ESOMAR 28 QUESTIONS

TGM proprietary panels are available in more than 85 countries, and we continuously expand our global coverage, reach and profiling. People join voluntarily and agree to participate in our market research projects. We have a panel management team applying strict panel rules to ensure the best panelist experience through their panel life cycle (from panel recruitment to panel health/communication and quality checks). We also have a support team to help members with their questions/requests.

We use several methods to invite panelists to join the TGM panel: this included

• opt-in email,
• organic traffic,
• referrals,
• mobile apps,
• marketing campaigns,
• social media
• as well as external affiliate networks.

TGM is following strict quality standards defined by ICC/ESOMAR Code on Market and Social Research.

We believe that reaching people on mobile is crucial for research; it gives much better representativity when every day becomes more mobile and allows us to connect with people at the moment. Thus, as an addition to our TGM panels, we use a global mobile network of publishers (access to more than 30,000 different mobile apps globally) like mobile social networks, games, apps, IM services, news sites, portals, mobile websites, etc.

This enrichment of our recruitment sources offers a true multisource blend and removes selection bias due to specific sources skewed to specific profiles. Dynamic sampling also counterbalances the "professional survey taker effect" because those respondents recruited at the moment are generally not part of any panels and never answered any surveys before.

In the rare cases where TGM Panels combined with dynamic sampling are not enough to reach our client's audience, we might involve our partners' panels. We only work with pre-defined and renowned companies that can comply with our international quality standards.

No matter where the sample comes from (TGM Panel, TGM Dynamic Sample or TGM Panel Partners), before entering the primary survey, each respondent is redirected to our TGM hub and pre-screening questionnaire. System checks if they are not a bot but human, check uniqueness using digital fingerprint tech, ask screening and quality check questions. Qualified respondents are redirected to the main survey afterward.

The optimal blend of sources will depend on many components but primarily the complexity of the study and target audience to reach. We will use our experience and knowledge to determine the feasibility of each available source we have access to, finding additional sources if needed to fulfill fieldwork successfully. Then we will combine our technology to blend all different sources and apply our deduping/quality check digital fingerprint to ensure that one unique respondent can only enter once.

For tracking studies, we keep the same sources distribution percentage for data consistency and comparability purposes.

Thanks to our multisource recruitment method, sourcing challenging targets is a relatively straightforward process. We have agreements with multiple region-specific affiliate websites where the target audience can register and invite that group to join the panel after completing an initial profiling survey. Once they are panel members, we can segment their profiles and identify those who qualify for that target.

Thanks to the ubiquity of internet access and smartphones in every country TGM operates, our panels cover most targets. However, we might require working with an external panel partner in the rare case where TGM panels with the TGM Dynamic Sample are not enough to reach the client's target objectives.

We at TGM Research are creative and never take no for an answer. For a specific hard-to-reach audience, we can run a customized recruitment campaign using specialized media. We believe that all methods have their pros and cons, and sometimes offline approach will work better (BtoB, specific store or shopping location). Using mixed methods is quite common, and we will not hesitate to consider and suggest different options to our clients. Once we have traction, we know that referrals can also be a good solution, using the right incentive and study methodology. Lastly, we can also liaise with our extensive network of trusted partners to help us gather the data we need (BtoB profiles included).

In accordance with our and industry guidelines, we base our decision to use pre-authorized suppliers exclusively on the following criteria:

1. Quality (reliability of the data obtained by the supplier, methods used in its creation and panel management are the initial indication of anticipated quality, along with the fraudulent-answer rate).
2. Delivery (the ability to recruit samples in the requested countries).
3. Price.
4. Speed and reliability (ability to quickly begin fieldwork; meeting deadlines).
5. Purchase reciprocity.

Before entering the primary survey, samples coming from Panel Partners will follow the same steps as TGM Panel or TGM Dynamic Sample. Each respondent is redirected to our TGM hub with both technical check and pre-screening questionnaire (system checks if they are not a bot, check uniqueness using our unique fingerprint technology, ask screening questions including traps).

Qualified respondents are redirected to the primary survey afterward. We only work with pre-defined and renowned companies that can comply with our own and international quality standards.

On top of that, we also monitor and track panel partners' metrics and match their KPIs against ours. We maintain a list of top performers that evolves overtime to make sure we use the best options when working with a panel partner is required.

Samples get randomly selected from among panelists who meet project requirements. This selection is made in proportion to related project quotas. Additionally, we apply multiple quality-control policies that prevent over-surveying and infra-surveying so that all the panelists are surveyed with optimal frequency.

