You spent six months planning a runway show. Shot it from eight angles. Now you're editing a 90-second video to send retail buyers.
Which cut keeps them watching? Where do they zone out? You're guessing.
Facial coding gives you second-by-second data on what buyers actually feel when they watch. Not what they say afterward. What their faces show while it plays.
Why runway video matters more than the live show
Most retail buyers don't attend your show. They watch the video you send.
That 90 seconds decides if they book an appointment. If they don't feel something in the first 15 seconds, they close it.
The live show has music, energy, champagne. The video has their laptop screen at 3pm on a Tuesday. Different emotional context entirely.
What facial coding actually measures
EmotionTrac uses your laptop or phone camera (with permission) to capture micro-expressions while people watch video. It maps those expressions to FACS (Facial Action Coding System) units.
You get a timeline showing joy, surprise, confusion, interest, or disengagement at each second.
It's not reading minds. It's measuring what faces do. Eyebrow raises. Mouth corners. Nose wrinkles. The involuntary stuff that happens before people control their reactions.
Testing three edits of the same runway footage
Let's say you have three cuts of your Fall collection video. Same footage, different edit choices.
Edit A opens with a wide shot of the venue, then cuts to close-ups of the first three looks.
Edit B starts with a model's face in close-up, pulls back to reveal the outfit, then shows the walk.
Edit C leads with your hero piece (the finale coat), then rewinds to show the full collection in order.
You think Edit C is bold. Your creative director prefers Edit B. Your sales team wants Edit A because it's "safe."
Run all three through EmotionTrac with 30 retail buyers (or buyer personas if you're pre-testing internally).
The data might show Edit B loses attention at second 8 when the camera pulls back. Too slow. Edit A gets consistent interest but no surprise peaks. Edit C spikes joy at second 3, holds attention through second 40, then drops.
Now you know Edit C works but needs tightening after 40 seconds. Not a guess. Measured facial responses.
The 15-second test
Before you test full edits, test your opening 15 seconds in isolation.
Show just the first 15 seconds of each version. Watch for the exact moment faces disengage. It's usually between seconds 6 and 12.
If confusion or boredom shows up before second 10, your opening doesn't work. Retail buyers will close the video before they see your best pieces.
This sounds harsh. It is. But you'd rather know before you send it to 200 buyers.
Testing music and voiceover separately
Music changes emotional response more than most teams expect.
Test the same edit with three music options. Classical piano. Electronic beats. No music, just ambient runway sounds.
Facial coding will show which score creates the emotional arc you want. Maybe piano reads as "boring luxury" to younger buyers. Maybe electronic feels try-hard. Maybe silence lets the clothes speak.
Same with voiceover. Test your video with and without narration. If the voiceover says "inspired by 1970s tailoring" while faces show confusion, the narration isn't helping. It's creating cognitive load.
Try it: Schedule an EmotionTrac demo and see second-by-second emotion tracking in action. Or visit Fashion for more information.
Where buyers stop watching (and why it matters)
The drop-off point is more important than average engagement.
If 60% of viewers stop watching at second 47, something at second 47 breaks the experience. Maybe it's a jarring cut. Maybe it's when you show a weak look. Maybe the pacing slows.
Find that moment. Fix it. Test again.
One brand found buyers dropped off when the video showed backstage prep footage. The brand thought it added authenticity. Buyers thought it wasted time. Cut it, retention improved.
Testing for different buyer segments
Department store buyers respond differently than boutique buyers. Menswear buyers respond differently than womenswear buyers.
Segment your test audience. Run the same video past each group. Compare emotional timelines.
You might find your hero piece excites boutique buyers but confuses department store buyers. That tells you which edit to send to which list.
Or you find both groups disengage at the same moment. That's a universal problem. Fix it for everyone.
What to do with the data
You'll get a timeline with emotion intensity at each second. High points, low points, flat sections.
High points: Keep these moments. They're working. If second 23 shows a joy spike, that outfit or that cut or that model's expression is landing. Protect it in future edits.
Low points: These are confusion, boredom, or negative response. Investigate why. Is it the look? The transition? The pacing? Test variations.
Flat sections: Sustained attention without emotion. This might be fine for establishing shots. But if your hero piece gets flat response, you have a problem.
The goal isn't to make every second a joy spike. That's exhausting. You want an arc. Build interest, peak emotion at key moments, let viewers breathe, close strong.
Common mistakes when testing runway videos
Testing only with your internal team. Your team has context buyers don't have. They know the collection story, the inspiration, the backstage drama. Buyers see the video cold.
Testing too late. If you test after you've already sent the video to half your list, you can't fix it. Test before the send.
Ignoring the drop-off data because you love that section. Your opinion doesn't matter here. Faces don't lie about engagement.
Testing with too few people. Five people isn't enough. Aim for 25-50 per segment for reliable patterns.
How one brand cut video length by 30 seconds
A contemporary brand tested their 2-minute runway video. Facial coding showed engagement dropped at 1:15 and never recovered.
Everything after 1:15 was filler. Buyers had already decided. The brand cut those 30 seconds, retested, and saw completion rates improve.
Shorter video, better results. They would have kept the 2-minute version if they'd only asked buyers "what did you think?" Everyone would have said "great."
FAQ: Does this work for lookbook videos too?
Yes. Same process. Lookbooks often have different pacing than runway videos. Slower cuts, more product detail shots. Test whether buyers want to linger on details or move faster through looks. Facial coding shows where patience runs out.
FAQ: What if our buyers don't want to be recorded?
Use panelists who match your buyer demographics. You don't need your actual buyer list for testing. You need people who think like them, buy like them, and respond to fashion content the same way. EmotionTrac works with opt-in panels who consent to facial capture.
Sources
Höfling, T. T. A., & Alpers, G. W. (2023). Validation of facial expression analysis in emotional video content. Frontiers in Neuroscience, 17. DOI: 10.3389/fnins.2023.1125983