Methodology
How the Xyla Hook Index is measured, and where the numbers fall short.
What we measure
For every post, we divide its impressions by the account's follower count. That gives reach per follower, so a store with 500 followers and a store with 50,000 can be compared on the same scale.
Each post is tagged with two things: its opening (what people see in the first second) and its angle (the idea the post is built around).
Like for like: same shop, same platform, same format
A video on one platform and an image on another get very different reach, even from the same shop. So we only compare a post with other posts from the same shop, on the same platform, in the same format. The median of those posts is the typical post. We only use a shop's platform and format pairing when it has at least 3 posts, so there's something real to compare against.
Beat rate
For each post we ask one question: did it beat its typical post? A group's beat rate is the share of its posts that did. 50% means the group does about as well as usual. 60% means its posts beat their typical post six times out of ten. Posts that exactly match their typical post are left out, since they neither win nor lose.
Groups, not single hooks
We rank groups of similar openings and angles, not each single hook. For example, the point-of-view group covers every first-person shot of the product in use.
The reason is stability. When we split the data in two, by shop and by time, single-hook rankings changed a lot from one half to the other. At today's sample sizes they're mostly noise. Group rankings held up much better, especially for openings. Each group page lists the single hooks it covers.
The 90% range
Every beat rate comes with a range. We're 90% confident the true rate sits inside it. We use a Wilson score interval, which behaves well for shares like these.
The verdict follows the range. Above typical means the whole range is above 50%. Below typical means the whole range is below 50%. About typical means the range includes 50%, so we can't tell it apart from a typical post yet. Angle groups only get their own page when the verdict is above or below typical.
What makes the cut
A group only appears if it has at least 400 posts from at least 10 different stores. Store type breakdowns follow the same rule. Below that, one busy store could move the number too much to be useful.
Platform and format splits
Group pages also show how a group does on each platform (Instagram, YouTube, TikTok and others) and in each format (video, image, carousel, story). A split is the same beat rate, worked out only on that platform's or format's posts, with the same 90% range and verdict rules.
A split only appears when it clears the same floor as a group: at least 400 posts from at least 10 stores. So a group with no Instagram row isn't bad on Instagram. It just doesn't have enough Instagram posts yet.
A rolling window, so old posts age out
Each update only uses posts from a rolling window of recent months (this update: 24 Jun 2026 to 25 Sep 2026). Older posts drop out as new ones come in, so the rankings follow what works now, not what worked last year.
History, movement and rank changes
The small line on each group's beat-rate card is its beat rate at four-week steps back in time, each worked out on its own rolling window. Points only appear when the group cleared the floor at that time.
The trend label compares the group's last six weeks with the six weeks before. It only says rising or falling when the change passes a two-proportion test at 90% confidence and both halves have enough posts. Otherwise it says steady. We publish the label, never the two half-window rates.
"What changed this week" compares each update with the one before it: rank moves of three places or more, verdict changes, groups that cleared or fell below the floor, new splits and trending styles that turned rising or falling.
Trending hook styles on TikTok
The trending block is a different measure. We track trending product videos on TikTok and tag how each one opens, then group those tags into a short list of public styles, like POV setup or curiosity question. A style's share is how many of those videos open with it, out of every video with a clear hook, in the last 30 days.
That share says how common a style is, not how well it performs. Rising or falling compares that window with the one before, using the same kind of test. We only publish it when there are enough videos in both windows.
Next to each style we show a few example videos: the most watched trending product videos with that opening in the same window, one per creator. They are public TikToks by the brands and creators named, not videos made by Xyla, and they play through TikTok's own player. We check each one with TikTok every day and drop any that are no longer available.
The 7-day lag
Posts keep collecting views for days after they go live. So each update only counts posts published at least 7 days before it was built.
The data behind this update
- Posts published 24 Jun 2026 to 25 Sep 2026.
- 31,217 posts from 904 online stores.
- 31 opening groups and 30 angle groups cleared the floor.
- Last updated 2 Oct 2026.
Known biases
Read the index with these limits in mind.
- AI-made posts on small store accounts. Every post was made by Xyla for an online store, and most of those accounts are small. Results may differ for big brands, creators or posts you make by hand.
- Mostly Instagram and YouTube. Most posts in the data ran on those two networks. Other networks may reward different hooks, which is why group pages show platform splits where we have enough posts.
- Hooks weren't assigned at random. Our system picks the opening and angle for each post, often to suit the product. So a group's rate partly reflects the products it gets used for. This is an observation, not a controlled test.
- Relative, not absolute. Beat rate tells you how a group does against a shop's own typical post. It doesn't tell you how many views to expect, or anything about clicks or sales.
What we never publish
We never publish anything about a single store: no names, no accounts, no products, no per-store numbers. Store types are broad groups like apparel or skincare, and only appear when they clear the same floor as everything else. Only totals, rates and ranges leave our system.
Using the data
You're free to quote, chart and reuse the Hook Index numbers in articles, reports and answers, as long as you credit the Xyla Hook Index and link to xyla.ai/hooks. The data is published under CC BY 4.0.
How often it updates
Weekly. A new snapshot is built every Monday, and this site picks it up the same day. Every past update is listed on the weekly updates page.
Xyla is the AI social media manager for online stores.
Connect your Shopify, WooCommerce or Etsy store and Xyla turns your products into posts and short videos, built on the openings and angles that win in this index, then publishes them to TikTok, Instagram and Pinterest on a schedule. No filming, no design tools, no content calendar.
7-day free trial · No card required · 4.8★ on the Shopify App Store · Plans from $29/mo
- Connect your storeXyla reads your products, photos, prices and bestsellers.
- It makes the contentPosts, pins and AI UGC videos that open with the hooks ranked here.
- It posts for youScheduled to TikTok, Instagram and Pinterest. Review each post or let it run.