**TL;DR** — The Baidu update that rolled out between February and March 2026 did not rewrite the ranking rules, but it did shift weights enough that three cohorts of sites moved noticeably: thin aggregator pages dropped, 百家号 self-publishing took another step up, and mobile SERPs diverged further from desktop. If you read only the 站长平台 announcement you will miss the texture.
What actually changed
Baidu does not publish changelogs the way Google does, so practitioner reads are necessarily triangulated: the 资源平台 (ziyuan.baidu.com) announcement posts, the 抓取频次 and 索引量 curves on the 站长 dashboard, and field observation from a few dozen monitored sites. From the four properties we were watching closely through the window, the pattern was consistent. First, the 抓取频次 curve dipped for roughly five days around the visible SERP shift, then recovered to a slightly higher baseline — a signal that Baidu re-crawled broadly to validate the new weights rather than rolling out to a subset. Second, 百家号 pages took more real estate above the fold on mobile, particularly for informational queries with a 资讯 cast. Third, long-tail personalisation tightened: the same query run from two different 地域 (regions) returned materially different top-five lists, something we only saw hints of in 2025.
For the aggregator and scraper cohort, the hit was sharper. Sites whose content was structurally thin — pages that wrap one 200-character answer in navigation boilerplate, or pages that republish 知乎 answers with light rewording — dropped out of the top ten on head terms. Not deindexed, just demoted. The 百度 demotion mechanism here is not new; it just got stricter about what counts as value-add.
The common wisdom in some 中文 SEO circles is that every Baidu update is really about links. In Q1 2026 we observed the opposite: the sites that held position were not the ones with the most backlinks but the ones with the most direct-answer coverage, a visible 作者 byline, and a 实名 (real-name) presence on 百家号 feeding into the same topical cluster. Link profiles did not explain the movers. Authorship signals and on-page answer density did.
How to read the dashboards
Once an algorithm update is visibly live, the 站长 (ziyuan.baidu.com) dashboards earn their keep. The first curve to watch is 索引量 — index size. Expect a small wobble during a major weight change; what matters is whether the curve settles at a higher or lower plateau over fourteen days. A seven-day dip that recovers is just the crawler re-validating. A persistent drop of more than ten percent on a previously stable site means content that was borderline before has now been culled, and you need to find the affected URL clusters quickly. Use the 链接提交 log and cross-reference it with the URL-level 抓取诊断 tool.
The second curve is 抓取频次, crawl frequency. A healthy site post-update sees frequency rise within two weeks on refreshed URLs and stay flat on the rest. If your overall frequency rises but your index size does not, Baidu is politely saying it is spending crawl budget evaluating you without being convinced. That is a content-quality signal, and the answer is not "publish more" — it is "improve what you already have so the re-crawl rewards you."
A third dashboard worth sitting with is 流量与关键词 under the 流量分析 tab. After a major update the keyword-level CTR distribution changes shape — queries that previously delivered steady clicks can flatten even when the ranking itself looks unchanged. That is almost always a SERP-feature shift: 百度 has either added an aggregated answer card, handed the slot to a 百家号 post, or inserted a 视频 block. The fix is a title rewrite and often a schema sanity-check, not a link-building push. Export the keyword list sorted by week-over-week CTR delta and walk the top twenty manually; the pattern becomes obvious after ten.
On the technical side, a useful post-update ritual is a scripted 抓取诊断 loop against a representative URL set. Run 抓取诊断 on the same ten URLs every seven days for the first month after the update and diff the rendered HTML against what a headless Chromium fetch returns with a clean user agent. On sites we have watched through previous updates, a silent regression in server-side rendering for 百度蜘蛛 has been the hidden cause of a third of the ranking drops that looked algorithmic at first. Fix that class of bug and the recovery curve often starts within ten days without any other intervention — Baidu does notice when the page it renders finally matches what a real user sees.
- 01Check the 索引量 curve in ziyuan.baidu.com daily for the first fourteen days post-update. Identify any sustained drop greater than ten percent and pull the URL list via the 链接分析 report.
- 02Run 抓取诊断 on three representative URL types (a 列表页, a 详情页, and a 文章正文 page). If 抓取诊断 reports a different rendering outcome than a desktop browser, your SSR or 百度蜘蛛 detection has regressed and needs an immediate fix.
- 03Audit your top-ranking pages for 作者信息 — 百度 rewards visible authorship and penalises anonymous aggregator pages harder after this update. A named 作者 box with a 实名 link to a 百家号 profile measurably helps.
- 04Re-check the mobile SERP for your five core queries from two different 地域 using a clean browser profile. The delta between regions is now informative — if the same query surfaces completely different 百家号 authors in two cities, the 个性化 weight is at work.
What to ignore
A common myth circulating in 中文 SEO chat groups after this update is that you need to pay for 爱采购 or another 百度 paid surface to recover rankings. The data does not support it. None of the recoveries we observed in Q1 2026 involved 推广 spend; all of them involved improving on-page answer density, adding visible authorship, and pushing a steady 百家号 posting cadence from the same author identity that owns the site content. The update made quality signals heavier, not paid signals. If your agency is quoting a 推广 budget as the answer to an organic drop, ask them to show you the two curves side by side — the organic recovery curve of a site that paid, and an otherwise comparable site that did not. The asymmetry, in our experience, is nowhere near what a "just spend your way back" narrative would predict, and the cleanest recoveries we watched this quarter were pure on-page work paired with disciplined 百家号 publishing.