Preparing strategy, we utilize a library of multiple demographics census data for the specific countries where we operate. We always make sure that the final study sample distribution matches and reflects the population distribution. The reason is obviously to ensure statistical significance and data interpretation from the study at the population level. We can draw representative samples of the national population by combining age, gender, region, education and other socio-demographic criteria.

Sampling plan strategy is always discussed upfront with a client in full transparency, considering country specificities and complexity of the study/target to reach / associated quotas. In addition, we always provide several scenarios with costs, feasibility and timing implications so clients can do a trade-off of offered options.

For most of the projects, we recommend applying basic demographics quotas such as: age, gender, region, income or education. However, certain methodologies require specific brand/category/product quotas, which we can discuss with the client.

TGM Research technology does not include a router per se. Thanks to our profiling capabilities, however, our solution will automatically propose to every respondent a selection of surveys matching key demographics, profiles and the interest of the respondent. Each survey from the selection mention the LOI with associated incentive, and the respondent can choose the survey he would like to take. This process allows us to increase conversion, which means better respondent experience and better fieldwork execution.

We collect standard demographics for any respondent entering our TGM hub (technical check + screening standard qualifications) before any survey, no matter the source or origin.

To become a member of TGM's proprietary panels, members must provide:

• Basic - socio-demographic information: age, gender, region, occupation, etc.
• Enhanced – deep profiling: Most of our panel members also volunteer further information by answering profiling variables within the TGM panel portal at any moment of their panelist life cycle.
• The more information panelists provide, the more incentives they can receive and/or unlock specific features.
• Better profiling increases their chances drastically to be invited to surveys matching their personas and interests.
• Through our profiling surveys, we enhance our dataset with hundreds of additional background data items, grouped into several modules: household demographics, employment and company decision-making, electronic devices, automobiles, consumer finance, insurance, travel, technology, leisure, Internet habits, health, telecommunications, etc.

Socio-demographic information is updated annually, and modules for additional information are collected by profiling all panel members after survey completion.

We deal with low-incidence projects on a case-by-case basis. First, we look into the profiling data we have to see if we can pre-target some of our respondents. If the information is not available, we generally combine desk research and internal feasibility check (based on small sample size mini poll). Internal investigations grant us a clear view on project feasibility and either decide whether to run fieldwork or not or help a client with a best-effort basis if it's a sample-only project.

• Once a new survey is available, it will appear automatically within the panelist panel portal (personal interface) if the respondent profile is eligible to survey specifications. (see question 8)
• Every listed survey contains the LOI, the associated incentives and the remaining time (DD, HH) for completion. We prefer not to mention the topic to avoid selection bias for the respondent.
• To increase traffic and start a specific survey, we may use emails or push notifications to contact our members and notify them about a new survey. (They can change email/push notifications rules within their login interface).
• We have resting rules for respondents to keep the response rate at a reasonable level.
• Emails invitations contain LOI, incentives associated, remaining time for completion and a unique link to access our system before entering the survey and finally, an opt-out option.
• Respondents coming from dynamic sampling sources will see the same information (LOI, incentives, remaining time) before choosing to start their journey with us or not.

Incentives primarily depend on LOI and the complexity of the target/project, but of course, every country is different. As an example and for the same LOI, incentives for a high-income respondent in Switzerland and a student in emerging countries will not be the same. As experts in data collection and sampling, we know how to adapt incentives for any study, target or country to keep it fair for the respondent and motivating enough. Doing so leads to maintaining a good response rate and engagement through the survey and the panel in general.

We have different types of incentives depending on the country. For example, we use Amazon Gift Card, Gift Codes, Paypal, and the respondent can choose which method he prefers. We credit Incentives after the survey and once the quality checks for the respondent are successful. In addition, and mostly designed towards dynamic sampling, we use virtual and digital rewards such as unlock features within an app, free subscription or upgrade, virtual points or new levels (for respondents gathered via gaming apps).

Finally, panelists also accumulate TGM panel points when they answer profiling questions, feasibility checks, polls and then can redeem those points after a certain threshold to incentives described previously.

To give an accurate estimate of viability, we generally need to know:

• Geography / region.
• Target audience and screening / estimated incidence: we can help clients find the incidence (IR) through our database and/or feasibility checks.
• The number of completes required.
• LOI (length of the interview): increasing LOI will impact feasibility due to fatigue and drops, a restrictive pool of respondents, and limitations from mobile device traffic.
• The desired filters and quotas required: if not available, we can help with that and make suggestions
• Fieldwork duration.
• Other project-specific information such as type of study (Adhoc, tracking, diary, recontacts, module to install, qualitative part) and deliverables needed: sample only, questionnaire design, translation, scripting, data processing, analysis, reporting, presenting, etc.

