← Back to the journal

Zhihu links vs contact-info boundaries

Based on Zhihu's community and org-account rules, this guide explains the real boundary between reference links, contact details, and diversion on Zhihu, so you know what may remain and what should be removed before publishing.

Here is the short answer: Zhihu does not mainly punish the bare presence of a link. It punishes content designed to divert, capture contact paths, or push promotion. A reference link may sometimes remain, but URLs, QR codes, email addresses, chat handles, and their disguised variants are much riskier because they can function as contact infrastructure. For multi-platform content teams, the common mistake is not total ignorance of Zhihu's rules. It is treating “citing a source” and “leaving a conversion path” as if they were the same thing. Tools like OmniGoAI's OmniPost can distribute one article to many platforms, but the Zhihu version still needs a platform-specific judgment before you publish.

Many teams make Zhihu decisions by intuition. They leave a website URL in the body and call it supplementary reading. They add a final line like “message me for the template.” They break a chat ID into phonetic fragments and assume it no longer counts as contact info. Zhihu's rules are much less interested in how soft or indirect the wording feels. The real questions are: is the post trying to move the reader off-platform, is it leaving a usable contact path, and is the surrounding language promotional? Of those three, only the first category—limited reference-style linking—still has some room for discussion. The other two are where the rules become much harder.

This article answers five practical questions: why links and contact info should not be lumped together, which links behave more like references and which behave more like diversion, why Zhihu is especially strict about disguised contact methods, how to review a post before direct publishing, and why a website article usually needs a dedicated Zhihu rewrite.

A common summary goes like this: “Zhihu does not allow external links.” That summary is too crude to be useful.

If you read Zhihu Community Rules and Zhihu Org Account Usage Norms (Trial), the platform's main targets are:

  1. diversion, marketing, and other promotional behavior;
  2. contact information, including QR codes, URLs, email addresses, and phonetic, numeric, or symbolic disguises;
  3. attempts to evade review through prominent placement, special symbols, or image-based variants;
  4. the use of institutional identity or coordinated accounts for promotion and lead capture.

The key point is that the rules focus on intent and function, not only on whether a URL appears. That creates an important split:

  • some links are there to prove a fact or cite a source;
  • some links are there to move the reader elsewhere, capture contact, or drive conversion.

Those are not the same thing in Zhihu's eyes. The first category is not automatically safe, but it still has more room. The second is usually where the real risk begins.

The most useful question is not “can I leave a link,” but “what job is this link doing inside the article?”

If a link only supports a factual claim, cites a rule, or points to a verifiable source, it is usually safer than a conversion-oriented link. Typical cases include:

  1. linking to a platform's official policy page;
  2. citing a public announcement or research source;
  3. linking to a document directly tied to the factual point being made.

In a policy article like this one, an official rule link is there to help the reader verify the source language. That makes it easier for the platform to read the link as reference material rather than a funnel.

Even then, three cautions still matter:

  1. do not make external links the backbone of the article;
  2. do not combine source links with repeated “click here for more” nudges;
  3. do not turn a source link into a CTA.

Risk rises sharply when the link is doing things like these:

  1. “click here to get the template, join the group, or unlock the full version”;
  2. “go to the site to register, download, buy, or book”;
  3. “the rest is in our public account or private channel”;
  4. “message me and I will send you the link.”

The problem is not the URL itself. It is that the URL is clearly being used as an off-platform conversion action.

Once contact information appears, Zhihu is far less likely to interpret it as neutral supplementation. It is much easier to read it as diversion and lead capture infrastructure.

Under Zhihu's rules, the following are all high-risk forms of contact information:

  1. chat IDs, public-account names, and QR codes;
  2. email addresses, phone numbers, and website addresses;
  3. phonetic spellings, number substitutions, and symbol-separated fragments;
  4. disguised forms such as initials, homophones, or split strings;
  5. contact cues placed in images, covers, avatars, or bios.

One subtle but important point is this: a URL can be either a reference link or a contact/diversion path depending on how it functions. If the main role of the URL is to help the reader contact you, register with you, or enter your funnel, it stops being “just a link.”

Sometimes the body is relatively restrained, but the real problem sits in prominent surfaces.

1. The title, opening, and closing CTA

A post may look educational in the headline, then place a website address in the opening and end with “message us for the template.” That makes the whole piece easier to classify as acquisition content rather than knowledge sharing.

