The claim in the group chat

Search the number in quotes together with the topic. Confident round numbers that circulate widely — 90%, 70%, 10,000 hours — usually trace back to one source, and the trail is often short.

Pay close attention to the period. Failure rates over one year, five years and ten years are wildly different numbers, and the timeframe is the detail most often dropped when a statistic is shared.

A strong submission reaches an actual publisher — a statistics agency, a research body, or an academic paper — and quotes what it really said.

The most common finding on this kind of claim is that the timeframe changed. Business survival statistics are typically published over multi-year windows, so a figure describing failure over five or ten years frequently gets retold as "in the first year".

Also strong: noticing that "startups" and "new businesses" are not the same population, and that official survival statistics usually cover all new employer businesses, not venture-backed startups.

"Could not verify" is a completely acceptable verdict if you searched properly and no original exists. Many circulating statistics have no traceable source at all, and establishing that is a real result — it is the difference between "this is false" and "nobody knows where this came from".