The account had been posting three times a week for months. The feed looked consistent. The captions were considered. But nobody on the team could answer a simple question: which posts were actually working?
That question prompted a full audit: 128 posts, five metrics, no assumptions. This is what the data said, and what it forced us to change.
Why bother measuring
Posting without measurement is a habit, not a strategy. It feels productive. It keeps the feed alive. But it optimises for volume, and Instagram does not reward volume. It rewards distribution, which is a different thing entirely.
The Instagram algorithm distributes a post based on signals from the people who see it first. If those people engage, the post reaches more people. If they do not, it stops. So the question is not “how many posts did we publish?” but “which posts did the algorithm choose to push, and why?”
To answer that, you need to know five things about your posts: reach, views, engagement rate, saves, and shares. Each one tells you something different.
The five metrics that changed the strategy
Reach is the number of unique accounts that saw a post. It is the most honest measure of distribution. A post with a thousand likes from the same two hundred followers is not reaching anyone new. Reach tells you whether the algorithm promoted the post beyond your existing audience.
Views counts every time the post was displayed, including repeat views. For reels, views and reach diverge significantly: someone can watch a short reel several times before they decide to follow. For carousels and stills, views and reach tend to be close.
Engagement rate (as a percentage of reach) is the measure most accounts watch. But engagement rate means different things at different reach levels. A post seen by ten people that gets one comment has a 10% engagement rate. That is not the same as a post seen by two hundred people with 10% engagement. When you divide by reach, the two numbers become comparable.
Saves is the metric most accounts ignore and should not. When someone saves a post, they are telling Instagram: this is worth returning to. The algorithm weights saves heavily, because a save is a strong signal of value. A like takes a fraction of a second; a save requires a deliberate decision.
Shares is the most powerful signal available. When someone sends a post to another person, they are vouching for it. Instagram interprets a share as a recommendation. Sends-per-reach is, by most accounts, the strongest distribution signal after watch time on reels.
What the numbers showed
The audit covered 128 posts across three formats: reels (14 posts), carousels (16 posts), and still images (98 posts).
| Format | Posts | Median views | Median reach | Engagement % of reach |
|---|---|---|---|---|
| Reel | 14 | 123.5 | 105.5 | 4.3% |
| Carousel | 16 | 38 | 20 | 19.1% |
| Still image | 98 | 22 | 10 | 23.1% |
The first number that stopped the room: 77% of all posts were still images. The one format with the lowest median reach. The format that, by the evidence of the account’s own data, was reaching a median of ten accounts per post.
Reels achieve roughly ten times the views and reach of stills. Yet the account had posted seven stills for every one reel.
The second number: saves and shares were approximately zero across the entire account. Not low. Zero. No post had been built with saving or sending in mind. The two strongest distribution signals available were completely absent from the strategy.
What the top posts had in common
Three posts stood out above the rest.
The highest-engagement post was a carousel. The topic was Persian literature: a piece of lore, a story, a connection between the jewellery and something older. It reached 275 accounts and held a 24.4% engagement rate.
The second strongest was a reel. A personal story: the founder’s grandmother’s turquoise, where it came from, what it meant. It reached 256 accounts and stayed in the top three for five months.
The highest-reach post ever was a heritage reel: world-building, history, a piece of visual storytelling that had nothing to do with selling a product. It reached 237 accounts in the first 24 hours.
All three had one thing in common: they were about story and heritage, not catalogue products. None of them led with a price, a specification, or a call to action. All three were built to be interesting rather than to convert.
The watch time problem
Reels have one metric that images and carousels do not: watch time. The median watch time across the account’s reels was 3.9 seconds. The top-performing films held attention for 24 to 32 seconds.
That gap is a brief. A reel that holds someone for 25 seconds has earned distribution in a way a reel watched for 3 seconds has not. The first three seconds of every reel need to give a reason to stay. Not a hook in the copywriter’s sense. A genuine reason: something interesting enough that stopping would cost you something.
The two decisions the data forced
Decision one: stop posting catalogue stills as primary content. Stills are not the right format for reaching new people. They belong in the mix, used deliberately, not as a default. A still works when the image itself is the story. It does not work as a product shot with a price in the caption.
Decision two: build one save-target and one send-target post per week. A save-target post is something worth keeping: a guide, a reference, a piece of information that solves a recurring problem. A send-target post is something worth forwarding: a story, a piece of lore, a reel that makes someone think of a friend. Both require different thinking than “what should we post today?”
How to read your own data
If you have not done this for your account, the process takes one evening. Export your post insights from Instagram’s professional dashboard, or build a simple spreadsheet from the native analytics. For each post, record: format, reach, views, engagement count, saves, shares. Calculate engagement as a percentage of reach, not of followers.
Then sort by reach. Sort by saves. Sort by shares. The top of each list will show you what is actually working, stripped of the impression that a post is performing because it received a few comments from people who already follow you.
The account in this audit had been posting consistently for months. The data said: you have been doing the wrong thing, consistently. That is a better finding than no data at all, because it is fixable.
- 77% of posts were still images: the format with the lowest reach by a factor of ten.
- Saves and shares were approximately zero account-wide. These are the two strongest distribution signals available.
- The three top-performing posts all covered story and heritage, not products. None led with a sale.
- Median reel watch time was 3.9 seconds. Top films held 24 to 32 seconds. That gap is a content brief.
- Fix: one save-target and one send-target post per week, minimum.