How to add structured data to a website
Learn how to add structured data to a website, verify JSON-LD against visible content, and check the final page for duplicates and errors.
Free app. Runs on your Mac with the Claude or Codex plan you already have.
What it means to add structured data to a website
When you add structured data to a website, you label facts about a page and the things it describes so software can interpret them. A recipe, a business location, and an article each have different facts worth labeling. Schema markup uses a vocabulary such as Schema.org to give those facts consistent names. JSON-LD is a way to express the labels and values as a block of text in the page's HTML.
Read the page, check what the site already publishes, and choose the types that fit the content. Structured data validation covers syntax, vocabulary, and the markup that appears in the resulting page. Rich results are enhanced search listings with additional eligibility requirements. Record the evidence for each check.
What it does for the business
- Give developers a patch they can reviewGive the developer the replacement block and identify the ones to remove. They can estimate the work and decide where to apply and test the change.
- Catch content errors that affect shoppersIf the visible offer and existing markup show different currencies, the comparison gives your merchandising team a concrete issue to investigate. They can trace the value to its content field and fix the source when it is wrong.
- Reuse verified identity across the siteKeep a sourced record of the business name, publisher, and authors. Editors and developers can reuse it for new templates or during a migration.
- Decide where extra SEO work is worth doingSeparate invalid markup, missing content, and feature eligibility in the report. The client can fund a repair that affects many pages, assign a content change where it is justified, or defer a proposal with little use for the site.
How to add structured data to a website, step by step
Inventory the pages and the markup they already carry
Open the page and its source before deciding what to change.
What to do
- List the canonical URL and main purpose for each page in scope. The canonical URL is the address the site identifies as the preferred version. Record redirects and language variants so you know which version you actually inspected.
- Read the visible page, including the byline, price, breadcrumbs, and any content behind a tab a visitor can open. Record the exact text and location of the facts you might mark up.
- Inspect the raw HTML for JSON-LD, microdata, and RDFa. Microdata and RDFa describe entities through attributes on HTML elements. Check the rendered page too if scripts can insert markup after loading.
- Open the home and about pages for identity facts. Sample sibling pages to see which blocks come from a shared template or plugin.
Why
A theme and a plugin can both describe the same author. If an incorrect author URL appears across several articles, the underlying fault may be in the shared template. Reading sibling pages also shows which valid details you need to preserve.
Choose the types that describe what visitors see
The visible purpose of the page determines which entities belong in the markup.
What to do
- Identify the main thing on the page. An editorial article may describe its headline, author, and publication date. A product page may describe the item and its visible offer. Use a breadcrumb trail when the page shows that path to visitors.
- Make a keep, fix, remove, or add decision for every block. Look for the same entity described with conflicting names or URLs, and check whether separate blocks are actually describing different entities.
- List facts already in the markup that you cannot verify from the page, such as a profile link. Record their origin. Resolve contradictions with the owner and keep the evidence difference visible in the handoff.
- Reuse stable entity identifiers. In JSON-LD, @id links statements about the same entity across blocks or pages. For example, reference the same publisher identifier from each article on the site.
- Check the search engine's documentation for the target appearance, the properties it calls for, and any content or access requirements.
Why
A valid type can still describe the wrong thing. A list of products and a page offering one item for sale serve different purposes. Selecting by visible purpose keeps the description accurate even when there is no corresponding search feature.
Draft the block and validate the facts and syntax
A parser can check the syntax while the page provides the evidence for each value.
What to do
- Use a schema markup generator or write the block directly. Start with the page's own headline, image, and other supported values. If you inherit a working block, copy its sound details before making the smallest repair.
- Use full URLs for linked pages and assets. Check that the author URL reaches the person in the byline. Verify that publication and modification dates describe the right events.
- Parse the exact JSON-LD you intend to paste. Then check the vocabulary and any target search feature separately. Read each warning in context and decide whether it points to a missing fact, an invalid value, or an optional improvement.
- Read the block beside the visible page once more. Verify an offer's amount and currency for the item and variant the page describes. Remove newly proposed details you cannot support, such as an inferred rating or stock status.
Why
Each check catches a different problem. A trailing comma can break the block, a valid property can be used on the wrong type, and a well-formed price can describe the wrong variant.
Prepare the patch and verify the affected template
Verify the markup from the same page output that will carry the change.
What to do
- Identify each old block to remove by its location or opening lines. Specify where the complete replacement goes and which unrelated blocks stay. Give the person making the change enough detail to find the right template or CMS field.
- Trace shared values to their content fields. Fix shared logic when the defect repeats, and correct page-specific data in the page record.
- Apply the change in an authorized environment. Open representative affected pages and a page outside the change. Extract the markup from the resulting output so you can verify there are no duplicate emitters and the intended values survived rendering.
- Record the URLs checked, the validation results, and where the change ran. If you only have page access, provide the removal and insertion instructions with the validated draft.
Why
A replacement can pass in isolation while the final page still emits an old block. A shared template can also produce the right value on one page and the wrong author or price on another.
Review the handoff and mark the deployment state
A handoff should show the evidence for every check it claims.
What to do
- Ask a reviewer to parse the exact replacement, verify each changed fact, and spot-check values carried over from the old markup. Give the reviewer the original and proposed versions so they can see what was removed.
- List the replaced block, the reason for the change, and any content correction still needed. Assign issues such as an unconfirmed author URL to the person who can resolve them.
- Record draft, local application, and live verification separately. After an authorized release, open the public URL and repeat the relevant checks on that version. Record when you observed the live page.
- Tell the site owner which checks to repeat when a template or content field changes. Date any observations of search appearance.
Why
A template tested locally can differ from a cached public page. Separate records let the release owner trace a discrepancy to the draft, the template, or the version the public page serves.
Common mistakes
- Adding a second copy of an existing blockThe old block can keep emitting conflicting values when a new one is added through a different field.
- Copying example valuesSample prices, reviews, and author URLs can survive in a draft unless you check each field against the page.
- Using the same entity ID for different thingsA publisher, a person, and a product need distinct identities so references resolve to the right entity.
- Fixing one page from a broken templateA page-level edit leaves the same defect on the other pages the template generates.
- Treating a test as the release recordA check on the proposed block cannot show what a public URL served after deployment.
What a good result looks like
- A page inventory names each inspected URL, canonical address, and source of existing markup.
- Each change points to a visible fact or identifies a value carried over from current markup.
- The replacement preserves working blocks and clearly identifies the blocks to remove.
- The handoff records parser and feature checks and names any pages sampled after a source change.
- The release record shows who owns the next check and which version of the page was verified.
And then what? Using the result
- Put shared template fixes into the development queue and attach sample URLs and the replacement blocks.
- Assign visible content discrepancies to the people who own the affected pages or data fields.
- Have the release owner verify representative public pages and attach the observed output to the release ticket.
- Reuse the verified entity references when briefing the next set of pages and check their page-specific values.
- Review recorded search appearances after release and use them with other page evidence when planning the next SEO change.
When to run it
- You inherit a client website with several markup emittersStart with one page from each important template. The inventory shows where the same entity comes from several sources and gives you a basis for deciding which source owns it.
- You are changing a theme or content management systemRecord working markup on the old pages before moving them. Use that record to check whether the new templates kept the author links, page references, and offers you intended to carry forward.
- A product or article reports a validation issueInspect the affected URL and a sibling page. Decide whether the issue comes from the page's facts or from the way the template emits those facts before asking for a code change.
Questions people ask
How do you add structured data to a website?
Inventory visible facts and existing markup, choose the applicable schema types, and validate the JSON-LD before placing it in the page. Check the resulting output and a sibling that shares the template.
What is the difference between JSON-LD and schema markup?
JSON-LD is a format for expressing schema markup, which labels information using a vocabulary such as Schema.org. A type such as Article identifies the entity. Properties such as headline and author describe its details.
Can valid structured data guarantee rich results?
No. Valid structured data can help a page qualify for rich results if its content meets the search engine’s requirements. The engine decides whether it displays an enhancement. Check eligibility separately from vocabulary validation, and track observed appearances separately from either test.
Should I replace all the structured data already on a page?
Keep sound existing data and repair the blocks that are wrong, redundant, or incomplete. Identify which template or plugin supplies each block first. Confirm that your replacement preserves useful entities and properties before removing any emitter from a shared template.
Where does JSON-LD go on a website?
Put JSON-LD in a script element of type application/ld+json in the page's HTML. For a shared template, generate the values from content fields and inspect the markup on a few resulting pages.
How do you test structured data on a live page?
Open the released public URL, extract its current markup, and compare it with both the approved block and the visible page. Run your vocabulary checker and the relevant feature test against the released version. Save the observed URL and date so another reviewer can reproduce the test.
More jobs in SEO agencies
Add structured data to a website, on your own Mac.Free app. Runs on your Mac with the Claude or Codex plan you already have. Your files stay on your computer.
Download for Mac