2. Avatars, bios, banners, and image watermarks

Zhihu is especially sensitive to prominent placements. Even if the body text looks clean, QR codes, public-account mentions, or disguised contact strings in visible design elements can still create serious risk.

3. Comments and supplementary notes

Some teams assume that leaving contact info in the comments is safer than putting it in the main article. That is not a strong assumption. If the purpose is still diversion, lead capture, or off-platform contact, the risk logic does not disappear just because the location changed.

Use this review order before publishing:

  1. If you remove the link, does the article still stand? If yes, it may not be necessary.
  2. Is the link proving a fact or pushing an action? Fact support is usually safer than action prompting.
  3. Does the link leave a usable contact path? Homepages, signup pages, download pages, and private-community pages deserve extra caution.
  4. Is there CTA language around it? Once a link is paired with “reply for,” “message for,” “join,” or “download now,” the risk rises quickly.

If you run a content-distribution workflow, this decision is best made inside a Zhihu-specific rewrite rather than after copying the website version unchanged. Two related references are Three ways to publish Markdown to Chinese platforms and How to monitor published, reviewing, and offline posts. One helps with the rewrite-and-publish path; the other helps verify what happened after submission.

A safer rewrite pattern: keep sources, remove conversion paths

If you regularly repurpose website posts for Zhihu, the safer pattern is usually this:

  1. keep links that support rules, announcements, or factual claims;
  2. remove download pages, signup pages, private funnels, and direct contact paths;
  3. rewrite “click here to get it” into neutral guidance such as searching for the official rule or the brand name;
  4. limit brand mention to one natural mention early and one restrained mention at the end;
  5. make the article complete on Zhihu itself instead of forcing the key answer off-platform.

That is why Zhihu should not be treated as a simple mirror of the website version. The site is good at carrying richer internal-link structure and product paths. Zhihu is better when it answers the question directly and minimizes off-platform prompts. If you also publish to other Chinese platforms, see Xiaohongshu external link rules explained and WeChat Official Account link policy, decoded. Different platforms regulate links differently, but Zhihu is especially sensitive to promotional intent and disguised contact methods.

A pre-publish checklist you can actually use

Before sending a post to Zhihu, review these seven items:

  1. remove chat IDs, QR codes, emails, phone numbers, and disguised variants;
  2. remove “message me,” “reply with a keyword,” “join the group,” and similar action prompts;
  3. review every URL and ask whether it is evidence or an entry point;
  4. if the link points to a homepage, signup page, download page, or private funnel, remove it by default;
  5. check titles, covers, images, and bios for hidden promotional cues;
  6. turn the closing CTA into a restrained summary or brand mention instead of an off-platform action;
  7. verify the publish status afterward instead of assuming “submitted” means safe and live.

Frequently asked questions

That is too simplistic. Zhihu mainly targets diversion, marketing, and promotional behavior rather than every possible link. But links with clear conversion intent are much riskier.

It can be either. If it only supports a factual point, the risk is lower. If it mainly works as a path to register, download, contact, or enter a funnel, it behaves more like contact or diversion infrastructure.

Is a disguised chat ID safer than a normal one?

No. Zhihu's rules explicitly mention phonetic, numeric, and symbolic variants, so disguised forms are still high-risk.

Is it safer to put contact info in the comments?

Not really. If the underlying purpose is still diversion or lead capture, changing the location does not change the nature of the action.

What is the most practical rule for content teams?

Keep links that prove a fact. Remove links that move the reader toward conversion. Zhihu can be a place to answer the question, but it should not become the first step of an off-platform funnel.

If you distribute one article to Zhihu, CSDN, Juejin, and other platforms, the sustainable improvement is not merely “use fewer links.” It is to encode each platform's real boundary on links and contact cues into your publishing workflow. If you want to do that on a local-first stack, start with OmniPost: https://omnigoai.com/en/download/omnipost/

#Zhihu#link policy#content distribution#OmniPost

More from the journal

9 min

Why AI-assisted posts on Xiaohongshu should be labeled

Based on Xiaohongshu community rules and Community Convention 2.0, this guide explains why AI-assisted posts should be labeled and which combinations of hype, diversion, and template-like copy raise risk fastest.

Read