Beyond followers: how financial firms can win the WeChat recommendation algorithm
Key takeaways
Followers no longer determine reach on WeChat: industry estimates suggest 70–90% of an official account’s traffic now comes from non-followers.
WeChat’s algorithm, the Friends♡ feed and AI search now decide who sees content, rewarding articles that readers finish, share and save.
Small, expert-led accounts can now reach investors far beyond their subscriber base, while large accounts with dull content lose out.
Image posts and Channels video are growing fast, and articles written to answer real questions are more likely to be quoted in AI search.
WeChat now restricts fully AI-generated content, so a named human expert and content worth forwarding are what the platform rewards.
Everyone knows WeChat. After all, it is the world’s most fully-realised super-app with all-in-one functionality that covers messaging, payments, social media, as well as a whole range of third-party services. No wonder WeChat’s official accounts (公众号) are a key channel for financial firms looking to connect with B2C and B2B audiences in China.
But not all international financial companies know how to make the most of the platform. In fact, it is common to see companies operating official accounts that fail to incorporate many of the features that can significantly increase reach.
Although many financial companies are still running a WeChat account fit for 2018, a few simple changes can bring it up to speed.
Why followers no longer determine reach on WeChat
The old way of running an account focuses primarily on pushing content towards subscribers, with success measured by growth in the number of followers. That fails to capture the increasingly important role of platform recommendations for driving traffic towards an account, with industry estimates that as much as 70–90% of an account’s traffic can now come from non-followers, of which recommendation traffic accounts for roughly 50–60% in 2026.[1]
Recommendations from WeChat’s algorithm – which rewards content that readers complete, share, and save – now determine the reach of the content published on the platform, with users served content that aligns with their interests, whether or not they are a subscriber of an account. The result is that most of an account’s readers now come from outside its follower base:
The algorithm is not the only source of non-subscriber traffic. The “Friends♡” (Friends Recommended) tab, prominently placed near the top of the Official Accounts information feed, aggregates articles that a user’s WeChat friends have marked with the ♡ “Recommend” (推荐) button or recently read. When a reader taps ♡ on an article, it appears in their friends’ Friends♡ stream[2] and in the “Look” feed, turning every recommendation into a trust-weighted social signal that can carry content beyond the original follower base.
AI also provides readers: WeChat’s AI search draws on a pool of more than 30 million official account articles, plus 260 million Channels content tags,[3] quoting official account content that answers a user’s questions, while agents like Xiaowei (小微), WeChat’s in-house native AI assistant, summarise recent articles from the accounts a user reads the most.
The rise of recommended posts solves a problem that many financial firms faced on WeChat – namely, that content would only go to subscribers. For companies with a newly opened official account, or those with a niche focus, a small number of subscribers would translate into equally small reach.
But now, a B2C business with a small expert-led account can reach affluent investors far beyond its follower list. In the B2B space, content with depth and usefulness can be shared into the right industry groups.
In short, a small account can now claim the attention of people outside of its subscriber base, yet an account with a large pool of subscribers that posts unengaging content will see only a fraction of its more interesting counterpart’s reads.
Four ways financial firms can win WeChat recommendations
1. Post image posts and Channels video, not just articles
WeChat’s image posts (贴图), nicknamed the “Little Green Book” for their resemblance to XiaoHongShu, now have their own tab and their own place in search and in the discovery feed. Users published 116 million of these carousel-style posts in the past year, up 377%, and 52% of accounts active in the past month have used the format.[4]
Video is the other format that is growing in importance. Total user time spent on Channels (视频号), WeChat’s video platform, grew more than 20% year on year.[5] Official accounts and Channels are also converging: 2.82 million official accounts have now linked to a Channels account,[6] letting traffic flow between long-form articles, short videos and livestreams. Chinese fund houses have found that long conversations with their own experts, cut into short clips, outperform hard-sell videos.[7]
2. Write for WeChat AI search
Put the key point in the first half of the title, because Xiaowei’s summaries show only that much, and answer a real question rather than announce news. Lead with the conclusion, use clear subheadings and cite data. An article titled “What T+1 settlement means for Asian custody” is far more likely to be quoted in an AI answer than “Q3 market outlook released”.
