Video customer interviews and async video surveys both involve video, and search engines often pile them into the same SERP, but they are completely different methodologies. An async video survey is a one-way prompt: the participant gets a question, records a video answer, and submits. There is no moderator, no follow-up, no adaptive probing. An AI video customer interview is a two-way moderated conversation where an AI probes 5-7 layers deep in real time, with optional screen sharing against a live URL or prototype. Both are useful. They are useful for different jobs.
This piece is methodology framing rather than a product pitch. We sell one of the two methods (the moderated interview side); the other side has credible incumbents like Voxpopme and OneClickResearch that have shipped good async-video products for years. The honest answer to “which should I use” is that it depends on the question you are asking. Volume vs depth. Sentiment vs reasoning. Trend line vs decision-grade evidence. By the end of this guide you should be able to pick the right method without us, or know when both belong on the same program.
Quick note on what this is not. If you arrived from a search for “AI video interview” expecting hiring software, neither method covered here is for you. HireVue, Mercor, and Interviewer.AI screen job candidates. The two methods discussed here are for customer research only.
What’s the difference between video customer interviews and async video surveys?
An async video survey is a one-way prompt where the participant records a self-narrated video answer with no live moderator and no follow-up. An AI video customer interview is a two-way moderated conversation where an AI probes adaptively, runs 5-7 layers of laddering, and optionally screen-shares a website or prototype while the participant interacts. Async wins on volume and unit cost; moderated wins on depth and adaptive probing.
Async video surveys, defined
An async video survey is a research instrument that pushes a recorded prompt to a participant and collects a recorded video answer in return. The participant clicks a link, sees the question, records up to a defined length (typically 30 to 120 seconds), and submits. Some platforms layer in light AI on the back end to transcribe, code sentiment, and pull highlight clips. The participant never speaks to a moderator. The platform never asks a follow-up question.
Tools in this category have been around for over a decade. Voxpopme is the long-running incumbent, originally built around video survey and video diary collection. OneClickResearch sits in the same space. Several panel providers have bolted on async-video features. The format is well understood, the tooling is mature, and the unit economics are favorable for studies that need volume more than depth.
The methodology has a particular shape. You write a small set of prompts (usually 3 to 7), distribute the survey to a panel or customer list, and collect short videos at scale. Analysis happens after the fact: human coding, AI coding, or a mix. The output is a quantified pattern across hundreds or thousands of clips, plus a highlight reel of representative responses.
What async surveys are good at: capturing initial reactions, sentiment at volume, brand tracking, ad recall, satisfaction trending, persona shading, and any study where the question is genuinely closed-ended in spirit even if the answer is verbal. What they are not good at: anything that requires the moderator to ask “and why?” The single biggest limitation of async video survey methodology is the absence of a probe.
A note on participant quality. Across both async surveys and moderated interviews, the industry has a fraud problem: roughly 30-40% of online survey responses are now flagged as fake, scripted, or low-quality, depending on which panel and which study. Async surveys are particularly exposed because there is no live moderator to detect a participant who is reading an AI-generated script into the camera. Most credible async-video platforms now run identity validation and behavioral signals at the panel layer to filter this; the more responses you collect at unit-cost, the more important that defense is.
AI video customer interviews, defined
An AI video customer interview is a moderated conversation between a customer and an AI moderator over a browser-based video call. The AI handles greeting, structured probing, follow-up questions adapted to what the participant just said, optional screen-share setup, and the wrap. The participant joins via a link, gives camera and screen-share permission, and the AI runs the entire session. No human researcher in the room.
Three things distinguish this methodology from async surveys. First, the moderator is live and adaptive: when a participant says something interesting, the next question is built from what they just said, not from a fixed script. Second, the AI applies 5-7 layers of McKinsey-style laddering on every session, which is what surfaces the why beneath the what. Third, optional screen sharing lets the participant interact with a live URL, Figma prototype, or any web-accessible asset while the AI probes their on-screen behavior in real time.
