Platform strategy field study · 2026

Traditional content platforms

Can independent developers replace Douban, Zhihu, or Jianshu? Zhihu's 2025 revenue fell 23.6%, yet it still reports 953.9 million cumulative content items.

This report separates operating data, content stock, and app-store signals. The evidence does not support one person rebuilding a whole platform; it supports only bounded hypotheses about adjacent opportunities.

Background
Douban, Zhihu, and Jianshu remain visible despite product stagnation claims and newer social formats, raising questions about their remaining value and replaceability.
Purpose
Identify the value these platforms still hold, assess whether an independent developer can replace them, and determine whether the evidence reveals adjacent business opportunities.
Main sources
Zhihu filings and results, current platform and App Store pages, CNNIC and company disclosures, regulations, and platform-economics research.
Zhihu 2025 revenue
RMB 2.749B · down 23.6%
Cumulative content
953.9M · company-reported
Operating expense / gross profit
130.8% · based on 2025 reported figures

Brief mode: data findings, opportunity hypotheses, and evidence limits stay visible while detailed sources are condensed.

01

The data do not support one-for-one replacement

The public record shows contraction, large accumulated state, and high operating load—not a proven “best startup direction.”

Copying features does not migrate platform stock

The current evidence cannot show that they can. Zhihu's 953.9 million content items establish a migration scale; Douban and Jianshu lack comparable operating denominators.

Adjacent opportunities remain plausible, not established markets

The evidence points to narrower products rather than a full clone. Object curation, private-community operations, creator-owned publishing, and moderation tools each address a visible function, but comparable payment data are unavailable.

No single successor has absorbed every use case

Public data do not measure the reason directly. Large content stock, the expense structure, and multi-homing research support three mechanisms; they remain inferences rather than established causes.

Vertical object networks preserve a core incumbent value

This hypothesis extends an observed platform value: a trusted object database for one expert or enthusiast niche, paired with compare, shortlist, review, submit, procure, or report work that users repeat. Payment and retention remain unmeasured.

02

Content stock, identity, and distribution remain the durable assets

The public evidence shows a large accumulated corpus, continuing paid demand, and unequal platform footprints.

953.9M cumulative Zhihu content pieces reported at 2025 year-end
-23.6% Zhihu 2025 total-revenue change
56.0% of Zhihu’s 2025 revenue came from paid membership

All three major Zhihu revenue measures declined

Expressed as a share of 2024, 2025 total revenue retained 76.4%, paid-membership revenue 87.3%, and marketing-services revenue 67.7%. Marketing services contracted fastest.

Total revenue76.4%
Paid membership87.3%
Marketing services67.7%

Calculation: 2025 ÷ 2024. Year-over-year changes are -23.6%, -12.7%, and -32.3%. Operating results do not by themselves identify why a product may be replaced.

China App Store observed 2026-08-12; cumulative ratings are not MAU or retention.
PlatformCumulative ratingsRatingSocial-networking rank
Doubanabout 1.59M4.6#19
Zhihuabout 1.74M4.7#9
Jianshuabout 87K4.8#142

Douban: cultural memory and taste identity

Its object catalog, ratings, reviews, lists, tags, consumption history, and groups make it a reference layer and a public record of taste.

Zhihu: searchable expertise and question routing

A large indexed Q&A corpus is connected to identity, topic expertise, reputation, recommendation, search, and paid membership.

Jianshu: low-friction long-form publishing

Long-form and serial writing, Markdown, rich text, interest spaces, subscriptions, and an archive keep the path from writing to publishing simple.

Current functions and bounded evidence; company figures are not treated as independent quality measurements.
PlatformDurable assetCurrent observationEvidence boundary
DoubanStructured cultural-object graph and taste historyBooks, films, music, podcasts, marks, reviews, groups, topics, local eventsNo current audited audience or financial figure found
ZhihuQ&A corpus, expertise, reputation, and routing13.5M average monthly subscribers, 80.3M cumulative creators, 953.9M content pieces reported for 2025Company filing; does not measure satisfaction or content quality
JianshuAccessible long-form publishing and archiveCurrent iOS app lists Markdown, serial writing, interest spaces, subscriptions, and virtual currencyCurrent active users, creator earnings, retention, and profit unknown
The database is not free seed capital

Douban’s legal statement covers object metadata, scores, reviews, counts, marks, topics, and groups, and prohibits unauthorized extraction and derivative services. The lawful source classes are licensed, public-domain, first-party, or user-consented data.

