Telephone research appears straightforward while the contact database is small. A team prepares questions, distributes numbers among operators, conducts conversations and transfers answers into a spreadsheet. With a large sample, the process becomes a bottleneck: three or four operators can conduct only several interviews at once, unanswered calls require further attempts, and records made by different people must be brought into a single structure.
VMTech DOO Beograd develops and implements VMTech Survey for automated telephone questionnaires. The solution conducts a conversation using an agreed questionnaire and retains structured answers for later analysis. In suitable campaigns, parallel calling can reduce the active calling phase from weeks of manual work to several hours. This refers specifically to active calling, not the entire research cycle: methodology, database preparation, a pilot, quality control and interpretation of results all require separate time.
What automated telephone surveying is
Automated telephone surveying is the collection of answers through a voice script. It asks approved questions in sequence, follows the required branches and records responses in an agreed format. Before launch, the script’s behaviour needs to be defined for different situations: a response option is selected, a respondent asks for the question to be repeated, remains silent, declines to participate or ends the call early.
The value is not simply in dialling a large number of phone numbers. It lies in executing the research logic consistently. Wording, question order, branching and the criteria for a completed questionnaire are fixed in advance. When manual interview results remain in notes and spreadsheets with different labels, they are difficult to combine and compare. A structured dataset reduces manual consolidation, although it still requires checking and cleaning.
VMTech Survey is intended to automate structured research. It should not be confused with an operational AI Call Center, which handles day-to-day enquiries and business communications. In telephone research, the questionnaire, sample, quotas and interpretation rules are central. If one call simultaneously becomes a questionnaire, a service request and a sale, the research objective becomes unclear and the answers are harder to assess.
How an automated survey differs from traditional CATI
In a conventional CATI study, an interviewer speaks with the respondent, reads questions from a screen and enters answers into a software system. A person can notice a misunderstanding, repeat the wording or provide an explanation permitted by the methodology. Operator involvement is particularly important for long interviews, sensitive topics, open narratives and situations where not every meaningful clarification can be described in advance.
In the automated format, an agreed sequence is delivered by a voice script. This approach is better suited to short, standardised questionnaires that are easy to understand by ear. The main branches are described before launch: which question follows a particular answer, how often wording is repeated, how an uncertain answer is handled and under which conditions the conversation ends.
Why active calling may take hours rather than weeks
With verified telephone infrastructure, a prepared script, agreed contact conditions and approved methodology, VMTech Survey is designed for approximately 1,000 simultaneous calls. This is the technical capacity of a specifically prepared environment, not an unconditional guarantee for every campaign. Actual concurrency depends on the telephony configuration, database volume and quality, questionnaire length, branching complexity and repeat-contact rules.
For a suitable study, such a configuration can reduce active dialling to several hours compared with weeks of work by a small group of operators. Yet the number of calls started is not the number of completed questionnaires. Results are affected by respondent availability, willingness to participate, invalid numbers, conversation length and the need for repeat attempts. Timelines are therefore assessed after database review and a pilot, rather than from one technical capacity figure.
What must be defined for research in Serbia
For a campaign in Serbia, uploading phone numbers and translating questions literally is not enough. The first local consideration is the language of the conversation. The client needs to determine which language or languages the audience is comfortable using for answers. A Serbian script should be checked for natural wording, pronunciation and whether scales are understandable when heard. If the database contains groups with different language preferences, questionnaire versions are best separated and tested before a broad launch.
The second consideration is the origin and composition of the contact database. For market research in Serbia, it is important to understand which regions, settlement types, buyer categories or other relevant segments are represented among the numbers. A database of existing customers is appropriate for assessing their experience, but without separate methodological justification it does not describe the entire market or the country’s whole population.
The third consideration is the contact rules. Before launch, the client defines acceptable time windows, the basis for contact, the content of the introduction, the procedure for consent, repeat-attempt rules and the limits on use of answers. These conditions depend on the research purpose, questionnaire topic and database origin. Technical setup does not replace the organisational and legal decisions for which the campaign client is responsible.
What to check before a broad launch
Consent and a script that is clear by ear
The introduction needs to state the purpose of the contact clearly and include an agreed procedure for participation or refusal. A telephone question itself should be understandable after one hearing. Long constructions, professional jargon and extensive lists of options increase the risk of error. For every block, permitted answers, repeated wording, branches and end conditions are defined in advance.
Pilot and quality control
A pilot questionnaire makes it possible to hear the script from the respondent’s perspective. During the pilot, the pace, pronunciation, question length, branches and the format of recorded answers are checked. If participants repeatedly ask to hear one phrase again, choose an uncertain option or end the call at the same point, the cause needs to be addressed before scaling. At the same time, indicators of a complete, incomplete and additionally reviewable questionnaire are documented.
Research that can use a voice script
Market research and purchasing habits
Telephone questionnaires can be used to study awareness, purchase frequency, selection criteria and experience of using a particular category. The voice format is appropriate when questions are understandable without images or extensive explanations. The composition of the database sets the limits of the conclusions: answers from one company’s customers cannot automatically be extended to all consumers in Serbia.