User Intuition runs these video interviews on the same adaptive AI moderator that powers our voice and chat interviews. Studies start at $200 and return results in 24-48 hours. Sessions run concurrently and asynchronously across the platform’s 4M+ pre-vetted global panel, which spans 50+ languages. Pricing on the Pro plan is $20 per audio interview equivalent with visual capture stacking on top, and the platform holds a 98% delivery satisfaction rating across studies. We pay-for-quality only: sessions that fail any layer of the five-layer fraud and identity defense do not count toward usage.
The output is different from async too. Each session produces a replayable video clip synced to a verbatim transcript, scroll and click activity captured against on-screen behavior, and quantified themes derived from the laddering. You can jump from a finding back to the moment in the conversation it surfaced and watch what the participant did on screen at that instant. For a deeper read on the modality itself, see the video customer interviews complete guide or the video customer interviews platform.
The best mental model: an async survey asks “what do you think?” and gets an answer. A moderated AI interview asks “what do you think?” and then asks “why” four more times, watches what the participant does while answering, and pulls a verbatim citation tied to the moment.
Methodology comparison: depth vs breadth
The fundamental tradeoff is depth versus breadth. Async video surveys optimize for breadth: more responses, lower unit cost, faster fielding, simpler analysis. Moderated AI video interviews optimize for depth: layered probing, on-screen evidence, adaptive follow-up, decision-grade output per session. Most teams need both at different points in a research program; very few teams need only one.
| Dimension | Async Video Surveys | AI Video Customer Interviews |
|---|---|---|
| Moderator presence | None | Live adaptive AI moderator |
| Adaptive probing | No (fixed prompts) | Yes (5-7 layer laddering per session) |
| Screen sharing | No (or static stimulus image) | Yes (live URL, Figma, prototype, mockup) |
| Cost per response | $5-15 | $20-50 |
| Session length | 30-120 seconds per prompt | 20-35 minutes |
| Sample sweet spot | 500-5,000 responses | 50-500 sessions |
| Time to results | 1-3 days | 24-48 hours |
| Output | Quantified sentiment + highlight clips | Replayable clips + transcripts + themed findings |
| Best for | Volume, sentiment, trending | Depth, reasoning, on-screen evidence |
Here is the part most teams get wrong: the choice is not between cheap-and-shallow versus expensive-and-deep. The right framing is what question you are answering and which method’s data shape fits that question. A brand tracker run as moderated interviews wastes most of the depth and burns budget on follow-up questions you didn’t need. A concept test run as an async survey collects 1,000 surface reactions to a stimulus nobody actually engaged with, and you ship the wrong design. The methods are not interchangeable at different price points; they answer different questions, and the cost difference reflects the depth of the answer rather than a quality gap. Pick the method that maps to the question first, then defend the budget on the basis of what the data is actually going to be used for downstream.
Cost comparison
The unit-cost gap is real but narrower than it looks once you account for what each method delivers. Async video surveys typically price between $5 and $15 per completed video response across mainstream platforms, depending on screening complexity and incidence rate. Diary studies (multiple recordings per participant over a window) push higher because the panel pays per occasion. AI moderated video customer interviews run $20-50 per session on platforms with adaptive moderation; User Intuition is $20 per Pro audio rate equivalent for moderated audio, with visual capture stacking for video.
The math gets interesting when you express both as cost-per-finding rather than cost-per-response. An async survey at $10 per response with 1,000 responses costs $10,000 and produces a directional sentiment pattern plus highlight clips. A moderated interview program at $40 per session with 100 sessions costs $4,000 and produces 100 individually probed conversations with verbatim citations, on-screen evidence, and quantified themes. The async program reaches more people; the moderated program produces more depth per dollar.
Neither is universally cheaper. A volume study run as moderated interviews would be punishingly expensive (1,000 sessions at $40 = $40,000, often unjustifiable). A depth study run as async surveys would deliver shallow data the team cannot defend (100 video responses with no probing rarely answer a concept-testing question). User Intuition is built around the moderated band specifically, which is why the unit pricing only makes sense once depth is the goal.
