Launches audited acrossevery vertical

A launch enters the RADAR corpus when it is a public product or company launch post whose engagement metrics are public and complete enough to read. RADAR records views, likes, reposts, replies, and quotes for each one, then re-pulls those figures after the launch has settled, so the 30 launches in the corpus were all measured the same way rather than screenshotted on launch day and left there.

Three rules decide whether a launch is in. First, the post has to be the launch itself, put out by the product account or the founder, not a recap thread or a press pickup. Second, every figure has to be a matured one. A like count read in the first hour is still climbing while the impressions have already run, and reading the two together manufactures a views-per-like gap that says nothing about how the reach was built. Third, nothing is modelled. Where a metric is not published, the launch stays out of the corpus instead of being estimated into it.

The corpus currently spans launches from October 2025 to June 2026. Per our own record of those posts it carries 110.4M+ views of launch reach, sitting on 145,493 likes, 20,088 reposts, 23,152 replies, and 9,642 quotes. Those four engagement layers are what a reading is built from, which is why a launch with impressive reach and no published engagement is not a launch RADAR can read.

Every launch that clears the rule gets its own page with the source post cited and the reading shown in full. The ranked view of the clean set is the verified organic directory, and the whole set is indexed on the RADAR hub.

Each bar is a count of launches read, not a share of a market. RADAR has audited 30 launches across 6 verticals, which is enough to show the method travels and nowhere near enough to describe an industry. The honest limit is visible in the grid itself: marketing and sales carries 9 readings while crypto and web3 carries 2, so those two rows do not deserve equal confidence.

The bar length is the audited count and nothing else. It is not weighted by reach, so a vertical holding one very large launch does not stretch past a vertical holding several mid-sized ones. The bone segment inside each bar is the share of that vertical that read organic. A short bar with a long bone segment is a thin read that came back clean, not a strong result.

Selection is where the honest caveat sits. RADAR reads launches that put up enough public reach for a views-to-likes comparison to mean something, which tilts the corpus toward launches that were already visible. A launch that reached a few thousand people is not absent because it read badly, it is absent because at that size the ratio is noise. Read the grid as evidence that one method holds across markets, and read the launch authenticity report when you want the corpus-wide distribution rather than the per-vertical spread.

Every count here traces back to the same signals, applied identically in every vertical. The load-bearing one is views per like. Across this corpus the ratio runs from 98 to 12,075 with a median of 821, and roughly 500 views per like is the line RADAR treats as the top of the organic range, with a heavier band opening near 2,000.

Two more signals sit underneath it. The timing shape says whether reach arrived on a curve that engagement followed or in a block that engagement never caught up with. The corroborating layers, meaning replies and quotes rather than likes alone, say whether real conversation formed under the post. A launch that clears the ratio test but has almost no replies at its reach reads differently from one carrying a full conversation, and both facts go into the reading.

None of that is proprietary. The thresholds, the calibration, and the confidence rules are published on the methodology page, the ratio itself is benchmarked in our views-to-likes benchmark, and the wider picture of how launch videos are performing sits in the state of launch videos. A method you cannot check is a method you should not trust.

Paying for distribution is legal and common, and RADAR measures how reach was built rather than anyone's honesty. On this page that policy has a specific shape: every per-vertical figure is an aggregate count, the remainder of each vertical is never given a public label here, and only the 10 launches that read organic are named and graded.

The products in this corpus are real and so are the founders behind them. A reading describes a distribution signature, which is a statement about mechanics, not an accusation about a person. Where RADAR reads a lift, the wording is a signature, never a claim that anyone lied, and a launch RADAR cannot read with confidence is labelled as such instead of being called either way.

Counts are also not a ranking. A vertical with more readings is not a worse or better market, it is a market RADAR has spent more time in. If you want the per-launch detail, each reading lives on its own page with the source post cited and its confidence stated, and the amplified reads are worked through one at a time in the launch authenticity study.