Customer and buyer questionnaires
A customer satisfaction questionnaire helps gather feedback about a product, service and communication. Asking about specific stages of the customer journey is more practical than limiting the survey to one general score. Answers about consultation, purchase, delivery and support can identify an area needing further review. A numerical score shows a direction but does not always explain the reason, so open questions or separate interviews may sometimes be needed.
Employee satisfaction research
A standardised telephone survey can be used for questions about working conditions, internal communications, clarity of tasks and attitudes to processes. Confidentiality, access to data and the permitted level of detail in results should be defined before launch. A consistent questionnaire logic makes answers comparable, but employee trust depends not on technology but on clear rules for handling information.
Product testing
Telephone questions can cover product-use experience, frequency of using particular functions, clarity of purpose, reasons for choice and an overall assessment. They can also test a response to a briefly and neutrally described concept when every respondent hears the same wording. Such questions should be short and should not require complex mental comparisons.
If a participant needs to assess packaging, design, an interface, a prototype, an advertising layout or a specification table, a voice channel alone is insufficient. Visual material should then be shown separately or another research method selected. A telephone script should not replace a demonstration with a detailed verbal description: different respondents may imagine different objects.
Public opinion surveys
Automated telephone surveying can technically be used to collect standardised opinions on public issues. However, conclusions depend on methodology, sampling, quotas, contact rules, treatment of unanswered calls and data interpretation defined in advance. Automation by itself does not provide representativeness, statistical validity or the right to generalise a result to Serbia’s population.
Where conclusions about public opinion are required, the contact database must be defined before calling: whom it represents, how participants are selected and which groups must be present in the final dataset. Without these decisions, the system collects structured answers from reachable respondents but does not automatically turn them into a substantiated picture of opinion across the country.
How structured answers shorten the path to analysis
In VMTech Survey, the result structure is defined together with the questionnaire. Each answer is linked to a specific question, and branches match the agreed conversation logic. The team receives an organised dataset without first transferring information from several operators’ notes. This does not remove the need to check completeness, clean data and control quality, but it shortens the technical path from interview to initial analysis.
Campaign results are checked first: questionnaire completion, branch correctness, recognition quality and compliance with agreed criteria are assessed. The method and frequency of access to interim data are agreed for the particular configuration and do not replace review of the final dataset.
If a confirmed answer needs to initiate an internal action, such as reviewing a complaint, assigning a task to a responsible employee or arranging follow-up communication, this is a separate area of business process automation. Its action owner, trigger conditions and exception-handling procedure are defined separately. The research methodology remains an independent part of the project.
How VMTech runs a pilot questionnaire
Once the parameters are agreed, a pilot script is prepared. Test calls check wording, branches, conversation duration and the structure of recorded answers. Ambiguities are addressed after the pilot and a working configuration is determined. A broad launch is considered after the script and quality criteria are agreed, while permitted concurrency is selected for the specific campaign.
The purpose of the solution and the pilot approach are described in more detail on the project page for VMTech Survey for automated telephone surveys. An engineering assessment can help verify the suitability of the voice format and infrastructure, but it does not replace a researcher’s work on the sample, quotas and interpretation rules.
When automated questionnaires are not suitable
- The interview requires a long open narrative and unpredictable clarifications.
- The respondent must examine an image, document, interface or large list of options.
- The contact database does not make it possible to establish whose opinions are represented in the result.
- The research purpose, contact conditions and data-quality criteria have not been defined.
- One call must simultaneously perform research, service and commercial functions.
- A ready explanation of causes is expected from the system rather than structured material for analysis.
- Guaranteed representativeness is required without preparing a sample, quotas and methodology.
Frequently asked questions
Can VMTech Survey be used for CATI studies?
VMTech Survey can reproduce the pre-agreed logic of a standardised telephone questionnaire, but this does not mean it meets every CATI requirement. The need for an interviewer, open clarifications and special research procedures is assessed for each project. Human involvement or a mixed format is preferable for long or flexible interviews.
Do approximately 1,000 simultaneous calls guarantee campaign completion within several hours?
No. This is the planned capacity of a prepared configuration with verified telephony and a suitable script, not universal throughput. The number of completed questionnaires depends on the database, respondent availability, conversation duration and contact rules. Several hours may apply only to active calling in a suitable campaign, not to the whole study.
Does automation ensure research representativeness?
No. Representativeness depends on methodology, sample formation, quotas, reachability of different groups and data-processing rules. The platform helps run an agreed questionnaire consistently and structure answers, but it does not compensate for contact-database bias or the absence of a research design.
What should be prepared for a pilot questionnaire in Serbia?
The required materials are the research objective, audience description, conversation language or languages, draft questions, contact-database origin, expected volume, quotas, contact rules and data-handling limitations. It is also useful to state what decision the team intends to make after reviewing the answers.
