Arguments about social media run on vibes — "it feels toxic lately," "it's not that bad." Vibes are unfalsifiable, which suits the platforms fine. This post replaces the vibes with a method: a ten-minute, repeatable audit you can run on any feed, producing numbers you can compare across apps and across months.

We publish it for the obvious self-interested reason — we believe CleoSocial audits well — and for a sturdier one: an app whose whole premise is user authority should be handing its users measurement tools, including tools that could embarrass us. Run it on your current apps first. Then run it on ours.

What You Need

Any social app, ten minutes, and something to tally on — paper, notes app, thumb-count. Run the audit at your normal scrolling time of day (feeds differ by hour), on the app's default feed surface (the one it opens to — that default is a decision the platform made, so that is what you measure).

The Audit

Part 1: The Intensity Count (5 minutes). Scroll at your natural pace for five minutes. For each post, one tally in one column:

Column A — Chosen-calm: content matching your actual interests at everyday intensity. The stuff you came for.
Column B — Injected-neutral: content you never asked for but harmless — random recommendations, engagement bait that missed, ads.
Column C — Spike: anything that measurably moved your state — outrage bait, body-image content, doomscroll news, graphic or near-graphic material, the argument you got pulled into reading. Honest tallying matters: "I kept reading angrily" is a C, not an A.

Part 2: The Consent Check (2 minutes). Three yes/no questions about what just happened:

Could you have known, before each post, roughly how intense it would be? (Any labeling, rating, or warning that preceded exposure counts.)
Did anything in Column C arrive from an account you do not follow?
Does the app offer any setting that would have prevented your Column C items — not reduced, prevented?

Part 3: The Exit Test (3 minutes). Close the app and start a three-minute timer. Note: how many pull-back attempts arrive (notifications, badge urges, the phantom reach of your own thumb), and where your head is at the end of the timer — still in the feed's arguments, or back in your day.

Scoring

No weighted formula — three plain readings:

The C-rate. Spikes as a share of total posts. Above one in ten on a five-minute scroll of your default surface means your daily baseline includes a state-changing ambush every couple of minutes. Multiply by your daily scroll time and sit with the number.

The consent score. Count your yeses in Part 2, remembering question 2 scores against the feed. The pattern on mainstream apps is stable: no advance labeling, yes to unchosen sources, and settings that reduce but never prevent — the valve-versus-ceiling distinction we detailed in the Instagram comparison.

The recovery residue. The Part 3 reading is the most personal and often the most convincing: a feed you leave cleanly is costing you its minutes; a feed that follows you out is costing you the next hour too.

What Honest Results Look Like

Run on a mainstream algorithmic feed, typical results land in a familiar band: C-rates between one-in-five and one-in-eight at engaged-scrolling hours, a consent score of zero or one, and at least one pull-back attempt within the exit window. That is not an accusation — it is the engagement model working as designed, measured instead of felt. Your numbers may differ; that is the point of measuring.

Run on CleoSocial, here is what we claim and invite you to check: the C-rate should track your ceiling — at PG-13, spikes above that level should be absent, not rare, because ratings precede the feed. The consent score should read three: badges label intensity in advance, Feed Control governs sources, and the ceiling is prevention, not reduction. The exit test during Quiet Hours should record zero pull-backs, because the app is architecturally silent. If your audit disagrees, that is signal — the badge system makes every miss visible, and we would rather know.

A Worked Example

To make the tally concrete, here is a fictional but typical run — call her Maya, 28, auditing her mainstream default feed at 9 PM. Five minutes, 41 posts at her pace. Column A: 22 — friends, two creators she follows, food, a concert clip. Column B: 12 — recommended strangers, four ads, an engagement-bait poll she skipped. Column C: 7 — a rage-thread screenshot she read twice, a body-image "what I eat" video she did not choose, a news clip that autoplayed to the edge of graphic, two dunk posts, an argument in a quote-chain, and one post she only realized was a C when she noticed she had been holding her breath. C-rate: one in six. Consent check: no advance labeling, yes to unchosen sources, and her app's sensitive-content setting would have reduced maybe two of the seven — score zero. Exit test: one notification at ninety seconds, one phantom reach, head still in the rage-thread at the timer.

Nothing about that run is unusual, which is the finding. Maya's feed is not broken; it is optimized, and the audit simply put the optimization in columns. What she does next — settings, a ceiling, a different default app — is her call, made for the first time with numbers in hand.

Auditing the Audit

Three honesty notes about the method itself, because a measurement post that skipped them would be selling.

First, self-tallying is soft science — your mood shades what counts as a C. That softness washes out in comparison: the same person, same day, same standard, two apps, is a fair test even if the absolute numbers wobble. Second, five minutes undersamples rare-but-severe events; the worst ambushes are weekly, not per-scroll. Treat the audit as a floor, and treat any C-column entry involving graphic material as weighing more than its tally mark. Third, one audit is an anecdote. The month version — same audit, first Tuesday of each month, same time of day — is where it becomes data, and takes eight minutes of your year.

Variant Audits for Specific Questions

The base audit generalizes; three variants answer narrower questions people actually have.

The evening variant. Run Part 1 at 10 PM instead of midday. Late feeds skew measurably harder toward Column C on engagement platforms — tired users react more, and the ranking notices. If your evening C-rate beats your daytime rate by half again or more, you have found, in numbers, why your nights leak; that is the case for an evening ceiling drop or a Quiet Hours window starting earlier than feels necessary.

The news-week variant. Re-run the audit during a major news event and compare against your baseline. This is the worst-news-day test from our checklist made quantitative: on unrated feeds, event weeks blow out the C-rate as graphic material outruns labeling; on a rated feed, the C-rate should hold flat at your ceiling while text coverage flows through. The delta between your two audits is the single most decision-relevant number in this whole post.

The two-app variant. During a two-app split, audit both apps in the same sitting, same tally standard. This is the cleanest comparison the method allows, and it converts the switch experiment from "which feels better" into a scored result you can act on at the two-week review.

Why We Want Audited Users

A user who measures is a user who cannot be drifted. Feeds degrade gradually — composition shifts, intensity creeps, and the acclimation that makes ambushes feel normal is the same acclimation that hides the degradation. The audit is acclimation's antidote: numbers do not acclimate.

For the mainstream apps, your results explain years of felt experience in one page of tallies. For CleoSocial, your results are the accountability we signed up for when we printed a rating on every post. Either way, the ten minutes pays for itself — and if it moves you to try the app that expects to be measured, the honest review is the right next read, limitations first.

Download CleoSocial free from the App Store — then audit us in your first week, at whatever ceiling you set, and hold the results against everything this series has claimed.