How to Build Research Highlight Reels Stakeholders Trust
Learn how to turn consumer interview footage into research highlight reels that show stakeholders the evidence, from clip selection to consent.
A reel only moves a decision when the room believes it. Senior decision-makers have sat through enough well-chosen moments that they now ask how the clips were selected and how anyone checked them. This guide covers four rules for research highlight reels, where they sit in a readout, and what a researcher checks when AI produces the first cut.
Why stakeholders doubt a highlight reel
Pushback in a readout usually arrives as one of four objections, and each one points to something the reel left out.
Cherry-picking comes first. Four articulate consumers who all agree with the finding look like the best four minutes out of thirty hours, and a skeptical brand director will assume the other hours said something else.
Then there is prevalence. A vivid clip shows that one consumer felt this way. It does not show whether that was one person in thirty or twenty, and a CMO weighing a repositioning needs that number before the clip means anything.
Context is the third gap. A consumer who says she would never pay that much may have been talking about a competitor’s price, or may have changed her mind two questions later. A twelve-second cut can remove the sentence that changes the meaning.
The last objection is that nobody can check the source. If nobody outside the insights team can open the full interview, the room has to take the clip on trust, and earning that trust was the reel’s job in the first place.
Four rules for a research highlight reel stakeholders trust
Each rule below answers one of those objections. Together they turn a reel into evidence for a claim.
One finding per reel
A reel that mixes three themes cannot be defended, because a challenge to one clip unsettles the other two. Build one reel per finding and state the finding on the title card as a plain claim. In a churn study, for example, the card might say that trial buyers cancel after month two because the refill schedule does not match how fast they use the product. Every clip then has one job, to support or complicate that sentence, and the editor cuts any clip that does neither. A concept test with three findings gets three short reels, so each part of the readout discussion rests on one claim and one set of evidence.
Show how many consumers said it
Put the count on the title card next to the claim, for example 14 of 30 consumers. With the count, the room reads the clip as one instance of a pattern. Without it, the clip looks like the most quotable moment in the study. This guide recommends a count instead of a percentage at this scale, because 47% of thirty interviews suggests a precision the numbers cannot carry.
Include a clip that disagrees
Add one clip from a consumer who saw it differently. Place it near the end and give it its own count, such as the number of consumers who preferred the current pack. Showing the dissent openly answers the cherry-picking objection before anyone raises it. The contrast clip also needs a line on its title card saying what the dissent represents, such as which segment it came from, so the room reads it as part of the finding.
Make every clip traceable to its interview
One link per clip answers both the context objection and the source objection. Each clip links to the full session and the point where it starts, so people in the readout can inspect the surrounding minute instead of trusting whoever edited the reel.
Consent and redaction happen at the same step. The MRS Code of Conduct 2023 requires that participants be told about any recording at recruitment and again at the start of the interview, and that, where consent is the legal basis, the consent information covers who is likely to have access to the recording. Before a clip goes to brand or marketing teams, the researcher checks that the consent covers internal sharing. If it does not, the researcher blurs faces and removes names spoken on camera or shown on screen, and leaves out any clip where redaction cannot adequately protect the consumer’s identity.
Where the reel sits in a readout
A reel belongs next to the written finding it supports, either embedded in the slide that states the claim or sent along with the summary that contains it. A written claim on its own gives the room a conclusion and asks them to accept it. A pulled quote adds one consumer’s words but still leaves the count and the context unstated. The reel puts the consumer’s voice behind the same claim, and its title card carries the count, so the claim, the quote and the clip work together as one argument.
Placement also sets the length. A reel shown in the meeting has to end with time left for the discussion it is meant to start. For a meeting reel, this guide recommends one to three minutes and four to six clips. A longer cut can go in the follow-up email, linked from the summary.
AI-generated vs hand-cut highlight reels
Cutting a reel by hand means searching hours of footage, marking moments, trimming them and stitching them together, and the work starts over for each finding. With an AI-generated reel, the researcher reviews a draft instead of cutting one.
What AI does
AI tools group clips by theme as interviews finish, link each clip to the session it came from, and assemble the theme-grouped clips into a draft reel that stays editable and shareable. For studies with dozens of consumer interviews, AI produces a draft reel for each finding, and the researcher reviews the selected clips rather than manually searching, marking, trimming and stitching every cut. The researcher spends that review time deciding whether each selected clip supports the finding and carries enough context to stand on its own.
What a researcher still checks
An AI-built reel is a first draft, and the four rules apply to it the same way they apply to a hand-cut reel. AI can miss, misread or overstate what a consumer meant, so a researcher reviews these six points and signs off on all of them before the reel reaches anyone outside the insights team:
- Fit: Each clip supports the finding on the title card. A clip can sound related to a theme while supporting a different interpretation, so the researcher removes clips that do not support the title-card finding.
- Context: The surrounding minute does not change what the clip means. The researcher opens the source session for any clip that sounds stronger than the finding.
- Speaker: The words in the clip are the consumer’s, not the moderator’s question.
- Count: The number on the title card matches the coded interviews across the whole study, not just the clips in the reel.
- Contrast: At least one clip shows a consumer who disagreed, with its own count.
- Redaction: Faces, voices and names are handled the way the consent requires.
Highlight reels on Strella
On Strella, reels come from AI-moderated interviews. Strella generates video highlight reels by theme as each interview finishes.
Every clip and verbatim quote links back to the session it came from, so the context check takes one click. Researchers can edit reels and share them with the team.
Chat With Your Research lets anyone on the team ask a plain-language question across selected sessions or the whole repository, and the answers come back with video clips. A brand lead preparing a readout can ask what consumers said about a new price point and save the clips from the answer as a reel. Video blurring and voice and name anonymization are available for any study where consent calls for them. To hear how the moderator follows up on what a participant says, try an interview.
FAQ
How long should a research highlight reel be?
For a meeting reel, this guide recommends one to three minutes and four to six clips, including one contrast clip. That length fits inside the discussion of a single finding. Longer cuts can go in a linked appendix for anyone who wants every clip.
How many interviews should support a finding before it gets a reel?
A finding earns a reel when it comes up across several consumers, not when one consumer says it memorably. There is no fixed threshold. A 2022 review of saturation studies by Hennink and Kaiser found that studies with fairly similar participants and narrow objectives stopped turning up new themes within 9 to 17 interviews. This guide recommends that a theme raised by only one or two consumers in a study of that size go in the appendix as a quote rather than into a reel.
Do consumers need to consent before clips reach leadership teams?
Yes. Consumers should know before the interview starts that it is being recorded and which teams are likely to see the recording. The consent wording should name internal sharing directly, for example that clips may be shown to the brand’s marketing and leadership teams. Where consent does not cover identifiable sharing, faces should be blurred and voices and names anonymized before a clip leaves the insights team, and a clip stays out of the reel if redaction cannot adequately protect the consumer’s identity.
Bring your research question to us and we will talk it through. See Strella in action:

