Review Declared Metadata, Links and Image Records

Alder Seed Library, a fictional neighborhood project, is preparing a spring collection page on its own website. The team has copied an older template and added reservation links, a location map and a decorative divider. Its example below turns declaration conflicts and eight supplied anchors into specific editing questions without labeling any destination working, broken or indexed.

Identify the document before interpreting its declarations

Ask where the source came from and which parts it includes. Alder Seed Library obtains a complete HTML snapshot from its website editing workflow. A body-only paste would still support useful link and image inventories, but it might omit the metadata supplied by the page template. Record that scope before investigating absent declarations. Not found in this source is a defensible observation; missing from the public website is a different claim requiring different evidence.

Use HTML mode for structural inspection. Plain-text mode counts supplied prose without inferring metadata, anchors or images. Writing a URL in a paragraph does not create an extracted link record in text mode, and writing Description followed by a sentence does not create a meta declaration. Obtain the relevant HTML yourself. The optional base is resolution context, not an address the checker will fetch to fill in missing material.

Analysis stays in the browser and does not run supplied scripts, load referenced files or render a live preview. Head, script, style, template and noscript content are excluded from prose, heading, link and image extraction; supported declarations are examined separately. This makes the inventory a reading of the supplied parsed document, not a reproduction of the site’s interactive state. A link inserted later by an absent script cannot be discovered from this snapshot.

Choose the website project label that helps your team recognize the work. All platforms and the nine named social contexts do not change metadata rules or connect to platform accounts. This seed-library page can support announcements across several channels without becoming an audit of any native profile. A useful source note identifies the draft, export date and intended page. Avoid treating a familiar project name as proof that the supplied file is the latest version.

Read title and description records as declared wording

The declarations table records the supported title element and meta description entries, with placement and source lines. Read each extracted value rather than relying only on a missing or multiple flag. Alder’s draft contains one title, Spring seed collection, and two description declarations. One description is empty; the other explains reservation and collection. There are two records to inspect even though only one contains meaningful prose. An empty declaration is not the same situation as no declaration at all.

Locate the settings responsible for the duplicates. One may come from a copied template and another from the page editor. Decide which should own the final value in the website workflow, then obtain a fresh snapshot after the change. The checker cannot choose the correct plugin setting or delete an element. Nor does a declaration outside the head become harmless merely because its words sound appropriate; placement remains a separate review question.

Judge the words against what the page actually offers. Alder’s description promises readers can reserve a packet and see collection arrangements. If the page only explains the project and has no reservation route, that promise needs revision. The tool does not read the offer for factual agreement, provide a keyword score or apply arbitrary description-length pass marks. A useful sentence is accurate and specific, not simply long enough to make an indicator disappear.

Keep search appearance separate. An HTML title is declared text, not proof of the title link currently shown in a search result. A meta description is a candidate summary, not a guaranteed displayed snippet. This inspection does not query search results or measure live appearance. Check those questions through the appropriate live workflow when needed, while retaining the source inventory as evidence of what the supplied draft actually declares.

Review canonical and robots intent without inventing index status

Alder’s older template declares a canonical for last season’s collection page, while another canonical points toward the new spring resource. The multiple-canonical finding makes that conflict visible. Read the raw references and their locations, then ask which resource the team intends this document to represent. Do not let the first-looking or longest-looking address decide the matter. A canonical declaration expresses a preference; it does not prove which URL a search engine has selected.

The checker also inventories supported robots and googlebot meta declarations. In this draft, a noindex value was retained from a preview version. That is a source-level observation worth assigning to the website owner before release. It is not a live indexation verdict. The tool has not inspected the public response, HTTP headers, robots.txt, redirects or a search engine’s stored state. A noindex declaration in supplied HTML does not reveal whether any crawler has encountered it.

Preserve each supported declaration rather than flattening several records into a reassuring conclusion. An additional directive should be read in its own context; do not assume one positive-looking value cancels another record. Missing, empty, multiple and outside-head findings describe different source situations. The owner needs to resolve the intended configuration in the actual website system. This checker does not calculate a complete effective indexing policy from all possible delivery layers.

Write the next action narrowly. Alder assigns Confirm intended canonical and remove preview-only instruction from public draft if appropriate. It does not assign Get page indexed because this report cannot establish the starting condition or guarantee that outcome. After implementation, inspect the delivered page separately. Keep publication decisions, crawler access and current index status distinct from the declarations you have just reviewed; combining them into one pass or fail would discard important uncertainty.

Resolve relative references using the entered page base

Enter the page context deliberately. Alder uses https://alder.example/library/spring/. Against that base, reserve/ resolves to https://alder.example/library/spring/reserve/. A root-relative /visiting/ resolves at the origin root, not below spring. The reference ../catalogue/#beans resolves to https://alder.example/library/catalogue/#beans. These are address calculations, not checks that any route or named section exists.

The trailing slash matters to the path context. A base ending in /library/spring treats spring as the final path segment to replace when resolving a relative child; the slash-ending base treats it as a directory path. Use the exact intended page URL rather than casually shortening it to a hostname. The entered base must be an absolute HTTP(S) page address without credentials. It can supply context, but it cannot compensate for the wrong source file.

Only that explicitly entered base controls relative resolution here. A base tag inside the pasted document is ignored and produces a review note. For example, an old source base pointing to an archive does not silently redirect Alder’s inventory context. If you intentionally want the archive context, enter it yourself and name the review accordingly. Changing the base changes the evidence for resolved addresses and clears primary review exceptions, even when the HTML text stays identical.

When no base is entered, the checker does not borrow one from the source or guess it from the canonical. Relative references need review because they remain unresolved. Absolute HTTP(S) references can still be inventoried, but their internal or external classification is unavailable without the entered origin. Keep original and resolved values together in the handoff. A computed absolute address should never be described as a visited destination merely because it looks complete.

Reconcile the anchor inventory with explicit origin arithmetic

Alder supplies eight anchors. Five references belong to the entered origin after resolution: reserve/, /visiting/, ../catalogue/#beans, https://alder.example/help/ and #collection. Two absolute references use other origins: https://events.alder.example/open-day/ and https://regional.example/growing/. The eighth is mailto:seeds@alder.example. With the entered base, that is five internal HTTP(S) records, two external HTTP(S) records and one non-HTTP(S) reference.

The arithmetic describes eight extracted anchor occurrences, not necessarily eight unique pages. The collection fragment retains the current page path, and two different references could eventually lead to the same place. The checker does not follow redirects or merge destinations by editorial intent. If several anchors repeat one target, inspect whether each is useful in its local context. A repeated address is not automatically needless, and a large internal count does not establish a good navigation experience.

Internal classification uses resolved origin equality with the entered base. The events subdomain is therefore external to this origin even though the organization controls it and the name looks related. Scheme and relevant port differences can also change the origin. Use the label for the defined comparison, not as a judgment about ownership, trust or business relevance. A helpful external resource can serve the reader better than an irrelevant same-origin page.

Remove the base from this example and the evidence changes. Four relative references become unresolved; three absolute HTTP(S) references remain recordable but unclassified; the email reference remains non-HTTP(S). Four plus three plus one still accounts for all eight anchors. There is no justification for filling unknown internal counts with zero. Restore the intended base before making the final handoff if relative resolution is part of the question the editor needs answered.

Use source locations to review anchor intent, not HTTP health

Inspect each anchor’s text, reference and source location together. An empty-anchor finding says the extracted anchor text needs consideration; it does not prove the published control has no accessible name. An image-only anchor may carry information through other markup that this literal text review does not assess. Locate the element and check its purpose in the website workflow. The tool is not a complete accessible-name computation or an accessibility certification.

Non-HTTP(S) references deserve classification rather than a blanket broken label. Alder’s mailto reference is intended to open an email action, not retrieve a website page. A different scheme may be accidental or inappropriate. Review the actual value and the promised action, while recognizing that the checker does not launch it. Invalid or credentialed URL findings likewise identify source questions; they do not establish the state of a remote service or run a security audit.

No destination receives an HTTP request. The report cannot tell you that a reservation form returns successfully, redirects to the right season or accepts submissions. A retained fragment is not proof that a matching target exists. Neither a plausible path nor a clean URL record confirms the content behind it. Keep a separate live-check task for important visitor actions, and record its result outside this local source inventory rather than retroactively describing the tool as having performed it.

For Alder, the editorial question is whether Reserve a packet actually points toward the intended reservation resource. The implementation question is whether that resource works when opened through the delivered page. Assign those as related but distinct checks. Pasted destinations remain non-clickable inventory text inside this tool. Do not expect it to discover buttons driven by scripts, follow a chain of pages or inventory every interaction a visitor could trigger on the finished website.

Review image descriptions without claiming to inspect the pictures

Alder’s source has three img records: a packet-label photograph without an alt attribute, a decorative leaf divider with empty alt, and a collection-location map with a nonempty description. Each record points to an element in the supplied source. The checker does not download those images, see their contents or confirm that they will appear on the delivered page. Three img elements are not proof of three visible images in the final layout.

Missing and empty alternative text need different conversations. For the packet-label photograph, ask what information a reader needs that the surrounding paragraph does not already provide. The editor can then propose appropriate wording based on the actual photograph. For the decorative divider, deliberately empty alt may be the correct choice. Do not add an unnecessary description solely to make an empty-alt review disappear; consider the image’s purpose before making a change.

A nonempty value still needs human reading. A map description saying image or spring collection gives little help with collection arrangements. The inventory can reveal that wording, but it cannot verify whether the description matches the map or whether equivalent directions appear nearby. Read the content and inspect the actual asset through your normal website tools. No source-only alt review can establish complete accessibility, factual image accuracy or suitability for every reader.

Treat source references as references, not a media availability report. A relative image path uses the entered base for resolution, while source base tags remain ignored. The checker does not confirm response status, dimensions, visual sharpness, loading behavior or which responsive candidate a browser would choose. It also does not discover CSS background artwork as img records. If the page appears to contain more pictures than the inventory, investigate how they are implemented rather than assuming extraction proves they are absent.

Keep source findings and live follow-up in separate handoffs

Use one queue for source decisions and another for implementation checks. Alder’s source queue resolves duplicate descriptions, confirms canonical intent and reviews the preview robots value. Its follow-up queue checks the delivered reservation journey and collection information. Only the primary source gets a review checklist. An optional comparison source receives a complete inventory and exact heading multiset difference, not a second checklist. Both inputs share the selected mode and explicitly entered base.

Mark a primary finding reviewed after considering it, not as a substitute for editing. The report keeps that status alongside the finding. Source, mode or base changes clear review exceptions. Per-source extraction caps are 200 native headings, 400 links, 200 images and 400 metadata records; the combined primary checklist cannot exceed 500 findings. An exceeded cap causes explicit refusal rather than silent truncation. Split genuinely separate documents and describe each scope instead of treating partial input as a full-page result.

Save JSON for both raw sources and compatible project state. Blank-source drafts can be saved, but inspection requires nonempty primary content. There is no autosave. CSV keeps all extracted inventory for both sources plus primary findings; copy checklist provides the primary task list. PDF contains complete extracted records, scope and primary findings with status, not raw source or a rendered website. Above 100 pages or 40 MiB, use JSON or CSV rather than an incomplete PDF.

After saving, return to the website workflow to implement and inspect the decisions. The heading comparison guide explains the separate literal outline check, and all free tools offers other preparation options. Checker exports do not publish a page or automatically enter SMTasker. Once the resource is ready, prepare supported social activity separately, using accurate descriptions of what visitors will find rather than treating source approval as proof that a campaign destination is live.

Questions about Metadata and Link Review

Continue with another content review guide

Inspect your supplied draft in the Content Structure Checker, or choose a focused guide below. More browser-based helpers are in the free tools directory.