporting.
Screener questions should not be shared with panellists in advance. Panel members who fail screeners repeatedly should be profiled — systematic screener failure patterns indicate fraudulent profiling.
All delivered data must have been processed through at minimum: speeder detection (LOI <33% of designed), straight-lining detection, attention check failure review, and duplicate IP/device check.
Vendors must provide a QC report alongside the data file. The report should state: number of records removed per QC criterion, total removal rate, and the thresholds applied for each check.
Where verbatim responses are collected, a sample must be reviewed for AI-generated content, copy-paste duplication, and gibberish responses. These represent quality signal, not edge cases.
Vendor replacement of removed completes must use the same screener and source conditions as original fieldwork. "Top-up" respondents from an easier-to-reach source compromise comparability.
All reported metrics (completion rate, IRR, data quality rate) must be defined in the proposal stage, before fieldwork begins. Post-hoc definition adjustment invalidates comparability.
Vendors should provide a log of fieldwork decisions made during the field period: quota adjustments, source additions, screener modifications. Each change affects the integrity of the final dataset.
For business samples, vendors should document the verification method used to confirm respondent role and industry. Self-reported panel profiles carry systematic over-reporting of seniority.
Where multiple panels are blended, buyers have the right to request source-level completion and quality breakdowns. Blended panels mask source-specific quality variance.
Every delivered dataset should include: field start and end dates, total invitations sent, total starts, total completes (pre-QC), total QC removals, and final delivered n.
All reported findings should be accompanied by confidence intervals appropriate to sample size and design effect. "N=200" without a design effect estimate understates margin of error for complex samples.
"A standard only functions as a standard when it is applied consistently, by a party with no financial interest in the outcome. HYMBS does not certify vendors. We publish criteria — buyers make the judgement."
These standards may be cited in RFP documents, vendor scorecards, or contract appendices. HYMBS does not endorse any vendor's claim to meet these standards. Buyer verification is required in all cases.
Apply These Standards
Contact HYMBS to discuss benchmark-based vendor evaluation, or to submit a standards revision proposal based on observed field conditions.