Being primarily a mobile panel, one of the most critical elements is understanding how the study can be built in for survey design. Thanks to our team of experts, we can use either script and host the survey to make sure it's fully accessible from mobile devices or give recommendations to our clients, helping them with their questionnaire design and programming.

In the rare situation where the study could not be fully adapted and taken from a mobile device, we will include this parameter in our feasibility estimations (penetration of Internet users via smartphone is much higher than desktop/laptop). It is even more true for emerging countries where excluding smartphone users (a study only accessible from desktops) would be statistically arguable and might generate a misleading interpretation of the data.

We also give great importance to being fully transparent with the study objectives. As researchers, we can certainly help clients by pointing out any limitations or concerns and adding recommendations or sharing ideas on how to maximize insights from the study.

One of our goals is to provide a high level of satisfaction to our panelists continually. We believe a big part of this is providing the right level of incentives (better than most of the competitor's benchmarks) and delivering top-notch customer service

We have a process in place at the end of each survey, asking respondents questions about survey satisfaction (interest, design, overall experience) and source of improvement. This set of questions is not mandatory.

We use this data to rank our top-performing clients and the ones with the lowest scores. In addition, we may use this information with our clients if we think they are concrete ways of improvement. In any case, this is always a great chance to discuss relationships between survey design, respondent engagement and data quality with our clients for future developments.

We at TGM Research aim to keep regular and transparent project communication with our clients throughout the project. After the study launch, we check that fieldwork metrics align with what was agreed upon at the quotation stage to avoid any feasibility or change in the pricing.

We are flexible and understand that our business can sometimes be unpredictable. However, we shall discuss with our clients if there are significant differences to take corrective actions. In addition, we will continue to monitor fieldwork until the project ends.

Survey data quality depends on many different factors. Including finding respondents who are providing valuable answers and eliminating those who may provide low quality. It's also crucial to ensure that surveys are well prepared, mobile optimized and engaging. We take various steps to ensure we are providing the best survey quality possible.

Early fraud detection within the TGM Hub system and before starting a survey (All sources):

• Use proprietary fingerprinting technology (widespread combination of device characteristics) for all users. It provides respondent identification (regardless of whether they cleared cookies or changed IP address). As a result, we can identify the uniqueness of the respondent – he can't take surveys more than once, knowingly or unintentionally.
• Additionally, we track IP and cookies to identify repeat or fraudulent manners before starting any survey. GeoIP technology to ensure that users are from the country we need.
• We automatically eliminate all users from suspicious IPs and traffic from proxy servers and bot suspicious.
• Use of logically related questions - as a part of the screener. (traps - same questions asked twice, different way, logical consistency checks). If the respondents fail to answer any of them, we remove them from the sample.

Only users who pass all quality checks described previously get redirected into a real-time survey. Devices with high numbers of violations are blacklisted and permanently removed from our surveys.

Full survey data validation process – when TGM Research scripts and host the survey:

• The time needed to complete the survey. Speeding through a survey is one of the most critical indicators showing that a respondent is not reading and/or answering the survey truthfully. Therefore, we measure the average time spent by respondents answering individual questions and disqualify outliers from the survey with a new sample.
• Trust Score algorithm – questions; that is flagging all respondents who have a typical straight-lining behavior (always click the first/last answers) and have patterned responses. In addition, trap/red herring questions, other internal consistency checks are part of the Trust Score.
• Response Quality score – machine learning human supervised checks to assess response quality, looking at gibberish (responses with non-word answers), fake answers and other illogical discrepancies.
• Both Trust Score and Response Quality are evaluated for each respondent and used to calculate a “Dirty” Score. Respondents with the highest Dirty Data Score are disqualified and replaced with a new sample.

Panelists who do not pass the quality control first get a warning; they are then temporarily removed from the panel and permanently unsubscribed in case of recidivism.

Although we may use emails / SMS or other push notifications to inform our panelists about a new study available, most of the participation comes from surveys available on each respondent's panel portal. This approach limits unwanted emails and gives respondents full flexibility and participation when this is "the good time" for them.

We suggest surveys that are likely to match the respondent's profile, but at the end, each panelist decides which survey to be part of and the frequency. We, however, put a limit for a maximum number of surveys that participants can complete per month. This limitation varies across sources and other specificities like the country for fieldwork or profiles.

Same as answer 19.

