Cintの品質管理ポリシー

As the industry’s largest sample exchange, we believe that it is our responsibility to actively maintain and ensure quality for every transaction. These high standards impact not only the health of the Cint platform, but can also have a positive effect on the insights industry as a whole. This is why we have clearly defined what quality means to us. We track it, measure it, and ensure it by having the best people, products, and technology working on quality every day, in real-time. Here’s how we do it.

Our commitment to quality

Cint quality charter
We have formalised our commitment to quality as a top business priority, with a cross-functional team dedicated to furthering our processes, technology, partnerships and products when it comes to quality. Our Quality Charter clearly defines a comprehensive approach, diagnostic performance measurements and methods for improvement.

Compliance with industry bodies
The Cint platform and its products comply with ESOMAR, MRS, ARF, MRIA, AMA, AMSRO and Insights Association standards. Cint also complies with ISO 20252 and is certified in Australia.

Dedicated team members
At Cint, we take accountability for quality on our platform to heart. That’s why we have staff whose only job is to ensure quality through operational and methodological support. They are backed up by a steering committee of individuals from our commercial, technology and operations departments to provide guidance.

Processes and procedures
Our data-driven approach gives us a window into quality at any time during a project, for real-time investigations and post-project reporting. Cint’s product suite for ensuring quality is comprised of both proprietary and third-party solutions that verify overall platform health, working cohesively to create better outcomes.

Transparency and privacy
These are two principles that are at the centre of Cint’s technology. We take data privacy seriously, are fully compliant with the most stringent regulations, such as GDPR. We facilitate consumer permissions and control via Cint’s Permissions Engine. When it comes to transparency, our clients have full, direct access to supply source information so they can monitor and evaluate recruitment quality.

Cint’s five components of quality

Respondent Quality
We evaluate respondents across multiple metrics to determine that individuals are real, unique, engaged and representative. All Cint panellists are opted in and we use a blend of personal, behavioural and demographic data points to ensure quality. A mix of advanced solutions ensure that, regardless of the number of accounts, users attempt a specific survey only once.

Supplier Quality
We carefully review and vet all of our supply partners for sample and research quality. They must comply with industry standards and codes of practice set forth by ESOMAR, Insights Association and others, plus adhere to requirements such as ISO 20252 and ISO 26362. Our partners have defined and standardised incentive requirements, survey invitations and more to boost quality. We share data and techniques to show suppliers any problems we see and then root out trouble, together.

Buyer Quality
Cint seeks clients whose surveys provide a good, engaging experience for respondents. Because we have set formal KPIs surrounding quality, we can discover which studies are performing poorly against our own stringent metrics. Our systems evaluate these indicators in field, which we share with buyers to help them understand how performance is impacting supply receipt and quality. We will shut down studies that create a bad user experience.

Operational Quality
Are we executing well and delivering on our commitments to clients and partners? We never underestimate our own, first-hand role in overall quality, so we ascribe to the golden rule when it comes to our internal operations. We measure our performance on both subjective and objective measures, such as timeliness and internal/external satisfaction.

Product Quality
As a company that provides the technology backbone for the insights industry, we know our products must provide industry-leading and future-proof solutions. We seek to be ranked consistently as best-in-class both through external benchmarks in the marketplace, and ratings by our clients and suppliers.

Fighting fraud from all directions

Proprietary technology
Our cutting edge AI-drive fraud detection system uses PII and profiling data to detect patterns that show fraudulent anomalies, such as user “surges” involving specific demographics, IP ranges or completion time frames. Additional solutions: provide digital fingerprinting to prevent duplicates; analyse data points to detect inattentive or disengaged behaviours among panellists; and automate panel cleansing and reconciliation. Our secure APIs and unique survey URLs limit common sources of fraud.

Third-party solutions
Imperium (Relevant ID)
MaxMind (minFraud, VPN/Proxy Detection, Geo-IP Verification)
Firehol (IP database)
Apility (Email scoring)
Google (reCAPTCHA)
SmartyStreets (Address verification)

Traditional techniques
Mainstream and more commonly used fraud mitigation tools still have their place, and provide an extra layer of protection. We combine our proprietary technology solutions with artificial blocks that require human intervention, such as Captcha; use of methods that machines would detect but humans would not, commonly called Honeypots; and open-end questions to determine genuine answers.

Fraud and quality checks at multiple instances

Registration
We apply a combination of proprietary Cint Fraud Detection Services and industry-standard 3rd party solutions – including GEO IP check and CAPTCHA – to ensure data quality is kept to optimum standards.

Sampling and survey
Deduplication through proprietary Cint Unique Respondent solution as well as Imperium RelaventID and Maxmind Minfraud. (Address verification and exclusion from banned database).

Rewarding
Running Cint proprietary Cint Fraud Detection Service before panellists are rewarded.

Proprietary technology

Cint Unique Respondent
Proprietary deduping solution.

Cint Fraud Detection Service
Proprietary solution to score and detect fraudulent behaviour among panellist through analysing data points, behaviour and patterns including speeding and straight-lining.

Customer Feedback
Real-time and post-survey feedback from customer data collection to feed into automated panel cleansing and self-learning Cint Fraud Detection Service.

Banned Database
Database for banned panellists containing identifiers for future automatic blocking.

Automatic detection of anomalies
Systematic detection and removal of fraudulent panellists

Unique Survey URLs and Secure API transactions with 3rd parties

Cint supply quality

All Cint’s supply panels are carefully reviewed and vetted before entering a partnership to ensure they meet specific contractual requirements that are essential for sample and research quality. 

