Rella just released its 2026 social media data from more than 100,000 posts, and one finding clears up something that confuses almost everyone.
Reels win reach. Carousels win engagement.

This year Reels passed single images as the most-posted format on Instagram, growing from 37% to 45% of feed posts. And 65% of marketers said short-form video performed best for them.
But look at engagement. Carousels average a 7.9% engagement rate, compared with 4.4% for Reels. That’s almost twice as high.

Here’s the easiest way to think about it:
Reels = impressions. They put you in front of new people. Absolutely keep making them.
Carousels = engagement. Every swipe is someone choosing to interact with your post. One swipe means that person engaged.
So you don’t have to choose between them. Plan both: Reels to get discovered, carousels to build connection.
How to build a carousel people actually swipe through
Slide 1: Ask a question or start an idea. Pose an open-ended question, tease a tip, or name the problem. Your only job here is to earn the first swipe.
Slide 2: Put your most helpful point here. Lead with your best tip, so the person who swiped gets a reward right away.
Slides 3 and on: Go in order of importance. Let’s be honest, people get bored. Put the essentials up front and the nice-to-knows toward the end.
Last slide: Make it easy to act. Ask them to save it, share it, comment, or check the link in bio.

Your challenge this week: take your best-performing Reel and turn it into a carousel. Same message, different format, double the opportunities.
Talk soon,
Alyssa
Data: Rella 2026 Social Media Report, platform data through Oct 31, 2026.

The voice-to-text app that built its own speech model
Most dictation apps are a wrapper around someone else's speech model. Wispr Flow trained its own.
It's called Canto, and it was built for the way people actually talk to AI: thinking out loud, changing direction mid-sentence, half-whispering in an open office.
Wispr tested it on 10 hours of real dictations from more than 2,300 speakers. Canto had the lowest word error rate of every model tested, including models from Google, OpenAI, AssemblyAI and Deepgram. That means your prompts to ChatGPT, Claude and Cursor come through the way you said them.
Talk naturally in any app and Wispr Flow types clean, formatted text, 4x faster than typing.