For a complete cost breakdown of moderated AI video interviews specifically, see the video customer interviews cost guide. For the broader AI moderated video customer interviews methodology, the full pricing math is laid out per session and per study tier.
Speed comparison
Both methods are dramatically faster than traditional moderated video labs, which take 4-6 weeks to clear 100 sessions because a human moderator caps at 10-20 sessions per week. Async and AI-moderated both collapse that timeline.
Async video surveys field in hours: the survey link goes out, participants record on their own time, and most respondents complete within 24-48 hours. Analysis takes another day or two, depending on volume and how much human coding sits in the loop. End-to-end, an async video survey study delivers in 1-3 days for a sample of a few hundred to a few thousand.
AI moderated video customer interviews field in 5-10 minutes (the time to set up the study and recruit) and return full results in 24-48 hours. Sessions run concurrently across hundreds of participants 24/7. User Intuition handles recruitment, fraud screening, moderation, transcription, and theming. The 24-48 hour window includes the time it takes participants to complete their sessions plus the analysis pass; you do not separately wait for fielding and then wait for analysis.
For most decisions, both methods are fast enough to fit inside a sprint cycle. The async approach is marginally faster on the clock, but the moderated approach delivers richer data in a similar window. The right speed comparison is against the four-to-six-week traditional lab study, where both methods win by an order of magnitude. For a head-to-head on the best video research platforms, the speed dimension is one of three or four that matter.
Sample-size comparison
Sample-size sweet spots reflect the unit economics. Async video surveys are designed for 500-5,000 completes, where the marginal cost of the next response is low enough that volume drives the analysis. Below 500, you are paying setup cost without enough data to justify the method. Above 5,000, returns diminish unless you are running a tracker that needs that volume to be statistically meaningful at the segment level.
AI moderated video customer interviews work best at 50-500 sessions. Below 50, you have not stress-tested your hypotheses across enough segments. Above 500, the marginal interview tells you less than the previous one did, and you are paying for depth you no longer need. Most User Intuition video studies land between 60 and 200 sessions for exactly this reason. For directional concept testing, 30-50 interviews per concept (90-150 total across three concepts) is the typical band. For prototype validation on a single design direction, 50-100 interviews. For hypothesis confirmation in a known segment, 30-50 interviews.
Both methods can stretch outside their bands. You can run an async survey at 200 responses if that is all the budget allows; you will get a directional pattern. You can run moderated interviews at 1,000 sessions if the decision warrants it; you will get more depth than you need but the data still works. The bands are about where each method delivers its best ratio of cost to insight.
A program that needs both volume and depth typically uses both methods in sequence, which we cover later. The hybrid pattern is the most common pattern among teams that have used both methods for a quarter or more.
Which method wins for which research job?
The most useful frame for picking between async and moderated is mapping common research jobs to the method that wins. The table below covers the eight jobs that come up most across product, design, marketing, and research teams.
| Research job | Best method | Why |
|---|---|---|
| Concept testing | Moderated wins | Needs probing on the why behind preference |
| Prototype testing | Moderated wins | Needs screen-share + on-screen behavior |
| Brand tracking | Async wins | Needs volume and cadence |
| Customer satisfaction trending | Async wins | Needs volume + repeated waves |
| Win-loss analysis | Moderated wins | Needs depth on multi-factor reasoning |
| Churn diagnostics | Moderated wins | Needs probing on the breaking point |
| Ad / creative testing | Either | Depends on whether you need volume or depth |
| Persona shading | Async wins | Needs volume across many segments |
Concept testing is the most common moderated win. The whole point is to surface the why beneath the preference, and the why never lives at the surface. An async survey asking “which concept do you prefer?” returns a vote count. A moderated interview asking the same question and then probing four layers deep returns the structural reason the vote went the way it did, which is what design teams need to ship the right thing. For more on this, see solutions/concept-testing/.