We prefer to give respondents full flexibility of participation frequency (we, however, track this metric to make sure this remains acceptable). Although we use emails / SMS or other push notifications to inform our panelists about a new study available, most of the participation comes from surveys available on each respondent's panel portal. This approach limits unwanted emails and gives respondents full flexibility and participation when this is "the good time" for them.

We suggest surveys that are likely to match the respondent's profile, but at the end, each panelist decides which survey to be part of and the frequency. We, however, put a limit for a maximum number of surveys that participants can complete per month. This limitation varies across sources and other specificities like the country for fieldwork or profiles.

Exclusions rules across/within categories are something we can do on a project basis if tracking and specific client's requirements.

Yes, we maintain contact and participation history data for a logic of panel management, panel health, and associated internal KPIs. In addition, we collect and use individual-level data information to understand and determine respondents' participation and analyze segment behavior to improve its incentive strategy.

As well as a satisfaction score survey, we collect all traffic information for every survey. We have no problem sharing this information with our clients; depending on the purpose and after we verified that information, we would share it as long as it complies with our policy regarding data privacy and protection for respondents.

We can supply specific data to the client while always prioritizing panelists' privacy and the confidentiality of their information.

This part goes back to question 18 related to quality checks and our process before the survey and at the survey data level.

In re-cap, we regularly employ the following mechanisms to avoid duplicates:

1. On the registration level - panel members with inconsistent data in their information fields (name, address, email, etc.) and registration profile are flagged and not invited to join panels until their identity is confirmed.

Our double opt-in registration combined with digital fingerprinting technology helps us detect fraudulent behavior early.

1. Response level - inconsistency in survey data and profiles get flagged for potential fraud detection.
2. Whenever panel members redeem their rewards, their address, street and city are manually checked against our records to prevent professional fraud and duplicates.
3. We use a cookie and digital fingerprint to ensure that two different panelist identities don't access our surveys from the same computer.

We don't differentiate between consumer or business samples when using these procedures.

TGM Research uses a double opt-in process for all TGM panelists to make sure they voluntarily want to join us. No matter the recruitment source or channels, all participants who want to join TGM Panel will start their journey on our landing page depending on the country panel the participant wants to join. (Example for the UK TGM panel landing page: https://tgmpanel.uk/join.html).

Participants enter their email and have full access to our Privacy Policy and Terms of Services they need to agree on (ticking box) before continuing the registration process. Following that, TGM Panel will send a confirmation email automatically to participants with a unique link to validate the subscription process to the TGM Panel. (that is the double opt-in part).

At this stage, the individual becomes a TGM panelist; he can answer profiling questions and participate in surveys. They can opt-out and unsubscribe from the TGM panel either from their interface or when they receive emails / SMS / push notifications.

For the cases where dynamic sampling gets used, respondents get directly redirected from our global network of mobile publishers. They click on our TGM Panel ad and land on our TGM hub for pre-survey quality process and pre-screening. They will also have to agree with our Privacy Policy and Terms of Services. Single opt-in gets used here.

You can find our TGM panel privacy policy at: https://tgmpanel.uk/privacy-policy.html for the UK. The content of the policy may vary for each country where TGM Panel operates to meet local regulations requirements. Kindly contact us for more information or Privacy Policy for a specifics country. Respondents can access panel privacy and terms of services anytime within their personal TGM interface.

All data and project materials provided by TGM panel members and clients get stored on secure servers. Only authorized personnel have access and only for administering TGM panel member accounts and surveys. The premises within which TGM Research panel servers are located is secured from public or unauthorized access both physically and electronically using the latest technologies and security systems, including but not limited to; firewalls, data encryption, IP-based permissions, CCTV, and swipe entry access control.

Data back-ups are subject to the same levels of physical security and authorized access.

There's no 100% reliable method for protecting confidential client material. For example, no one can prevent a panelist from taking photos of their screen. However, TGM Research takes the following security actions:

1. Disabling the right button of the mouse to avoid the use of its functions.
2. Continuous deleting of the clipboard to prevent copying through the use of ctrl+C / ctrl+V.
3. Prevention of survey printing.
4. Specific anti-copy software.
5. Anti-copy strategies. For example, we insert individualized watermarks in the pictures used so that if a panelist does distribute a picture, we can determine who it was.
6. Warning message.

We take the listed above and other measures to prevent and reduce the ease with which materials can be taken from a survey, for instance, encrypting file paths or video content. We also use unique links to ensure a survey is not available after the respondent completed it. If needed, adding a simple non-disclosure agreement at the beginning of the survey is also an option.