3. Put a named expert at the centre
With AI summarising content, a recognisable voice is what makes readers click through to the full piece and come back for the next. This matters because WeChat now actively favours human-made content.
In March 2026 the platform updated its Official Account Platform Operating Rules to add a clause on “non-real-person automated creation”, banning content that departs from a real creator’s expression and is generated, rewritten, stitched together or copied using AI, as well as content published in bulk or continuously through scripts or programmatic scheduling.
The ban targets fully automated creation rather than AI-assisted creation: WeChat still allows creators to use AI tools for research, error-checking and polishing.[8]
AI can help with research and polishing, but the topic, viewpoint and examples must come from a person, and AI-generated content on Channels must be labelled: since January 2026 Channels has encouraged creators to label AI-generated content in all scenarios, and since August 2026 such labelling has been mandatory for e-commerce short videos containing AI-generated content, with unlabelled videos having their distribution limited.[9]
For firms that machine-translate global research and post it at volume, that is a risk. Giving a strategist or product specialist a regular, signed column, or a recurring slot on Channels, is what the platform now rewards.
4. Make content worth forwarding
The ♡ “Recommend” button, shares and saves are among the strongest signals to WeChat’s algorithm: every ♡ tap surfaces an article to the reader’s friends through the Friends♡ card, while shares route it through Moments, group chats and one-to-one forwards, each placing it in front of a new network.
A clear chart, a checklist or a one-line takeaway gives readers a reason to forward an article to a colleague. Running content as a series, such as a monthly outlook or a weekly market explainer, gives them a good reason to come back.
About Vermilion Asia: Vermilion Asia is a financial services marketing and communications firm based in Hong Kong and Shanghai. Since 2015, it has helped international and Chinese banks, asset managers and fintechs reach investors through PR, thought-leadership content, events and social media.
[1]Friends♡ (Friends Recommended) card in the Official Accounts feed: it aggregates articles friends have tapped ♡ on — i.e., the public "Recommend" (推荐) button, which replaced "Watching/在看" in January 2025 and is distinct from the private thumbs-up "Like" (点赞) — or recently read. The card is often placed at or near the top of the feed, ahead of subscribed accounts. Sources: 36Kr (March 2025); NewRank (March 2025); thumbs-up vs ♡ distinction per Jianshu (July 2026).
[2]WeChat AI search draws on more than 30 million official-account articles and 260 million Channels content tags. See Baidu Baike, WeChat AI search and NetEase / 雷科技.
[3]Industry analysis of 2026 official-account traffic patterns, which puts the non-follower share of traffic at roughly 70%–90%, with recommendation-driven traffic at about 50%–60%: Sohu (June 2026). See also NewRank. This is an industry estimate, not an official WeChat disclosure.
[4]Official Accounts 14th Anniversary Creator Report, August 2026: 116 million image posts (贴图) were published in the past year, up 377%; 52% of accounts active in the past month used the format, and more than 10 million such posts have been published in total. Reported by 广州日报 and Sina Tech.
[5]Tencent Holdings, 2026 second-quarter results: total user time spent on Channels grew more than 20% year on year, and marketing services revenue rose 22% year on year to RMB43.565 billion. Reported by IT之家 and mpaypass.
[6]Official Accounts 14th Anniversary Creator Report, August 2026: 2.82 million official accounts have linked to a Channels account. Reported by 广州日报 and 新京报贝壳财经.
[7]Chinese fund houses’ content practice on Channels: Yicai.
[8]WeChat's response on the 27 March 2026 update to the Official Account Platform Operating Rules, which added a clause on “non-real-person automated creation” (非真人自动化创作行为). Reported by 齐鲁网 and 新浪 (3.27 条新规). The rules prohibit fully automated creation, while explicitly allowing AI tools to assist with research, error-checking and polishing.
[9]Channels rules on AI-generated content: from January 2026 Channels has encouraged creators to label AI-generated content in all scenarios, and from 10 August 2026 labelling is mandatory for e-commerce short videos containing AI-generated content, with unlabelled videos having their distribution limited. Reported by 网经社 and Torchtree.
Comments