These include:

  • Compliance with ISO 20252 and ISO 26362 quality requirements including panel operation, maintenance and hygiene requirements.
  • Compliance with all applicable industry codes of practice, including but not limited to ESOMAR and Insight Association.
  • Exclusion (removal of specific supply / remove specific panellists in case of issues)
  • Defined/standardised incentive requirements
  • Defined/standardised survey invitations

Quality and anti-fraud measures taken by Customer and Partner Success teams

Partner Success
Cint has a dedicated Partner Success team that works with suppliers to assure the highest level of quality from suppliers in the Cint Insights Exchange.

Partner Cooperation
Continuous follow up with suppliers which are over-represented in bad quality or fraud activity reports. Share of best practice in recruitment and fraud detection where suppliers can be excluded or put in quarantine until issues are resolved.

Partner Monitoring
Cint’s Partner Success teams monitor the performance of supply sources daily. Conduct further investigations at both panellists and panel level to determine current settings and security processes and any trends and performance patters. A panel can be put on hold until any data quality issues are resolved, or to choose permanently to remove them from the Cint Insights Exchange.

Client Cooperation
Cint does not host any surveys. We work diligently with our clients to ensure surveys get deployed after the appropriate engagement and validation checks. These include analysis of questionnaire completion time, data outliers, unanswered questions, straight-lining, red-herring questions, and report any quality concerns in real-time to Cint to learn and improve Cints automatic detection services.

Other quality and standards

Survey control checklists

  • correct implementation of re-directs
  • appropriate logic in the survey
  • correct set up of quotas
  • whether or not there is any personal, sensitive or illegal questions being asked
  • whether or not panellists can choose a ‘prefer not to answer’ option with any income or ethnicity questions
  • overall general quality before proceeding with fieldwork.

If any of these points are not in compliance with Cint’s quality standards, the survey is not launched.

Panellist rating by activity
All panellists are scored by their level of survey activity. If a score drops to a certain agreed level, panel owners can use this scoring system to automatically clean their panels.

Automatic housekeeping
Panellists are automatically removed from panels in Cint platform when their email address is proven invalid.

Random and stratified sampling
Within the required targets, sample is randomly generated as well as being stratified by high, medium and low responders.

Frequency of invitations to surveys
All panels in the Cint platform are given the option of deciding how often they want their panellists to receive survey invitations. The default number of days set is 10, however, panel owners or panellists can change this to be lower or higher.

Exclusion management
When sending sample to studies, panellists can be excluded on projects and specific statuses.

Refusal of panel abusers
Unique identification data is required in the validation process as well as to redeem incentives such as: ID number, home address and bank details.

Panel blending
Sample can be drawn from multiple panels simultaneously to reach hard-to-find target groups and eliminate source bias. This allows users to benefit from selecting sample generated by different recruitment methods, from CATI recruited panels to panels built from natural online communities, where members have a relationship with the panel owner’s brand.

Panellist survey rating
Panellists have the ability to rate each survey on length, language and logic as well as other errors in surveys. This feedback can help buyers to improve the quality of their surveys.

All surveys run in our system require a soft launch to obtain 10% of the overall sample size. This serves as an extra check to ensure any errors are detected before fully launching.

Maximum limits on survey lengths
Cint believes that the length of survey has a reciprocal relationship with the quality of data. Therefore, a longer survey may illicit lower quality results. Having stated this, Cint does not allow surveys over 35 minutes in Cint platform without the written approval from independent panel owners.

Performance and security
We strive to always produce the most advanced technology in terms of performance and security. If you have specific questions, please direct them to learnmore@cint.com. We would love to be of service!

Panellist quality control process
The Cint Platform takes special care to check every link for topic, content, language, length and logic before the initial launch. In addition to this, we have some of the strictest panel quarantine procedures in the industry. Our security measures and quality assurance efforts eradicate the common problems associated with sampling.

See our Platform Features to learn more about the technology behind these quality processes.

  • At registration
    Cint hosted registration pages can make use of Geo-IP verification technology to ensure that the registrant is located in the panel’s stated country. Captchas can be employed to stop automated (non-human) registrations.
  • When sampling
    When selecting panellists to invite to a project (sampling), panellists cannot be invited more than once to the same survey, and any other panellist using the same email address for more than one panellist account is also excluded from sampling. There is also the option to exclude panellists that have previously participated in other projects.
  • At survey start
    Cint tracks the status of each respondent, i.e. unique panellist invited to a specific project. Respondents can only start a survey if their status indicates that they have not previously responded or, if they have responded, have not completed or have been ‘terminated’ for some reason. Further, Cint relies on two de-duping technologies, ‘Cint Unique Respondent’ and ‘Relevant-Id’ which can determine if a respondent has started the survey before. Cint also has the ability to apply Geo-IP validation on survey start to ensure the respondent is located in the panel’s stated country when responding.
  • In survey
    Cint provides end-points (redirect URLs) that the client can use for terminating respondents that don’t provide quality responses (e.g. straight-liners).
  • At survey finish
    Upon reaching a survey end-point (Cint redirect URL), responses are checked for duplicate completes while ‘Speeders’ and fraudsters are automatically detected and tracked (as suspicious).
  • At reconciliation/project closing
    As each respondent is associated with a unique identifier it is possible to compare IDs between the Cint platform and the client. Any suspicious responses can then be investigated and, when required, excluded from the complete count to be marked as a bad response in Cint’s records.
  • Panel care
    Cint uses respondent records to regularly identify and remove/disqualify fraudulent panellists from the platform