Nanjing has many platform businesses in many shapes: a tea chain's ordering mini-program and an office-based studio signing a dozen livestreamers both count as platforms, and manual sweeps easily miss them. The city's tax service built a discovery method that combines regulatory data from market, development and publicity agencies with invoice records, traces leads backwards from payers of platform settlement income, and cross-checks third-party databases plus ICP and EDI licences and influencer follower counts. Staff wrote Python scripts to distil shared traits of known platforms into a batch identification model, producing candidate lists for case-by-case checks. The city has mapped 123 platforms and 64 MCN agencies, the largest register in the province. For reporting, the bureau compiled a common-mistakes guide covering nine forms and nearly 200 fields, and sent teams to help leading platforms such as Yunmanman, achieving 100 percent on-time reporting in both rounds.
News Detail
Nanjing leans on big data to supervise the platform economy
Nanjing's tax service has built a pipeline of multi-source collection, model identification and manual verification to bring internet platforms into tax reporting. It merges regulatory information from other agencies with invoice data, traces leads back from payers of platform settlement income, and cross-checks with third-party databases and licence records; staff also used Python to distil common features into a batch identification model. The city has now mapped 123 platforms and 64 MCN agencies, the most in the province, and both reporting rounds were completed on time.
2026-10-26