Skip to main content
Theme

Connectors

WordPress

The tool reads a WordPress site through its REST API, signed in with an Application Password for a user made for the move, and copies it to an Insites v6 instance. It only reads. Nothing is ever written back to WordPress. It copies every published page with its design, the posts and every other post type as a database table each with a collection page, the media library, the site's people, its forms, its redirects, and a WooCommerce shop, which sells on the new site through the site's own Stripe account. A site that cannot make an Application Password connects with its export file instead (below). What moves from WordPress, part by part.

What you need

  1. In WordPress, open Users, then Add New User. Make a user for the tool, named insites-migration-tool, with the Administrator role and an email address you read. Use a user made for the move, never the owner's own login, so it can be deleted afterwards.
  2. Open that user's profile (Users, then All Users, then the user) and scroll to Application Passwords.
  3. Type Insites Migration Tool as the name and click Add New Application Password.
  4. Copy the password. WordPress shows it once. In the tool, give the address visitors use (for example https://www.example.com), the user's name and the password.

An Application Password is a password for one app. It can be revoked on its own, and it cannot be used to sign in to the dashboard. WordPress gives Application Passwords only on a site served over HTTPS.

No Application Passwords section? A security plugin switched them off. Wordfence does by default: under Wordfence, All Options, Brute Force Protection, switch off Disable WordPress application passwords for the move, and switch it back on afterwards. Other security plugins, and some hosts, have a setting like it.

The Administrator role is what the move needs: WordPress lets only an administrator read the site's people with their email addresses, its menus, its plugins, its settings and the shop's orders. A user with a lower role, such as Editor, is refused, and the connection test names the role it has.

When the connection test refuses the password

The test says which of these it is, from what WordPress answered:

  • The Application Password is wrong. The user name is right. Copy the whole password again, or make a new one.
  • There is no user with that name. Type the user name exactly as Users lists it. Its email address works too.
  • Application Passwords are switched off on this site. A security plugin or the host switched them off. See above.
  • Application Passwords are switched off for this user. The site offers them, but a plugin keeps this user from using them.
  • WordPress did not accept this user name and Application Password. Some sites do not say which part was wrong. Check both, and if they are right, the site's server may be removing the sign-in from the request. Its host can fix that in the server's configuration.

The address can be the one in your browser's address bar while you are in the dashboard: the tool cuts it back to the site. It must be https://, on the standard port, with no user name or password in it. A site whose permalinks are set to Plain works too: the tool reads its REST API through ?rest_route=.

The address must be the site's own: the REST API says which site it is, and if the address names another site, the tool says which one and stops. The password is kept encrypted for as long as the migration needs it, and removed when you close the migration: how long a key is kept.

Without an Application Password: the export file

Some sites cannot make an Application Password: a WordPress.com site on a free plan has none, and a host or a security plugin can switch them off for good. Connect such a site With an export file instead:

  1. In WordPress, open Tools, then Export, choose All content, and download the export file.
  2. In the tool, give the site's address, choose the file under WordPress export file, and press Connect and check. The tool checks the file is WordPress's, and for the address you gave.

The posts, every other post type, the categories, tags and other terms, the comments, the menus, the media library and the authors come from the file. The pages are still read from the live site, with their design. Advanced Custom Fields come from the file whether or not their group was shown in the REST API, and so do Yoast SEO's titles and descriptions.

What a file does not hold, and so does not move this way: the plugins and the settings, the Redirection plugin's rules (the tool still finds redirects by asking the live site), a WooCommerce shop, and anyone who wrote no post, such as subscribers and customers. WordPress also leaves spam comments out of its export. Content comes as WordPress stores it rather than as it draws it: block markers are taken out, and a shortcode stays as text. The tool keeps the file for 14 days. A final sync reads the same file, so to copy later changes, export the site again and connect with the new file.

A multisite network

A network moves one site at a time, one site per migration. Give the address of the one site you are migrating, such as https://www.example.com/shop or https://shop.example.com: the tool reads only that site. The plan says when the site is part of a network. Each other site needs a migration of its own.

What is copied

On WordPressOn Insites
PagesEvery published page, read from the live site and copied to the same address, with its title, its description and its sharing tags.
Posts and other post typesPosts, and each other post type the REST API lists, become a database table each, with a row per item: the title, the content as WordPress rendered it, the excerpt, the date, the author's name, the categories and tags as lists of names, the featured image, and the item's old address. Each table gets a collection page, so editors add rows and pages appear. See Posts as collections below.
Page buildersA page made in Elementor, the block editor, Divi, WPBakery or any other builder is copied as the HTML it shows, with its styles and scripts, so it looks the same. It is no longer edited in the builder: on Insites it is a page of HTML you change in the instance's admin.
Lists and archivesEvery address the site's sitemap lists, such as the blog, its categories and tags, the authors, and the shop and its categories, copied as it looked on the day it was read. A list does not page or filter on Insites, and a new post does not appear in it. Each author with published posts keeps their author page, every page of it, even when the sitemap leaves the authors out; an author with no posts gets a redirect to the blog page. Every page of each category or tag list is copied too. A page number the old site does not have, such as /page/9 on a list with one page, answers 404 on Insites, as it did on WordPress. A page number below a page or a post that is not a list, such as /privacy-policy/page/2/, redirects to the page itself, as WordPress does.
Categories, tags and other taxonomiesEach taxonomy the REST API shows for a post type that moves becomes a database table, a row a term, with its name, slug, description, parent and old address. Each item's row lists its terms' names and, beside them, their WordPress ids, so a row can be matched to its terms.
Advanced Custom FieldsEach field of a field group with Show in REST API switched on becomes a column on its post type's table, in a Custom fields group: a number, a true or false, a list, or text. A field with structure, such as a group, a repeater, or an image or link returned as an object, is kept as JSON text. A field group with it switched off is not in the REST API at all, so it cannot be read: switch it on in the group's settings before the dry run.
CommentsApproved comments, and comments waiting for approval, become a Comments table: the comment, the name, email and website the person gave, its status, when it was left, the post it was left on, and the comment it replies to. They can be shown again on the new site. Spam and the trash stay on WordPress, counted.
Menus and widgetsIn every page's header, footer and sidebars, as the site drew them. Each classic menu's links are also copied into a Menu items table, in order, with the menu's name and the theme places it filled, for when the header is rebuilt. Changing a menu after the move means changing the pages.
DesignEvery file the pages load from the site (the theme's and the plugins' styles, scripts, fonts and images) is copied, and the pages name the copies. A file a stylesheet names beside itself, such as a font, is copied too, and every file the pages' scripts fetch as they run is found by loading each page in a browser.
Media libraryEvery file in the media library, the original only, copied to the new site. A smaller copy WordPress made is copied only where a page uses it.
SEOEach page's title, description and sharing tags, as Yoast SEO, Rank Math or any SEO plugin wrote them into the page. Each post's own Yoast SEO title and description are kept on its row, and its page on the new site prints them, with its own address as the canonical one. A tag naming one post's address that cannot be per post (its feed, its oEmbed links, Yoast's structured data) is left out of the layout rather than said on every post's page. The sitemap is the one Insites serves at /sitemap.xml, listing every page that was copied except the not-found page. The site's robots.txt is copied with its own rules, and points to /sitemap.xml.
UsersEvery WordPress user becomes a person with a CRM contact. See People below.
FormsEach Contact Form 7, WPForms, Gravity Forms or Elementor Pro form on a copied page is rebuilt in the Insites form builder. See Forms below.
RedirectsFrom the Redirection plugin, and found by asking the live site. See Redirects below.
WooCommerceProducts, variations, categories, orders and their payments in the Insites Ecommerce module, customers as CRM contacts, and a cart and Stripe checkout built on the new site. See The shop below.

Every row is made with an id from its WordPress id, so a final sync updates it rather than adding it again. A final sync of a site that has not changed writes no rows and copies no files: the redirects the move made keep the ids they were given, a stylesheet is sent again only when its text changed, and a file the old site no longer serves is asked about once, not read again. A file is sent again only when the old site serves it at another size: a script, stylesheet, SVG or font the old site sends compressed, with no size, counts as the same file. The checks class a file the old site no longer serves as broken on the source, not as a file the move lost.

Posts as collections

Each post type's table gets an Insites CMS collection page, with a details layout that draws one item's page from its row, the way Webflow's collections are moved. The tool makes the layout itself from one post's own page: the post's title, content and featured image become the row's, and the rest of the page (the theme's header, footer, sidebar and styling) stays as it was. Where that does not give every post's page back word for word, because the theme prints something else from the post, such as its date in the theme's own words, its categories, or the next and previous posts, the AI pass asks Claude to finish the layout from a few posts' pages, and the proposal is used only when every post's page renders word for word. The AI pass is the same one Webflow collections use: switched on per tool, and within the migration's AI limit.

When nothing settles the layout, you choose on Transform, on Pages and code: use the proposal as it is (the AI's, or the tool's own when the AI is off or over the limit), copy each post's page as a static page at its own address (what happens until you choose, so the site reads the same on day one, though a post added later has no page until a layout is written), or copy no pages and rebuild the layout yourself. The rows move whichever you choose.

Addresses. Where every item of a post type already sits under one prefix (/blog/my-post, /service/whitening), the collection page sits at that prefix and each item keeps its address. When the post type has an archive page at that prefix (/services), the collection page takes its address and shows the archive page, as it was copied, as its list. Where it cannot, because the items sit at the site's root (/my-post) or under a date (/2024/05/my-post), the collection moves to /<post type>/, such as /posts/my-post, and the tool plans a 301 redirect from each old address, which the Redirects check verifies after the load. Categories and tags are kept as lists on each row, not as linked tables; the category, tag and author archive pages are copied as static pages, frozen on the day they were read.

People

Every WordPress user becomes a person on the new site, with a CRM contact. WordPress never gives out a password, so nobody can sign in to the new site with their WordPress one: the To do list asks you to send each person the new site's password reset link after cutover. Nobody is given admin rights on the new site. Give the people who look after the site admin access yourself. The user the tool connected as is not copied. A WooCommerce buyer who ordered without an account becomes a CRM contact by the email on their order.

Forms

A form plugin draws its form into the page and sends it back to WordPress with its own script, so none of them works once the site has moved. Each Contact Form 7, WPForms, Gravity Forms and Elementor Pro form is built the way the Insites form builder builds one, so you edit it in Forms in the new instance's admin like a form made there:

  1. Its database table comes first, named after the form, one column a field.
  2. Then the form, on that table, with the fields the form required.
  3. Then its two emails: a workflow email for each submission, and an autoresponder to the person who sent it. A Contact Form 7 form's workflow email goes to the address its own settings sent to. So does an Elementor Pro form's, with its own subject, read from the form's own settings. The autoresponder is made switched off, so nothing is sent until you have read it and turned it on.

The page keeps the form as the plugin drew it, with the plugin's own classes and labels, so the site's styles still apply, and it shows the plugin's message once it is sent. The plugin's form scripts are taken off the page. A form used on several pages is one form, and each submission records the page it was sent from. A file upload field and a captcha are named on the plan and left off the rebuilt form: switch on spam protection for the form in the form builder. An Elementor Pro form keeps its classes and markup, and Elementor's own script stays on the page for the menus and popups it draws, with the form taken off its handler so it posts to Insites. A form drawn by another plugin (Ninja Forms, Formidable) stays on the page and no longer sends, and the plan names it.

Past form entries

Contact Form 7 and WPForms Lite keep no entries at all, so there is nothing to copy. Where a plugin keeps them:

  • Gravity Forms: read through its REST API, with the same Application Password, when the site has it switched on (Forms, Settings, REST API). Each active entry becomes a row of the form's table. With the REST API off, export them instead.
  • WPForms Pro: read through WPForms's abilities when the site offers a read-only one for entries. Otherwise, export them.
  • Elementor Pro: read from its Submissions, with the same Application Password. If that fails, export each form's submissions as CSV from Elementor, Submissions. Submissions of an old form that no copied page shows still move, each old form into a table of its own named after it, with a column for each field its submissions hold. No form is rebuilt for it and no email is made, and the plan names it as kept as records, the form itself no longer on any page.
  • Flamingo, for Contact Form 7: export them.

To export, use the plugin that keeps the entries (Flamingo for Contact Form 7, WPForms Pro, Gravity Forms) has no REST route for them, so export each form's entries as a CSV (Flamingo's Inbound Messages, WPForms's Tools, Export, or Gravity Forms's Import/Export) and add each file beside its form's table on the Form submissions step of Transform. Each column is matched to a field of the form by its name, and a column that matches no field is named, never dropped silently. A date the export writes without a time zone is read in the site's own time zone, from its WordPress settings.

Redirects

The tool cannot see every redirect a WordPress site has. Rules kept in the server's own settings or in a plugin with no API are invisible to it. When it finds no redirects at all, the Redirects step says that is not confirmed, in amber, and the To do list carries a job to confirm it, until you add a redirects export or tick This site has no redirects and say why. The reason goes in the report. It never stops the load.

Where the Redirection plugin is installed, its rules are read through its REST API. A rule that is a pattern (a regular expression), or that redirects only under a condition, is named and left out, because a redirect page answers one address. The tool also finds redirects from the outside: every old address the Internet Archive has seen on the site, and every link on its pages, is asked of the live site, and each one the site sends somewhere else becomes a redirect. That also finds the redirects WordPress makes itself when a page's address changes. Yoast SEO Premium and Rank Math keep their redirects without a REST API, so export them and add the file on the Redirects step:

  • Yoast SEO Premium: Yoast SEO, Redirects, export as CSV.
  • Rank Math: Rank Math, Redirections, Export Options, as Apache or Nginx. Rank Math writes server rules rather than a CSV; the tool reads both. An exact rule becomes a redirect; a rule that matches part of an address, a rule for a query string, and a "gone" (410) rule are named and left out.
  • Redirection: its own CSV works too, if you would rather add a file than let the tool read the plugin. A rule switched off in Redirection, and a pattern, are named and left out.

A redirect you add wins for the address it names.

The shop

A WooCommerce shop moves into the Insites Ecommerce module, which keeps a shop's products and orders and has no storefront of its own, so the tool builds one on the new site:

  • Product pages keep WooCommerce's design. Add to Cart adds the product, or the variation chosen, to a cart on the new site. A shop or category page's Add to Cart goes to the product's own page. A grouped product says to add each item from its own page; an external product keeps its link to the shop that sells it.
  • The cart is at /cart, inside the site's own header and footer. WooCommerce's own cart, checkout and account pages are not copied, and there is no customer sign-in: everyone checks out as a guest.
  • Checkout uses Stripe Checkout, through the site's own Stripe account. The new site prices every line itself and checks the stock. An order is marked paid only when Stripe confirms it was paid for the order's own total.

Before the shop can take payments, add the site's Stripe keys in the instance admin, under Integrations, Stripe, and in Stripe send the checkout.session.completed event to /shop/stripe/webhook on the new site. Shipping is free, no tax is charged and no coupon is taken until they are set up on the new site.

What stays behind

  • Files the site no longer serves. A theme's stylesheet can name a file the WordPress site itself answers 404 for, such as an old icon. It cannot be copied, so the plan names it, and the Addresses check does not count it against the new site, because the old site's pages asked for the same missing file.
  • Spam and trashed comments. Counted, and left on WordPress. A copied post's comment form no longer sends: the Comments table keeps the comments, and a new form can be built on the new site.
  • Plugins and the theme's code. No WordPress code runs on the new site. What each plugin drew on the pages moves with them. Anything it did when a visitor asked (site search, filters, bookings, memberships, a popup loaded from the server, a builder form) does not, and the plan names each plugin the site had switched on.
  • Shortcodes the old site showed as text. A shortcode from a plugin that is gone shows as text on WordPress, and on the copy too. The plan names each one and its page.
  • Drafts, private and pending items. They are not published, so they are not copied.
  • What a post page shows that is not the post. A link to the next or previous post, a related posts list or a comment count comes from WordPress at the moment it draws the page. A collection page's layout keeps the first post's version of it, and the plan names the difference.
  • Password-protected pages. Only their password prompt is public, so each is named on the plan, for you to rebuild.
  • The fields of a post type the REST API does not show. Its pages still move from the live site, because the sitemap lists them, but its own fields (custom fields, ACF) cannot be read.
  • Other languages. The main language moves. A multilingual plugin such as WPML or Polylang is named on the plan.
  • WooCommerce's shipping zones, tax rates, coupons, product reviews and order emails. Counted where WooCommerce counts them, and set up again on the new site: the To do list says how.
  • Every other site of a multisite network. One site moves per migration.

After the migration

The To do list in Finalize is written for a WordPress site. Each job names what this site has.

  • Domains and DNS. Add the site's domains to the new instance in Insites Console, and replace the records that point the domain at the WordPress host with the ones Console gives. Leave the email records as they are.
  • Forms. Open each form in the form builder: check who its workflow email goes to, read its autoresponder and switch it on, and send a test from the copied page.
  • People. Send each person the new site's password reset link after cutover, and give the people who look after the site admin access.
  • Stripe, shipping, taxes and coupons. For a shop: add the Stripe keys and the webhook, then set up shipping, taxes and order emails again.
  • Plugins. Go through the plugins the plan names, and rebuild what still matters.
  • Search engines. Submit /sitemap.xml in Google Search Console once DNS has moved.
  • The WordPress site. Keep it, and its hosting, until the new site has run cleanly through the 14-day watch. Keep a full backup with its database, close the migration on the Close step, delete the tool's user in WordPress (which revokes its Application Password), switch Wordfence's setting back if you changed it, and end the hosting.

If it does not connect

The tool saysWhat to do
WordPress did not accept this user name and Application PasswordCheck the user's name, make a new Application Password for it and paste it again. If it still fails, the site's server may be dropping the sign-in: ask the host to pass the Authorization header through to WordPress.
Application Passwords are switched off on this WordPress siteSwitch off the security plugin's setting that disables them (Wordfence: Brute Force Protection, Disable WordPress application passwords), make the password, and connect again.
WordPress accepted the password, but this user's role cannot read everything the migration needsGive the tool's user the Administrator role, and connect again.
That address belongs to another WordPress siteGive the address visitors use for the site you are migrating. For a site of a network, give that one site's own address.
Something in front of WordPress answered instead of WordPressA firewall, a security plugin or the host blocked the tool. Ask whoever runs it to let the tool's requests to /wp-json/ through.