03

Content stock, operating load, and fragmented use raise the cost of wholesale replacement

Public numbers do not isolate one cause, but they show why reproducing the interface is only a small part of the job.

953.9M · State
Observation: Zhihu approaches one billion cumulative content items; Douban adds object catalogs, ratings, and collections. Inference: replacement depends on lawful seed data and coordinated supply and demand, not interface work alone.
130.8% · Operating load
Observation: Zhihu operating expense exceeded gross profit; sales and marketing equaled 45.6% of revenue and R&D 19.1%. Inference: the relevant cost base includes acquisition, moderation, support, and engineering.
18–20× · Visible-footprint gap
Observation: Douban and Zhihu have about 18.3 and 20.0 times Jianshu's cumulative iOS rating count. Inference: the targets are not equal in observable footprint; ratings still cannot substitute for MAU.
Unknown · Causality
No cross-platform migration, willingness-to-pay, or Douban and Jianshu operating-cost dataset exists here, so these mechanisms cannot explain the full absence of a successor.
Broad community is not lightweight SaaS

Zhihu reported RMB 2.749B in 2025 revenue, RMB 195.2M GAAP net loss, RMB 1.252B sales and marketing expense, and RMB 525M R&D expense. Even at scale, the operating model is heavy.

04

Adjacent tools fit the evidence better than a general-purpose clone

Each product shape maps to a durable platform function; public data do not establish its market size or rank the alternatives.

HypothesisObserved basisDecisive missing evidence
Vertical object + expert networkDouban's object-centered structure and Zhihu's expert content show durable catalog and knowledge functionsNiche willingness to pay, contribution rate, and return behavior
Private-community operationsPublic-platform operating costs and governance research show that moderation and coordination are material functionsWhether administrators would replace current group tools and what budget exists
Creator-owned publishingJianshu's publishing role and platform-dependence concerns support an ownership and portability hypothesisWhether writers pay for ownership, export, and distribution rather than only publishing
Personal library / curationCollections and object histories create value that does not require a public feedStandalone retention, switching cost, and data-import feasibility
Moderation / reputation toolingZhihu's expense structure and regulatory duties imply non-trivial governance workActual governance time, accuracy requirements, liability, and buyer budget
AI lowers software production cost, not migration or governance cost

AI can compress CRUD, authentication, search UI, payments, notifications, admin tools, normalization, and moderation queues. It cannot create licensed data, credible contributors, norms, distribution, retention, or production judgment.

05

What remains unknown

These gaps separate the observed platform facts from claims about replacement or commercial opportunity.

Comparable current usage
No current audited MAU, retention, or content-production series was found for Douban and Jianshu, so the three platforms cannot be compared on the same denominator.
Cross-platform migration
The evidence does not show how many users want to leave, what data they would move, or whether they would rebuild relationships elsewhere.
Willingness to pay
No comparable dataset links the proposed adjacent functions to buyer budgets, conversion, or durable subscription revenue.
Contribution and retention
Public content stock does not reveal how a new service would acquire credible contributors or retain readers without the incumbent network.
Operating cost
Zhihu discloses a heavy expense structure, but the share attributable to acquisition, moderation, compliance, support, and infrastructure is not comparable across platforms.
Rights and regulation
Data portability, catalog reuse, user consent, content rights, and regulatory duties depend on the exact product and cannot be inferred from interface similarity.
06

Sources and limits

Official platform materials and filings establish current functions and economics; research literature supports the mechanism analysis.

Research limits

No current audited audience or financial figures were found for Douban or Jianshu. Zhihu figures are company filings and do not measure content quality or satisfaction. The opportunity table describes evidence-bounded product shapes, not success probabilities or a comparable-sample ranking. Specific regulatory applicability requires qualified legal assessment and remains outside this report.