Brand tracking is the canonical async win. You need to know whether brand sentiment moved between Q1 and Q2, and you need the answer to be statistically defensible across segments. An async tracker fielded quarterly at 1,500 responses per wave delivers the trend line cleanly. A moderated study at the same scale would cost five times as much and deliver the same headline number with extra depth that the tracker doesn’t use.
The two ad / creative testing rows are intentionally split. If you need to know whether creative resonates across a target population, async wins because you need volume to detect the signal. If you need to know why creative resonates, the emotional triggers, the comprehension gaps, the misread cues, moderated wins because you need probing.
Win-loss and churn are particularly clear moderated wins. Buyers and churners both have multi-factor reasoning that does not surface in a single video answer. They cite the apparent reason first; the structural reason takes 3-5 follow-up questions. An async survey gets the apparent reason and stops. The moderated interview gets the structural reason, which is what your sales and product teams need to act on.
Common mistakes mixing methods
Three mistakes show up repeatedly when teams pick between async and moderated.
Using async when you needed depth. A team running a concept test ships an async video survey because the unit cost is lower, gets 800 responses, finds 60% prefer concept B over concept A, ships concept B, and finds out at launch that the reason for the preference was a copy line that did not survive the redesign. The depth was the point; the survey did not have it.
Using moderated when you needed reach. A team running a brand tracker decides moderated will give richer data, runs 100 moderated interviews per quarter, spends three times the budget, and discovers the segment-level patterns are not statistically resolvable at that sample size. They needed 1,500 responses per wave, and they bought 100 deep ones instead.
Treating diary studies as moderated interviews. Async video diaries (longitudinal, multiple recordings per participant) are still async surveys methodologically; the participant is recording into the void each time. Some teams treat diary data like moderated interview data and over-interpret what the participant said because there were five recordings instead of one. Five async recordings is still five async recordings, not one moderated conversation.
A fourth, less common mistake: outsourcing the method choice to a vendor without understanding the tradeoff. If a vendor sells only async, every study looks like an async study. If a vendor sells only moderated, every study looks like a moderated study. Pick the method first, then pick the platform.
The hybrid play: async survey to triage, moderated interviews to dig
The most sophisticated programs use both methods in sequence. The pattern: async first to triage at volume, then moderated to dig into the segments or hypotheses that surfaced. The async pass tells you what is happening across thousands of customers; the moderated pass tells you why for the few hundred who matter most.
Concrete example. A consumer brand wants to understand a sales decline in a specific region. Step 1: an async video survey at 2,000 responses across the region’s customers, asking three short prompts about recent purchase behavior, brand consideration, and competitor exposure. The pattern surfaces in 48 hours: a competitor’s new campaign is moving share in 25-34 year-olds. Step 2: a moderated AI video interview program at 100 sessions with 25-34 year-olds in the region, screen-sharing the competitor’s campaign and the brand’s own creative, probing 5-7 layers deep on which messages resonate and why. The async tells you where to look. The moderated tells you what to do.
Another pattern: pre and post a launch. Async tracker before launch (volume, baseline sentiment), moderated interviews two weeks after (depth, what is working and what is not). Or: moderated interviews to develop a concept (50 sessions, deep probing on three directions), async survey to validate at scale (1,500 responses, statistical confirmation across segments).
The hybrid play has economic logic too. Volume work is cheaper per response, so you let it triage. Depth work is more expensive per response, so you reserve it for the questions that need it. The total program cost is lower than either method run at full scale, and the data quality is higher because each method is doing the job it is designed for.
For programs that lean heavily on the moderated side, User Intuition’s video interviews platform handles the depth work, and the AI-moderated interviews methodology overview covers the broader story across voice, chat, and video formats.
What this isn’t: hiring video tools (HireVue, Mercor)
A periodic reminder because the SERP keeps mixing these up. Neither async video surveys nor AI video customer interviews are hiring tools. HireVue, Mercor, and Interviewer.AI screen job candidates for employment decisions. Their AI scores applicants on competencies, generates hiring recommendations, and integrates with applicant tracking systems. The buyers are recruiting and talent acquisition teams. The compliance regime is employment law (EEOC, fair-hiring algorithms, GDPR-employment).
