Saves are the metric that matters. Likes are the metric you see.

Open the LinkedIn analytics dashboard and you'll see a wall of green numbers. Impressions. Engagement rate. Reactions, broken down by ten flavours of "like." It looks like a measurement system. It is not a measurement system. It is a feedback loop optimised for the platform's revenue, not yours — and the longer you write in response to it, the less the work compounds.

There is a better signal hidden in the same dashboard. It is harder to find. It is much less flattering. And once you've seen it, the rest of the metrics become hard to take seriously.

Why "likes" is the wrong window

A like is the lowest possible cost a human can pay to acknowledge a piece of content. It costs less than a comment, less than a save, less even than the time required to read past the second sentence. It registers a vibe. It does not register that the work mattered.

This wouldn't be a problem if likes correlated with the things you actually care about — getting into a senior reader's head, being remembered when a role opens up, getting cited in someone else's thinking six months later. The problem is that they don't. A post can collect 800 likes and zero opportunities. A post can collect 22 likes and end up being the reason a Distinguished Engineer DMs you in November.

Likes measure how it felt to scroll past your post. Saves measure that someone intends to come back to it. — Field note, internal

The mechanical difference is small. The signal difference is enormous.

The save-to-like ratio, defined

The signal we've found most reliable on senior technical writing is the save-to-like ratio — saves divided by likes, on a per-post basis, then aggregated across a body of work. The numerator is intentional: someone reached past their dopamine reflex and decided they wanted this back later.

For a typical LinkedIn post written by a senior IC, the baseline save-to-like ratio sits around 0.03–0.06. Most posts. Most authors. Most quarters. Above 0.10 the post is genuinely useful to a meaningful slice of its readers. Above 0.20 you've written something a senior reader expects to need again.

What the ratio tells you that the headline number can't

Run the calculation across your last fifty posts and three things tend to fall out:

1. The posts that compound aren't the posts that trended. The high-save posts are often the ones with mediocre like counts — the deep technical breakdowns, the unfashionable opinions held with conviction, the "here is exactly how this went wrong" autopsies.
2. Your authority is concentrated. One or two topics typically carry most of your save weight. The rest is noise. That concentration is the thing your readers actually associate with you, whether you've named it yet or not.
3. The cost of writing more is higher than you think. Each additional low-save post drags your mean ratio down and dilutes the topic concentration above. Cadence is not free.

Write less. Concentrate. Measure the work that gets reopened, not the work that gets scrolled past.

What to do with the signal

Two things, in order. First, audit the body of work you already have. Pull the analytics export. Rank by saves. Read the top decile in one sitting. Whatever theme runs through them — and there is almost always a theme, even when you're sure there isn't — is the authority you've already built. That's your line.

Second, change what you write next. Not the cadence. The composition. Trade three thin posts a week for one post a month that ends up in the top decile of your own corpus. The math on this is unforgiving: a single 0.15-save-ratio post is worth thirty 0.04-ratio posts. Not in vanity. In the only thing that matters — being remembered usefully by senior readers.

None of this requires a tool. You can do it in a spreadsheet in an afternoon. The reason we built LinkedIQ is that almost nobody does. The dashboard you log into shows you the loud number; reading the quiet one is friction. The product makes the quiet number unmissable.

If you want to see what your save-to-like distribution actually looks like, run a free scan. The first five posts are free; the rest is a different conversation.