The metrics that matter after you publish

After a post goes out, the network returns a handful of numbers. In EverFeed there are seven: impressions, reach, likes, comments, shares, saves and views. They are kept with history, which already solves half the problem — a lone number says nothing; the same metric over time says a lot.

The other half is knowing what each one means. Let us go through them.

Reach and impressions are not the same thing

This is the most common confusion, and it inverts conclusions.

Reach is how many distinct accounts saw the post. Impressions is how many times it was displayed — the same account can count several times.

The relationship between the two is informative. Impressions well above reach means people came back to the post: it was revisited, or shown again to people who had already seen it. On a carousel or a saved post, that is a quality signal. Impressions roughly equal to reach means each person saw it once and moved on.

Neither measures your content in isolation, because both depend on how much the platform decided to distribute it. That is why they read better as context for the other numbers than as a result.

Likes are the least useful number

Liking is cheap. It costs half a second and no reflection, and so it correlates poorly with almost everything that matters: a like does not say the person read it, remembered it or would come back.

Likes are good for one thing: relative comparison between similar posts, in the same period, on the same account. Beyond that, it is the number that goes up when you post a dog.

Saves are the best signal of useful content

Saving requires the person to recognise future value in what they saw. It is a higher cost than a like and a clearer intent: “I am going to need this later”.

Educational posts, recipes, checklists, comparisons, tutorials — the number that validates those formats is saves, not likes. If you produce useful content and only look at likes, you will conclude it does not work and stop doing the thing that was working.

Shares are the best signal of future reach

Sharing is the highest social cost: the person is putting their reputation next to your content. It is also the gesture that most feeds distribution, because it carries the post beyond your audience.

A post with many shares and few likes is often the most valuable of the month, and the easiest to discard by mistake if you sort your feed by likes.

Comments measure conversation, not approval

Comments are a strong signal, but an ambiguous one: the same metric rises when you write something excellent and when you write something irritating. A high comment count without matching growth in saves and shares deserves a read before you celebrate.

What matters more than the count is the nature: a question is a request for information the piece was missing — and every repeated question is a ready-made topic for the following week.

Views, on video, depend on the platform definition

This is where cross-network comparison falls apart. What counts as a “view” varies — some platforms count a few seconds, others require more, and the rules change over time.

So: compare views from the same network with the same network, over time. Comparing views on one platform with another is comparing two different definitions that share a name.

A zeroed collection almost never means nobody saw it

This is the point that generates the most wrong conclusions, and it deserves the longest section.

When a metric comes back zeroed, the most likely explanation is not that the post got zero of anything. It is that the platform did not return that number. The common reasons:

• It is too early. Many metrics only become available after a delay. Collecting five minutes after publishing brings back legitimate zeros.
• The metric does not exist for that post type. Not every format exposes every number.
• The account does not meet the minimum. Some platforms hide certain data from small accounts, for privacy.
• The insights permission was not granted. Publishing and reading metrics are different permissions; you can have one without the other.
• The token expired between publishing and collecting. Common with long-lived tokens past their 60 days.

That is why collection in EverFeed is a separate job that you fire when you like and that keeps history: a zero on the first collection and a number on the second is the normal situation, not a bug. If you treat the first reading as truth, you will file as a failure a post that has not even been distributed yet.

The three readings worth doing

With seven numbers and history, three comparisons give you almost all the value:

1. Saves per reach, across formats. Divides useful value by distribution and answers which format deserves more of your week — carousel, reel or single post.

2. The same piece, two collections. The curve matters more than the peak. A post that keeps accumulating reach after three days has a long life; one that stops on day two is immediate consumption. Those are different strategies, not different qualities.

3. Shares by subject. Group pieces by the kit field that originated the topic — persona, offer, differentiator. The subject that generates shares is the one that should become more topics next month, and that link only exists if topics record where they came from.

What none of these metrics measure

Sales. None of the numbers on the list knows whether anyone walked into the shop, called or bought.

Worth saying because the temptation to optimise what is easy to measure is strong, and it leads whole accounts to produce content that performs well and sells nothing. Network metrics are for choosing between two similar pieces and for discovering which format works. They do not replace asking the customer how they found you.

Anyone wanting to close that loop needs tagged links, a coupon per piece or the question asked at the counter. It is manual work, it is tedious, and it is the only one that answers the question that matters.