I got tired of guessing what Amazon rewards. So I spent weeks pulling data on more than 11,000 products, 37,000 Amazon Influencer videos, and 13,000 creators to find out what actually correlates with ranking in the product carousel — and what actually gets people to click and vote helpful. Not opinions. Not what somebody’s coach said in 2023. Numbers. Here’s everything I found.
Table of Contents
- Ranking and Engagement Are Two Different Games
- How Long Should Your Amazon Video Be?
- How Often to Mention the Product
- Horizontal vs Vertical: Which One Wins?
- The Title Formula That Earns Clicks and Helpful Votes
- Do Custom Thumbnails Actually Matter?
- Comparison Videos Are the Sleeper Winner
- The Winning Formula for Your Next Video
Ranking and Engagement Are Two Different Games
Before any of the numbers make sense, you need to understand that there are two separate scoreboards here, and most creators only play one of them.
Ranking is where your video lands in the carousel on the product page. Slot one, slot two, slot five. Engagement is what happens once you’re there — do people click your video, and do they vote it helpful?
A great slot means nothing if nobody clicks. And clicks mean nothing if you never rank in the first place. Some of the things I tested help one and hurt the other, so I’m going to keep them separate as I go.
How Long Should Your Amazon Video Be?
This was the clearest signal in the entire study, and it goes directly against what a lot of people are teaching.
Videos under 30 seconds ranked the worst on average. And as video length went up, ranking position got better — consistently. The best average carousel position, around 2.2, belonged to videos four minutes and up.
Remember there are only five slots in that carousel. Landing at an average of 2.2 is strong. Anything under four minutes ranked progressively worse the shorter it got, and the trend really starts moving in your favor once you clear the two minute mark.
So the short-form-only advice you’ve been hearing? The data doesn’t support it for Amazon shoppable videos. Make it at least two minutes. Three to four is the sweet spot.
How Often to Mention the Product
This one surprised me with how cleanly it lined up.
Videos sitting in the number one slot averaged eight product mentions. Videos in the lowest slot averaged 5.9. That’s a meaningful gap, and it moves in a straight line as you go up the carousel.
I’m not telling you to keyword-stuff your way through a review. But if you’re filming a three minute video and you say the product name twice, you’re leaving something on the table. Say it naturally, say it often, and make it obvious what you’re reviewing.
Horizontal vs Vertical: Which One Wins?
About 78% of all the videos in the dataset are already horizontal. So the raw numbers were always going to favor horizontal just because there’s so much more of it. I had to adjust for that imbalance before the comparison meant anything.
Unadjusted, horizontal took the top slot 82% of the time. After adjusting for how much more common horizontal is, horizontal still wins the number one position 57% of the time.
That’s the honest number, and 57% is still well over half. Horizontal wins.
But here’s the interesting wrinkle. Vertical isn’t getting crushed — it’s landing in the mid-40s. That’s close enough that I suspect Amazon may be deliberately seeding some vertical into the carousel to see how it performs. Worth keeping an eye on. For now, though, if you’re choosing, shoot horizontal.
The Title Formula That Earns Clicks and Helpful Votes
Your title is 100% in your control, costs you nothing, and takes about fifteen seconds to get right. Here’s what the data says.
Use one word in all caps. Titles with an all-caps word averaged 9.5 helpful votes. Titles with no caps dropped to 8.1. One word. That’s the whole change.
Ask a question. Framing your title as a question — “Does This Thing Actually Work?” — showed 9.5 average helpful votes with it versus 8.8 without. That’s about an 8% lift, which is still worth taking.
Use punctuation. Colons, pipes, and question marks all appeared to help. They break the title into scannable chunks and they read as more deliberate.
Aim for 50 to 69 characters. This was a clear sweet spot. The shortest titles, under 30 characters, averaged only 7.2 helpful votes. Worth noting: Amazon warns you once you go past 60 characters, so in practice I keep mine under that line and stay in the upper half of the range.
One caution — not every title trick helped. At least one element I tested came back at roughly 6.9 average helpful votes and actually performed better when it was left out, a 25% swing in the wrong direction. More isn’t automatically better. Pick the elements above and stop there.
Do Custom Thumbnails Actually Matter?
About 61% of videos in the dataset already have a custom thumbnail, so again I had to adjust for that before comparing.
After adjusting, custom thumbnails win the position 53% of the time. Head to head inside the same carousel rank, a custom thumbnail beats an auto frame grab 55% of the time.
So for ranking, it’s a real edge but not a dramatic one — a little over half. Where thumbnails earn their keep is engagement: an 18% increase in helpful votes from using one.
Make the thumbnail. It’s a few minutes of work and it pays on the side of the equation most people ignore.
Comparison Videos Are the Sleeper Winner
If you take one tactical thing away from this study, make it this one.
Averaged across all positions, comparison videos won 66% of the time. But at the top of the carousel it gets dramatic — a 74% edge at slot one and 86% at slot two.
The engagement side is even more lopsided. Comparison videos pulled roughly 4.4 times the helpful votes of non-comparison videos — 30.6 average helpful votes versus about seven. That 30.6 is the single biggest helpful vote number anywhere in this entire study.
Fair disclosure on the sample: this finding comes from 154 comparison videos within a transcript sample of just under 4% of the dataset. Smaller sample than the rest. But the size of the effect is hard to ignore.
Honest review framing also performed well. Review-style videos skewed toward the top of the carousel, taking 39.6% of slot one versus 29.2% of slot five, and earned 87% more helpful votes in the same sample.
The Winning Formula for Your Next Video
Here’s everything above, collapsed into a checklist you can run before your next upload:
- Film horizontal. It wins the top slot 57% of the time after adjustment.
- Go at least two minutes, ideally three to four. Four minute plus videos averaged the best carousel slot in the whole study.
- Name the product often. Slot one videos averaged eight mentions.
- Compare it to alternatives whenever you can. Biggest single lever in the data.
- Write a 50 to 69 character title with one all-caps word, ideally framed as a question, using a colon or pipe.
- Upload a bold custom thumbnail. 18% more helpful votes.
A three to four minute horizontal video that names the product often and compares it to alternatives, published with a bold custom thumbnail and a 50 to 69 character title containing an all-caps word. That’s the formula the data points to.
Here’s the full breakdown with every chart on screen:
Knowing the formula is half of it. The other half is finding products actually worth filming and keeping your storefront clean while you scale. That’s exactly what I built Oink for — Storefront Cross Check, Comparison Video Schedule, and Unavailable Video Matching are all built for this workflow. Grab it at oinkforinfluencers.com.
Let me know in the comments which finding surprised you the most.