Both methods covered in this guide are for customer research. The participants are existing or prospective customers giving feedback on products, prototypes, concepts, or experiences. The buyers are product, design, marketing, and research teams. The compliance regime is consumer-research and panel-management law (consent, panel disclosure, GDPR-research).
If you arrived at this page from a hiring SERP, the right destination is HireVue, Mercor, or Interviewer.AI. If you are running customer research, you are in the right place; pick async or moderated based on the rest of this guide. The two categories share the words “video” and “interview” and almost nothing else.
Five questions to choose between async video surveys and moderated video interviews
When the methodology choice is genuinely unclear, run through these five questions in order. They resolve almost every case.
- Does the answer require follow-up? If the question is “why did you switch?” or “what made you hesitate?” or “walk me through what you were thinking when you saw that,” the answer requires probing. Moderated wins. If the question is “rate this ad on a 1-7 scale” or “which concept do you prefer,” follow-up is optional, async is in play.
- Do I need to see what they did on a screen? If the stimulus is a live URL, a Figma prototype, a design mockup, or any web asset where on-screen behavior is part of the data, moderated wins. If the stimulus is a static image, a video clip, or no stimulus at all, async is sufficient.
- Is volume more important than depth? If you need 500+ responses for the analysis to be statistically defensible across segments, async wins. If you need 50-500 sessions for the depth to be decision-grade, moderated wins.
- Will the data drive a high-stakes decision? If the output of the study determines whether a product ships, a redesign happens, or a feature gets killed, the cost of being wrong is high and the depth premium is justified. Moderated wins. If the output is a tracker, a sentiment check, or a directional pattern, async is fine.
- Do I have time for 24-48 hours? Async returns results in 1-3 days; moderated in 24-48 hours. Both are fast. If you genuinely need the answer tomorrow, async is marginally faster. If 24-48 hours is acceptable, the moderated depth premium is worth taking.
Three or more “yes” answers to questions 1, 2, or 4 means moderated. Three or more “yes” answers to question 3 or strong “yes” on 5 with a “no” on 1, 2, and 4 means async. Mixed answers mean the hybrid play: async first, moderated second.
For the /compare/voxpopme-vs-user-intuition/ head-to-head, the breakdown is largely a moderated-vs-async-incumbent breakdown rather than two products built for the same job; the comparison page covers where each fits in a program.
Where User Intuition fits in this picture
We sell the moderated AI video customer interview side. We are not the right tool if you need 2,000 async video survey responses for a brand tracker; Voxpopme and similar incumbents are mature in that category and we recommend them honestly. We are the right tool when the methodology you need is moderated: concept testing, prototype work, win-loss, churn, on-screen behavior testing, anything where the why matters more than the what.
What you get from User Intuition specifically: an adaptive AI moderator running 5-7 layer laddering on every session, optional screen-share against any web-accessible asset, replayable video clips synced to verbatim transcripts, recruitment from a 4M+ pre-vetted global panel across 50+ languages, five-layer fraud and identity validation, $20 per Pro audio rate equivalent for moderated interviews, 24-48 hour delivery, 98% study satisfaction, 5/5 G2, and 5/5 Capterra. We pay-for-quality only; sessions that fail any layer of the fraud and identity defense do not count toward usage.
If you have a program that needs both methods, run async first with whichever incumbent you trust, then bring the depth pass to User Intuition. The hybrid play does not require single-vendor consolidation; it requires picking the right tool for each step. For a deeper read on the moderated side specifically, the video customer interviews complete guide and the video customer interviews platform are the two best entry points. For the broader AI-moderated interviews methodology across voice, chat, and video, the platform overview is the canonical reference.
The honest answer to “async or moderated” is “depends on the question.” The honest answer to “where does User Intuition fit” is “on the moderated side, and we will be the first to tell you when async is the right call instead.” Pick the method first, then pick the platform.