You open LinkedIn, X, or YouTube meaning to check one thing, and ten minutes later you close it not having found it. That feeling is easy to notice and hard to explain. Was the feed actually bad, or were you just in a mood? The honest way to answer that isn't a screen-time app that tells you how long you scrolled. It's a two-minute count of what you scrolled past.
How do you actually run the audit?
Open the feed you use most, scroll for two minutes without liking, clicking, or opening anything, then go back through what you saw and count: out of the first fifteen or twenty posts, how many were from people you actually follow, and how many were ads, suggestions, or something you didn't ask for?
That's the whole test. No extension, no dashboard, no login required — just a stopwatch and a bit of honesty about what you're actually looking at. The reason it works is that it turns a vague feeling ("this feed feels off lately") into a number you can compare week to week. A feeling is easy to dismiss. A number that says six of your last fifteen posts were things you never asked to see is harder to argue with.
What counts as clutter, and what doesn't?
A post from someone you follow doesn't count, even if you don't like what they wrote. Clutter is anything the platform inserted on its own: promoted posts, "people you may know," suggested videos, and content that showed up because an algorithm guessed you'd engage with it, not because you asked for it.
Ads are the easy part, because platforms are required to label them. The harder part is everything that isn't labeled but still isn't something you chose. I wrote a longer breakdown of this for LinkedIn specifically — what actually clutters a LinkedIn feed in 2026 — and the short version is that the unlabeled stuff usually outnumbers the labeled stuff by a wide margin. When you're counting during the audit, count both. If a post exists because you followed someone, it's not clutter, no matter the topic. If it exists because a recommendation system decided you might like it, it counts, whether it's an obvious ad or not.
- Open the feed. Don't refresh it first — you want what it actually shows you, not a curated re-load.
- Scroll for two minutes, hands off everything except the scroll itself.
- Scroll back up and count the posts you saw, one by one.
- For each one, ask: did I choose this, by following the person or channel, or did the platform choose it for me?
- Divide the second number by the total. That's your clutter ratio for today.
What does the number actually tell you?
One or two unwanted posts out of fifteen is a normal feed — every platform inserts something. Four or more, and the feed has quietly shifted from showing you what you follow to showing you what keeps you scrolling, which are not the same goal. Run it again on a different day before deciding, since one bad scroll can just be a bad scroll.
The number also tells you where the problem is. If most of what you counted were unlabeled suggestions and "you might like" modules, that's a different fix than if it was mostly labeled ads, and different again if the volume of AI-generated posts is what's crowding out the people you actually follow. That distinction matters, because AI content in a feed isn't a quality problem — plenty of it is fine. It's a volume problem: when there's enough of it, you start scrolling past real posts from real people without noticing them, which is exactly what the audit is built to catch.
What do you do once you know the number?
If the audit turns up mostly ads and algorithmic suggestions, a filter that hides those two categories fixes most of it directly. If it's specific topics or accounts, a keyword block handles that. If it's the sheer volume of AI content burying the people you follow, giving the feed a scheduled break from that specific kind of noise brings the rest back into view.
That's the same audit that shaped how Zenfeed's toggles are split up: hide promoted, hide suggested, block by keyword, and Rest from AI as a timer rather than a permanent switch, so it comes back on its own. On LinkedIn and X all four work today. On YouTube, Zenfeed does less on purpose — Rest from AI, keyword blocking, and an opt-in Shorts-shelf toggle, on the Home, Subscriptions, and search pages, with the promoted and suggested toggles wired but inert there since YouTube's ad and recommendation surfaces don't map cleanly onto the same rules.
Turn on "Hide promoted" and "Hide suggested." That alone accounts for most bad audit numbers on LinkedIn and X.
Add them to your custom keyword list. It matches case-insensitively across every post the scan sees.
Turn on Rest from AI for a set number of minutes. It's a break from the noise, not a judgment on any single post.
On LinkedIn, "Hide company pages" filters posts by the byline, not the text, so a company mentioned in a real post stays visible.
Worth saying plainly: the audit isn't a Zenfeed pitch. It works the same whether you install anything or not, and running it once a month, even with no tool at all, will tell you more about your feed than a week of vague "I've been on this app too much" guilt ever does. What you do with the number is up to you.