Understanding your Instagram analytics

Reach, impressions, accounts reached: don't mix them up

Instagram separates "accounts reached" (the number of unique people who saw your content) from "impressions" (the total number of views, where the same person can count several times). Confusing the two distorts everything: high impressions with few accounts reached means the same people keep coming back, not that you're reaching wider.

The metric to watch first is the share of reach coming from non-followers. It tells you whether the algorithm is pushing your content beyond your current audience — the real engine of growth. A healthy account regularly sees 40 to 70% of its Reels reach come from non-followers.

• Accounts reached: unique people — your real audience indicator.
• Impressions: total views — useful for measuring repetition.
• Non-follower share: the signal of discovery and growth.

Engagement rate, but relative to reach

Many people calculate engagement against their follower count. That's misleading: on Instagram in 2026, content is distributed by interest, not by following. The relevant formula is (interactions ÷ accounts reached) × 100. It tells you what share of people who saw the post reacted.

An engagement-on-reach rate of 5 to 10% is solid. Below 2%, your content is seen but sparks nothing: the format or the hook needs rethinking. Above 10%, you have a post worth replicating — analyze what made it special.

Never look at likes alone. Shares and saves weigh far more in the algorithm: they signal lasting value. A heavily saved but lightly liked post is often a future evergreen worth expanding on.

Reels retention: the metric that changes everything

For Reels, the retention curve is the richest insight. It shows at which second people drop off. The most critical fall is in the first 3 seconds: if you lose half your audience there, the algorithm will stop distributing, no matter how good the rest is.

Also watch the replay rate and the average watch time relative to total length. A 20-second Reel watched 15 seconds on average performs better than a 60-second Reel watched 25 seconds, even if the second absolute value is higher.

• Retention at 3s: is your hook working?
• Average watch ÷ total length: viewing "completeness".
• Replay rate: sign of dense content or a successful loop.

Follower growth: read the net, not the gross

The number "+120 followers this week" means nothing on its own. Instagram gives you both follows AND unfollows. A weak net gain despite strong reach often reveals a mismatch: your content attracts curious viewers who don't subscribe, or your profile fails to convert visitors.

Cross profile reach (visits to your page) with new followers to get a profile conversion rate. If it's low, the problem isn't your content but your bio, your photo or your feed's first impression — the first screen a visitor discovers.

Choosing your posting times from data, not myths

Generic "best times to post" circulate everywhere, but your audience has its own rhythm. Your insights show the days and slots when your followers are most active. Publish 30 to 60 minutes before the peak, giving the algorithm time to test your content.

Be careful though: this data only covers your existing followers, while a large share of your reach comes from non-followers. Use it as a starting point, then test over three or four weeks and let your own results decide.

Turning numbers into recurring decisions

A one-off analysis is useless. What matters is the trend: track 4 to 5 metrics each week in a simple table to spot underlying movements rather than the noise of a single post.

Gathering these numbers by hand is time-consuming, especially if you also post on other networks. That's where a unified dashboard helps: in CreatorBag, analytics from all your networks land in one place, with week-over-week evolution, so you can compare formats without juggling apps.

• Track the 4-week trend, not one post's performance.
• Always compare a format to itself (Reel vs Reel).
• Write down one hypothesis to test each week and check it.