Research interviewing is dialing with a consent record
A B2B research project lives or dies on completed interviews, and completing them means getting a busy professional to spend ten minutes on the phone. The dialing is unglamorous: gatekeepers, wrong numbers, disconnects, and a small number of people who are willing to talk.
What separates research interviewing from sales calling is the record. Every interview needs to show that the purpose and length were stated, that the respondent agreed, and that any recording was permitted. That record is the product as much as the answers are.
DialBreeze rings up to three numbers for one interviewer and records every completed interview. The interviewer conducts the conversation. The AI writes the administrative note.
Three queues a field team uses
- Fresh sample. Client-supplied companies and screened contacts. The goal is a completed interview or a defined refusal.
- Callback queue. Respondents who named a time. These are the highest completion rates in the project.
- Partial interviews. Respondents who started and had to stop. A short callback finishes them.
What belongs in the record
- Consent: the purpose, the length and the sponsor were stated, and the respondent agreed to continue.
- Recording permission: asked and logged, because several states require all-party consent.
- Respondent role: the qualification that makes the interview usable.
- Topics covered: which sections of the questionnaire were completed.
- Incentive: whether an honorarium was offered, in what form, and where it goes.
- Refusals and do-not-contact requests: permanent and project-wide.
- Next step: a callback time, a follow-up interview, or document delivery.
After each recorded interview, DialBreeze writes a transcript and then fills those fields with a short summary. Verbatim answers belong in the questionnaire system, and the calling record should follow the project’s own retention policy. Because the sample is often a client’s data, access and retention should match whatever confidentiality terms apply.
Is this research or telemarketing
The distinction matters more here than in most categories. Under 16 CFR 310.2, telemarketing means a plan or campaign conducted to induce the purchase of goods or services or a charitable contribution. A genuine research interview that does not induce a purchase does not meet that definition. A call that induces a purchase does, regardless of what the project calls itself.
That is a design decision, not a script detail. If a client wants buying intent measured, that is a research question. The moment the interviewer pitches, the rule set changes.
The rules that still apply
Business-to-business calls that do induce a purchase fall under the narrow exemption in 16 CFR 310.6(b)(7), which is not a TCPA exemption. 47 CFR 64.1200(a)(1) restricts autodialed and artificial or prerecorded calls to wireless numbers, and business respondents increasingly answer on cell phones. Where a call is a telephone solicitation, calling hours run 8 a.m. to 9 p.m. local time at the called party’s location under 47 CFR 64.1200(c)(1). Where more than one line can connect at once, 47 CFR 64.1200(a)(7) caps abandoned telemarketing calls at three percent over a 30-day campaign and treats a call as abandoned if a live interviewer is not connected within two seconds of the greeting, so the line count should match what your interviewers can answer.
Why the throughput number matters less here
Measured in production use, 3-line sessions, 90 days: a median of roughly 85 dials per active calling hour and roughly 600 dials per operator day. For a research project the more useful metric is interviews completed per hour of interviewing, which depends on the incidence rate in the sample. The dial figure tells you how many records a block can cover, not how many interviews it will yield.
What you need to start
- An approved consent and recording script.
- Your own Telnyx account with numbers and caller ID.
- A sample CSV with company, contact, role and consent status.
- A refusal and suppression procedure that holds across waves.
The 14-day trial runs in a sandbox. Load a small sample and run a block to test the script and the note quality before the field period starts.
Incidence rate decides the plan, not dial count
In B2B research every project has an incidence rate: the share of contacted companies that actually qualify. A study with a five percent incidence rate needs far more dials per completed interview than one with twenty percent, and the field plan should be built from that number rather than from a general throughput figure.
The consequence is practical. If a project needs 200 completed interviews at a five percent incidence rate, with one interview per twenty answered calls, the arithmetic is unforgiving and the client needs to hear it before the field period starts. A calling workflow that shows attempts and outcomes per hour lets a project lead recalculate the plan early instead of at the end.
Panel and client-supplied sample behave differently
Client-supplied sample is usually cleaner: the client knows these companies and the contacts are real. The problem is that contacted companies may already be over-researched by other vendors, and a company that has been called about four surveys this quarter may refuse on principle.
Panel sample has the opposite profile. Contact details are verified and consent language is usually settled, but respondents are professionals at being interviewed and their answers can be shorter and less specific. Field teams should plan refusal follow-up, callback strategies and sometimes an incentive structure per sample source, and the dispositions should distinguish the two.
What a client actually wants back
A client buying a telephone study wants three deliverables: completed interviews that match the screener, an auditable record that the interviews happened with consent, and a completion report by segment. The audio recording and the structured note are what make the second deliverable possible.
That is why recording permission belongs in the consent language from the start. A study that finishes with 200 completed interviews but no clear permission record, or an unclear statement of what respondents were told, is exposed at the exact moment the client audits it.