How should this matter be evaluated?
First check if you have products suitable for video demonstration and stable sales channels, then record orders, refunds, gross profit, and production costs for each video in the same spreadsheet. The 1,596 orders cited are the author’s self-reported result for a single video and cannot be directly extrapolated to your profits or the performance of your next video.
Orders look good, but why shouldn't we decide to proceed immediately?
A single piece of content may benefit from product selection, account, channel, and timing simultaneously. First, break one success into an operational chain: someone sees it, someone clicks, someone orders, someone leaves an order, and only then comes profit. If any segment isn't recorded, it's easy during a case study to mistake accidental viral hits for stable business capabilities.
- Deal closedRecord orders and transaction amounts driven by videos, distinguishing between platform display metrics and actual payments.
- FulfillmentMonitor refunds, after-sales service, and stockouts; quantity sold does not equal final retained orders.
- CostPut products, ad placements, tools, assets, manual review, and after-sales costs into the same ledger.
- ReproduceAfter publishing continuously, focus on median performance rather than letting the single best post represent the entire account.
Tasks AI can assist with versus those merchants must handle themselves
AI is suitable for organizing product information into scripts, generating multiple expression versions, or handling repetitive editing. However, merchants must verify whether product selling points are accurate, if visuals match the actual items, and if promises can be fulfilled. Bilibili editing cases retain human final review; the lesson to learn is this quality boundary, rather than fixating solely on the '30 minutes' figure.
- First define the product.Select a product for which you can verify inventory, gross margin, and real-world usage scenarios.
- Do content againCreate different narratives around the same selling point, recording the production and review time for each piece of content.
- Look at the deal at the endSimultaneously verify playback, clicks, orders, refunds, and net revenue, then modify scripts or pages upon discovering bottlenecks.
When should you not scale up first?
If product profit margins are unclear, after-sales capabilities are insufficient, or you have play counts but cannot attribute sales, do not procure materials and tools at the volume of a viral hit. Another case mentioned tens of millions of views, which is the author's self-reported dissemination result and cannot replace sales data. Running small-scale tests to success makes it easier to identify problems than building a batch generation system first.
