Picture a head of insights in Chicago (or London, or Sydney) with three quotes from Singapore agencies open in three browser tabs. The scopes look alike, the day rates look alike, and every credentials slide uses the same words: bilingual, proprietary panel, senior team. She has a call on Thursday and roughly forty minutes to prepare, so what should she actually ask? I have taken versions of that call many times (often at what is 6 a.m. for the buyer), and I think most buyers ask the easy questions rather than the revealing ones.
Here is the figure that changed how I think about this. In a 2023 Quirk's case study, the research firm KNow Research reported a 19% rise in bad actors in its virtual fieldwork, then cut mid-fieldwork fraud to 2.7% by checking people at three stages: screening, scheduling, and the session itself. I read those numbers twice because they sound almost too tidy (and they are one firm's own results, not an industry average). Still, I would be surprised if Singapore were exempt, and the broader point stands. You cannot see recruitment quality in a quote or in a credentials slide, so you have to ask for evidence.
Here's the tension. An overseas buyer cannot walk through the facility, meet the moderator beforehand, or listen in on a recruitment call, which means proof has to travel by email. A good answer arrives in writing, with a document attached. A vague answer is information too, and it costs you nothing to collect.
Why a polished quote tells you so little
A quote compares well on the things it lists: the number of groups, the number of participants, the incentive, the venue, the report. It compares badly on the things it leaves out. It rarely says who will moderate, how the participants were found, where the recordings will sit afterwards, or which professional code the agency answers to. Those four unlisted items decide whether your findings are usable, and whether your legal team is comfortable.
I would also resist treating a low quote as a warning and a high one as reassurance. Actually, that is too simple. Price is a weak signal in both directions, and the cleaner test is whether an agency can price your scope line by line. Our guide to what market research costs in Singapore explains what moves those lines.
The Five Proofs
Over the years I have asked the same five things whenever a buyer wants to check us (or anyone else) before commissioning. Let's call them the Five Proofs, though I'm still not sure five is the right number, since I tried seven first and merged two. Each can be checked in about a week without flying anywhere.
The Moderator Proof
Who scoped the project, who moderates, and in which languages. A name, a CV, and a sample of how they probe.
The Panel Proof
Who owns the panel, how participants were found, and what stops a professional respondent slipping through.
The Data Proof
Who controls recordings and transcripts, where they are stored, and how the PDPA applies to your project.
The Outside Proof
Which professional code binds the agency, which reviews you can verify yourself, and which referees will talk.
The Quote Proof
A line-by-line scope with currency, validity, and payment terms, so any two quotes can be compared fairly.
The Moderator Proof, who is actually in the room
Start with the question buyers ask least: is the person who scoped your project the person who will moderate it? In many agencies the answer is no, and that is not automatically bad (a junior moderator with a strong guide can be excellent). It just changes what you ask next, which is the moderator's name, languages, and a short sample of their work. At Assembled, Felicia Hu scopes, moderates, analyses, and presents every project herself, so for us the answer is a single name, and her background sits on our expertise page.
Language is where the stakes are highest. In Singapore, a participant might move between English, Mandarin, Hokkien, and Singlish inside one sentence, and a polite refusal rarely sounds like a refusal. "Can consider" usually means no. A moderator who misses the register will write "open to the idea" in the notes, and your report will carry that error into a board paper. Is an interpreter enough? Often. Actually, that depends, because someone translating a beat behind cannot probe in real time. Ask which you are getting.
I think the best test is a short anonymised transcript excerpt that shows the moderator following up on a hesitant answer. Our guide to analysing focus group data shows what a good read looks like, and our focus group and in-depth interview pages describe how we run each format.
The Panel Proof, who owns the people you are paying to recruit
A panel is only as useful as the answers to two questions: who owns it, and how does a person get in? Our panel of 100,000+ members was built organically rather than bought from a third-party vendor, so we can describe how each person arrived. You should ask any agency, us included, to describe it in writing. What you want is a method you can picture: referral, community outreach, a screening call, and a re-check on the day.
The failure you are guarding against has a name. Quirk's glossary defines a professional respondent as someone who attends many sessions by joining different facilities' recruitment lists, and notes that moderators want them removed because they answer less objectively. A Greenbook webinar cites an estimate that 15% to 30% of responses are fraudulent, a range that mostly describes online data, so treat it as a warning and not a Singapore figure.
The checks that worked in the Quirk's case study can be requested by name: phone prescreens, trick questions in open-ended answers, video audition questions, locked invitation links, IP and device checks, and a re-screen in the virtual waiting room. Ask which of those a Singapore agency runs, and which it repeats on the day. Our posts on paid research panels in Singapore and qualitative fieldwork recruitment go deeper.
The Data Proof, where the recordings go after the session
Overseas buyers sometimes assume data protection is the agency's problem. It is partly yours. Singapore's Personal Data Protection Act (PDPA) governs how organisations collect, use, and disclose personal data, and the PDPC's overview of the PDPA is the plain-language place to start. A recording of a face and a voice tends to count as personal data. In regulated sectors such as financial services, other rules sit on top.
The roles matter more than the jargon. The PDPC explains that a data intermediary processes personal data on behalf of another organisation, and must protect it, avoid keeping it longer than needed, and report breaches promptly. Whether your agency is one depends on the contract. Ask in writing for processing terms, storage location, retention, breach timeline, and consent wording. I am not a lawyer, so have counsel read them.
