Cullen vs Reelyzer: sorting Reels by views finds the biggest post, not the best idea.

What Reelyzer gets right

Reelyzer is unusually generous and unusually clean. There is no sign-up wall on the free tier, the first 25 Reels of any profile you can view are free to analyse, MP4 download is free rather than a paid upsell, and CSV and Excel export work on both tiers. The posting-time heatmap and the duration view are useful, honest summaries of the metadata. It does one platform and does it tidily, and it reads through your own browser session rather than asking for your Instagram credentials.

Reelyzer facts from reelyzer.com (home page and its plan comparison table; reelyzer.com/pricing returns 404) and its Chrome Web Store listing, checked August 2026. If anything here is out of date, tell us and we will correct it.

Engagement rate is still an absolute number. It just looks relative.

This is the trap worth naming, because engagement rate feels like a normalized metric. It is not normalized against anything on the account's own timeline. Divide by views or by followers and you still get one flat league table where a big account's routine Tuesday outranks the post that actually broke something open. Nothing in Reelyzer's published feature set names a baseline, a median or any per-account normalization, so there is no way for a Reel to be marked as unusual for this account at this moment, which is the only definition of a winner that transfers.

Cullen's question is narrower and far more useful: did this post beat the posts immediately around it in this account's own timeline? Because those neighbouring posts shared roughly the same follower count, follower size cancels out. A post that did 38 times its account's normal numbers gets flagged as a breakout even if it sits 200th on the likes list. And the rare post that did not just spike but permanently lifted the account's baseline afterward gets its own label, the level-up, which is usually the most important post on the whole channel.

Twenty-five Reels is not enough history to know what normal looks like.

Reelyzer's free tier caps at 25 Reels per profile and 3 analyses a day. That is a fair free tier, and it is also structurally too shallow to establish what an account's normal actually is, let alone to reach back into the earlier, smaller-numbers era where the genuinely instructive breakouts usually hide. Cullen captures up to 500 posts in a single scan with no manual scrolling, which is what makes a per-post baseline possible rather than a guess.

It also covers all three short-form platforms today. If your competitive set lives on TikTok and Shorts as well as Reels, and most short-form creators' does, a single-platform tool means running the research three times in three places, or not running it at all.

Duration and posting time are correlations. A quoted comment is evidence.

The closest Reelyzer gets to why is telling you the ideal Reel length and the best time to post. Those are properties of the metadata around the video, not of the video. They can tell you a 14-second Reel at 7pm did well; they cannot tell you it did well because the transition visibly failed twice before it landed and the comment section filled with people quoting the fail.

The honest gap: Reelyzer downloads and exports, and Cullen does not. Free MP4 downloads and CSV or Excel export on every tier are genuinely good, and if a swipe file of video files and a spreadsheet is the deliverable, Reelyzer wins that outright.