Content recommendations that keep every visitor reading.

Software that picks the next article, video or guide to show each reader, from what they and readers like them have read — and keeps learning from every read and click. A returning reader gets picks from their own history, someone new gets what readers where they are read most. Personyze does it on your site, in your emails and in your apps.

One for each kind of page, with fallbacks behind it: on the home page Content Recommended for You, then Popular in the Visitor’s Country and Trending Now; on an article Read Next, then Visitors Who Read This Also Read and Similar Content (AI); inside a section Most Popular; on search results Matches Their Search. Or leave a widget on Automatic — new widgets start there — and one widget chooses for every page by itself. See every algorithm.

Yes. Personyze reads every article once and records what it is about: its topics, the people, companies and places in it, whether it is news, a guide, an opinion or a case study, its level, and whether it dates quickly. It blends that with what your readers did, so a new article is recommended from its topics the day it is published, and a popular one mostly from what readers read next. Each reader is offered the right next article: more on the topic, or the natural next step — a news story leads to the explainer, a guide to a case study, a series to its next part. It adapts to your kind of site: freshness first for news, from learning to the demo on a B2B blog, by level for courses, the next episode for video. Nothing to configure.

Both run on the same engine, each on its own catalog. Product recommendations pick from your products and are measured by purchases; content recommendations pick from your articles, videos or guides and are measured by clicks and the content goals readers reach. Personyze does both, so publishers, e-commerce stores and hybrid sites can recommend whichever fits each page.

Yes — Personyze works on any website, whatever it’s built on: one tag goes into any CMS, site builder or custom code. The WordPress and Joomla plugins put it on in one click — on WordPress the plugin also keeps your catalog current every day — and Google Tag Manager and Webflow take one click too.

Yes. The widget translates its own words — buttons, titles, labels — by itself, and your feed can carry more than one language: set a language and map which feed fields go with it, so each reader gets articles in their own.

Install the tag and point us at your articles, and Personyze does the rest: it sees what each reader opens and maps it all. Install the tag: Works on any site, whatever it’s built on: paste one tag into any CMS or custom build — or let the WordPress or Joomla plugin, Google Tag Manager or Webflow put it on in one click. It is what lets Personyze see what each reader reads. Get your articles: We crawl your site: Personyze finds your article pages through your sitemap and links, reads each one’s title, picture, date, author and section, and reads them again on its own, so new articles follow. Or send a feed or files — RSS or Atom, a CSV, Excel, XML or JSON file at its address, by SFTP or over the REST API — and on WordPress the plugin’s feeds keep the catalog current every day. We see what each reader reads: The tag knows an article page by its address or by the ID on the page — one click, nothing to change on your site — and records the read against that article. We map it: Each article’s topics — your own categories, or keywords Personyze pulls from its text — become a reader’s interests as they read. Similar Content (AI) reads every article and finds its neighbours by meaning, so a new piece has related reading before anyone has opened it. Each article is kept by its Internal ID. Similar Content (AI) is switched on in Catalog settings and uses AI credits.

A new reader gets picks from their very first page: Popular in the Visitor’s Country needs only their location. A new article can be recommended by Recently Published straight away, and by Similar Content (AI), which matches articles by meaning from their titles, topics and descriptions; it is switched on in Catalog settings and uses AI credits.

Yes. A recommendation block in the emails you already send picks each reader’s articles when they open the email, not when it was sent — what was published since their last visit, for example. A sign-up popup shown only to readers of an author or a topic grows the list it goes to. See the open-time email guide.

Yes. In iOS and Android apps, the SDK — Content Recs for Apps — brings the same engine as your site. For anything else, the JSON feed returns a reader’s picks, found by their email or your own user id, for your front end or email platform to show in its own design; each item carries a link that counts the click.

Ask the assistant in your panel to build it, point at the spot on your own page in the editor, or start from one of six article formats in the gallery — in-article, continue reading, paging, most read, magazine or overlay — under or inside the article, in a sidebar or as a popup. Then change its layout and style in the designer, ask Edit with AI, or edit the HTML yourself.

Yes. Put two widgets — two algorithms, layouts or placements — in an A/B test. New tests call a winner at 98% confidence and at least 1,000 sessions, and can give the winner all the tested traffic.

For every widget and every article: times shown (only when it was really on screen), clicks, click rate and the content goals readers reach within 7 days, with what is rising or falling. Recommendation sets show what the engine learned — each article’s reads, comments, likes and goals, and the articles most often read together.