What changed for you as a user of the app or the plugin, newest first. Rebuilds you cannot see are not listed.
The plugin has its own page in the documentation: installation, the site key, the wp-config constants, how Yoast and Rank Math are handled, and a checklist for when nothing changes on a page.
Read The WordPress plugin.
The search field on the URL page has a mode: URL or Content. A row used to match on its address, its title or a word in its body all at once, so the list could not be explained from the field. Rows for addresses like /?p=1 now show their query instead of appearing as /.
A WordPress site arrives with wp-admin, feeds, ?p= links and tracking duplicates in its crawl, scored as pages and listed as "missing title". The URL view now opens with a card grouping them: admin, api, feeds, checkout, search, WordPress ID links, tracking parameters and files. One click excludes them, and every count in the app reads the same definition of which pages count. Three numbers in the Overview that were wrong for three different reasons are fixed.
A format, such as how a date is written, is set on the variable and applies everywhere it is used: in the preview, in the insert panel and in the published title. Rules can now use two conditions that existed in the engine but were missing from the menu: the query string contains a value, and a date falls in a range.
The variables page says the same thing everywhere: Your own, From your site and Rule. The list is grouped by the question you are asking, suggestions from the crawl have a single place to be approved, and a new site is offered a title suggestion built from its pages' own fields instead of an empty box. At the top, the template is shown on three real pages from the site.
The link analysis did not distinguish "the page is gone" from "the server declined to answer". A site whose host answered 508 to 62 of 262 requests showed 1,859 broken links and had one. Now only real errors count as broken, and a refusal is shown as a refusal.
The help panel is on every page and describes the page you are on. The documentation on dynamicseo.com answers what the panel cannot: what the product is, how discovery, categories, crawling and publishing fit together, and how the score is computed.
Read the documentation.
A newly added site could sit uncrawled with no explanation. Now it is discovered, crawled and shows what is happening in the Overview. Keywords, Integrate and Visualize are reachable from the menu on every page.
The robots directive used to be published with whatever value the crawl happened to see on the page, so a site's own noindex could come along and be served by us. Now robots is published only when you set it, per URL, and a published correction is recomputed immediately instead of after the next crawl.
The keywords view starts from conflicts: phrases that two or more pages target. With Search Console connected, conflicts Google has actually measured are marked apart from the ones we inferred from page content. Shared words, translations and your own brand name no longer count as conflicts. The list handles thousands of rows without slowing down.
The same thing had eight names on the variables page. Now the three types are called the same thing in tabs, lists, filters and badges. A template with a token that cannot be resolved can no longer be published.
The Performance view shows per URL what Google says about indexing and which canonical Google chose, compared with the one you published. When the account has several properties, the domain property is preferred.
The plugin takes over one field at a time: only the fields you published are replaced, the rest is left to Yoast or Rank Math, so a social image you set there stays (4.5.1). Rank Math's advanced robots tag is silenced when Dynamic SEO delivers the directive, so the page gets one robots tag instead of two (4.5.2). The crawler always sees the site's original page, never the injected output, so a title template cannot stack on itself at the next publish (4.5.3).
The Test connection button under Settings → Dynamic SEO says whether content is being served, not yet published, or the key was rejected. Sites with www in their address match their registered domain. A canonical pointing off the site is dropped instead of risking de-indexing. The plugin's own cache is five minutes, so a missed purge heals itself.
Every cache layer is cleared when you publish: in the plugin, in our API and in our own cache, and the new version is warmed so the first request hits the cache. A page could previously serve the old version for up to an hour.
Under Content you can import a CSV. Rows are matched to crawled URLs, fill the SEO fields and group the pages into categories. A preview shows what would change before anything is written, and you choose whether the file completes or replaces what is there.
The site's structured data is drawn as a radial graph around the page, with a side panel that is always there, even while the view is loading or empty.
The three ways to define a category, URL pattern, page content and hand-pick, share one editor next to the URL table. Hand-pick works end to end: ticked rows are assigned when you save. Search the pages' content and capture the matches as a category.
How blocks connect is your decision, not a guess in the background. The builder lets you pick blocks freely and say how they connect; the engine only writes the references. Fields with unfilled placeholders are left out instead of being published with the braces still in.
A site can be put in review mode: new URLs the crawler finds land in an inbox instead of straight in the overview. Filter rules can include or exclude, and excluded pages are hidden automatically.