Cross-border sharing is the part buyers overlook. The PDPA's Transfer Limitation Obligation in section 26 prohibits sending personal data outside Singapore unless prescribed conditions are met, which require the recipient to protect it at a similar standard. If your team in Chicago will receive full video recordings, someone needs to think about that transfer before the first session. Ask who owns the recordings and transcripts afterwards, and get ownership, access, and deletion written down.
How worried should you be? I am not sure the numbers settle it. The PDPC's 2023/24 data breach analysis found that reported large-scale breaches, those affecting more than 500 individuals, rose 41% within a year. That covers all sectors, so it probably says little about research agencies, but it is a reason to ask what happens to files after a project closes. The IMDA's Data Protection Trustmark is one way to show accountable data protection, so ask whether an agency holds it.
The Outside Proof, codes, reviews, and references you can check yourself
Professional codes give you something outside the agency to hold it to. ESOMAR publishes the ICC/ESOMAR Code, which it describes as the global standard for ethical market research and data analytics. The UK's Market Research Society publishes its own Code of Conduct, and its 2023 version says anyone describing a sample as representative must report the sampling parameters. The MRS also publishes guidelines for qualitative research.
One caution. Writing for MRS Research Live and ESOMAR Research World, as Felicia has, is not the same as being bound by their codes through membership. Ask every agency, us included, which code it works to and for written evidence.
Reviews are the cheapest proof to check and the easiest to misread. We have 700+ Google reviews, and our reviews page groups quotes by service. For any agency, sort by newest, look for reviews that describe a specific session, and check whether the dates cluster suspiciously.
References come with a catch. Many agencies, us included, cannot name clients, because confidentiality is part of the job. Ask instead for a referee who can speak to process, or an anonymised case study showing the brief, the method, and the decision it informed.
The Quote Proof, what a clear quote must include
A quote you can compare has to name its parts: groups or interviews and participants; who moderates and for how many days; the recruitment method and incentive per participant; the venue or platform; translation; recording and transcription; report format and revision rounds; the presentation; and the currency, GST treatment, validity, and payment terms. If a line is missing, ask before comparing totals. Our guide to writing a research brief shows how a tighter brief makes every line easier to price.
For scale, here are our published starting points for three common programmes. These are "from" prices that move with scope, and the conversions use the 30 October 2026 rates (1 SGD = 0.78 USD, 0.58 GBP, 1.11 AUD), which will have shifted by the time you read this.
| Programme | From (SGD) | Approx. USD | Approx. GBP | Approx. AUD |
|---|---|---|---|---|
| Focus group programme (4 groups, 32 participants, with recruitment, moderation, venue, incentives, recording, transcription, and report) | SGD 12,000 | US$9,400 | £7,000 | A$13,400 |
| Online survey | SGD 8,000 | US$6,300 | £4,700 | A$8,900 |
| In-depth interviews (8 interviews) | SGD 5,000 | US$3,900 | £2,900 | A$5,600 |
Compare scope before totals. A quote at half the price that leaves out incentives, transcription, and a presentation can cost more once those lines are added back. For a full entry plan, our guide to entering the Singapore market covers sequencing.
Scoring the Five Proofs
Here is the scorecard I would use. Score each proof 0 (no answer), 1 (a verbal answer only), or 2 (a written answer with a document you can check), out of 10. I chose the thresholds by feel, not data, so adjust them to your appetite for risk.
| Proof | Ask for | A strong answer looks like | A red flag looks like | Score (0, 1, 2) |
|---|---|---|---|---|
| Moderator | The moderator's name, languages, and a short transcript excerpt | A named person who also scoped the project, languages stated in writing, a sample showing follow-up probing | "Our team" with no name, translation by an interpreter nobody introduces | |
| Panel | Who owns the panel, how people are recruited, which re-screening checks run | A recruitment method described step by step, a re-check on the day, a cap on sessions per person | Recruitment outsourced with no checks described, no limit on repeat participants | |
| Data | Processing terms, storage location, retention, breach timeline, consent wording | Written terms, a defined retention period, a clear answer on overseas transfer | "We follow the law" and nothing in writing | |
| Outside | The code followed, evidence of membership, reviews you can verify, a referee or anonymised case | A named code, recent reviews with specific detail, a referee or case willing to speak | Directory badges only, reviews posted in one burst, no referee | |
| Quote | A line-by-line scope with currency, GST, validity, and payment terms | Every line itemised, assumptions listed, the same scope across all bidders | A single total with no assumptions |
A total of 8 to 10 suggests you can proceed to a brief. A total of 5 to 7 means ask follow-up questions in writing. Below 5, I would pause. Apply it to us too.
What I would do before Thursday's call
If I were that head of insights, I would send the same five requests to all three agencies on Wednesday, in one email, and ask for written replies by Friday. I would not judge on price at that stage. I would read the replies for specificity, and probably phone one referee. I would also write the brief first, since our research brief guide helps define the decision the research must inform.
Is that enough? I am not certain this catches everything. Chemistry matters, and a scorecard cannot score it. Treat it as a filter that removes agencies you should not brief, not a machine that picks the one you should.
What overseas buyers ask before briefing a Singapore agency
How do I check a Singapore market research agency from overseas?
Who is responsible for personal data when a Singapore agency records our focus groups?
What is a professional respondent and why does it matter?
Does a Singapore research agency need to belong to ESOMAR or the MRS?
What should a Singapore market research quote include?
Checking a Singapore agency before you send the brief
A good vetting call ends with documents, not promises. If you want written answers to all Five Proofs, plus a line-by-line quote you can compare against others, tell us the decision your research needs to inform and we will send both, whether the project is a single study or a full market entry programme.
Request a quote →