init.news●A FIELD JOURNAL
FOR HUMAN OPERATORS
Applied craft / FIELD GUIDE9 MIN READ

Your best project is invisible if nobody can explain it.

A practical release worksheet for Framer portfolios: give each project a stable address, readable evidence and crawlable paths without flattening the experience.

Get the useful answer ↓

THE USEFUL ANSWER

Keep the expressive homepage. Give every substantial project a stable detail page that explains what it is, who made it, the question or constraint, the method, the evidence and the next relevant path.

In Framer, use the CMS when the record repeats, set unique page metadata, expose the important text in the returned page, use ordinary links, and verify images, canonicals, status codes and structured data before release.

Motion can carry mood; the project record must still make sense when motion is reduced or scripts fail.

A beam of blue light reveals an ivory page inside a large circular concrete aperture.
A useful page must be accessible and understandable before it can be cited.AI-GENERATED EDITORIAL ART

WHAT YOU'LL LEAVE WITH

A Framer control map, a project-record worksheet, a URL-pair diagnostic and an image-release acceptance sheet that separates access, meaning, metadata and observed discovery.

  • Give each project a stable page and each image a specific, honest job.
  • Check public image paths, contextual descriptions and metadata against the same approved work.
  • Access and correct markup are release checks, not proof of indexing, citations or sales.
Give the work a page that explains itA conceptual publishing path from a creative work through an approved project record to a readable page, with a source-return path. It does not represent an indexing or citation outcome./ APPLIED CRAFTA WORKING MODELILLUSTRATION / NOT MEASURED RESULTSTHE WORKCLEAR RECORDREADABLE PAGEKEEP THE EVIDENCE ATTACHED
Give the work a page that explains itILLUSTRATIVE WORKFLOW

01Before the new images go live, check one complete path

Choose one project page and three images: the lead, an explanatory detail and a decorative accent. Give each a job before asking AI to write descriptions. The worksheet below is our proposed acceptance test, not a portfolio test we performed. Keep image access, reader understanding and search outcomes in separate columns.

Swipe or scroll to compare

RecordInspectAccept or hold
Role and permissionProject URL, asset owner, approved use and whether the image is evidence or a conceptual illustration.Hold material without publication permission. Label generated concepts; do not imply they document a finished client project.
Public image pathFetch the exact image URL without login. Inspect the HTML img src fallback as well as responsive candidates.Google documents HTML image discovery and a src fallback for responsive images. A CSS background alone does not provide that path. [4]
Description in contextCompare alt text with the actual image and its job on this page. Read the caption beside the relevant explanation.Describe informative content; use empty alt for decoration. A linked image may need its destination or action explained. Put complex information in nearby text too. [8]
Preview and structured recordCompare the preferred preview and schema image with the visible subject, not just the brand palette.Use relevant, representative imagery. Google discourages generic logos and text-heavy preferred images. Markup must describe visible facts; selection is not guaranteed. [4][9]
Sitemap and crawl accessCheck the image location under the intended page and whether robots rules permit that image URL.An image sitemap is a discovery aid, not indexing proof. Do not rely on deprecated caption, title or license sitemap tags. [7]
Handoff evidenceRecord image URL, page URL, actual check time, result, correction owner and recheck trigger.Keep unknowns explicit. A successful fetch does not establish indexing, an AI citation or a qualified enquiry.

Example, not a client result: a studio uses a generated sculpture to explain a permission boundary. Its caption should call it a conceptual illustration, not evidence that the pictured system exists. A screenshot of an actual interface needs an accurate description of what it shows and permission to expose its contents. The palette can unite the set; it cannot supply missing evidence.

Run the sheet on the three images, then ask a willing reader what each adds to the project explanation. Count corrections and total preparation, review and maintenance time. Stop adding artwork when it repeats nearby information or obscures the decision. Recheck after an export, URL or permissions change. Use the image-handoff guide when provenance must survive delivery; that is a separate check from discoverability.

02Keep the experience. Add a project record.

The practical failure usually appears before the platform choice: the work has no stable page, or the page cannot explain itself outside the homepage sequence. Treat the visual experience and the project record as two views of the same work. One earns attention; the other preserves identity, context, evidence and a path forward.

03Choose the smallest page that can carry the meaning

Swipe or scroll to compare

SituationUseDo not force
One substantial projectA stable project page with a short answer, evidence, credits and related paths.A long essay when a clear record is enough.
Several projects share the same fieldsA CMS collection with a detail page for each project.One giant portfolio page that gives individual work no address.
A technical or material decision deserves explanationA linked process note that answers one specific question.Repeating the same sales paragraph across projects.
Work is private, embargoed or lacks publication rightsA private internal record or an intentionally excluded page.Invented credits, outcomes or permissions to complete the template.

Framer documents CMS detail pages as crawlable pages and supplies automatic sitemap.xml and robots.txt files. It also exposes page and CMS metadata, semantic elements, alt text, video posters, JSON-LD and redirects. [1] Those controls create the opportunity; they do not prove a page is clear, accurate, indexed or cited.

04Copy this project-record worksheet

Swipe or scroll to compare

FieldWhat to publishRelease check
IdentityActual project title, creator or studio, date or period, and the kind of work.The title is unique and still identifies the work outside the homepage.
Useful answerTwo or three sentences: what it is, the situation, and the result or intended experience.A reader can understand the project before interacting with the motion.
Question or constraintThe commission, problem, material limit, audience need, or artistic question that shaped it.The wording is specific and supported by public information.
Method and decisionMaterials, tools or techniques; one meaningful choice and why it was made.Separate the documented fact from the studio’s interpretation.
Evidence and creditApproved images, captions, dates, collaborators and public sources.Every claim and asset has a source or an owner; missing facts remain missing.
Media descriptionAlt text for informative images; empty alt for decorative images; a useful video poster and nearby context.Descriptions explain the depicted work rather than repeat target phrases.
NavigationA crawlable link from the work index and one or two genuinely related next pages.Links are ordinary anchors with href values and meaningful labels.
Release identityStable path, unique title and description, canonical, index choice, status code and redirect history.The public URL returns the intended page without login and old paths resolve deliberately.

05Map the worksheet to Framer controls

Swipe or scroll to compare

Framer controlUse it forWhat still needs judgment
Page or CMS metadataUnique title, description, path and search visibility for each project.Write for the actual project; a generated field is useful only when it stays distinct.
CMS detail pageRepeat the record across a body of work while keeping one address per project.Keep the collection fields small enough that someone will maintain them.
Semantic text and elementsOne clear H1 and meaningful page structure.Visual scale does not decide document order; inspect the published result.
Image accessibilityDescriptions for informative images.Decide whether an image conveys information or is purely decorative.
Video posterA useful preview before playback.The poster cannot replace a written explanation of the work.
JSON-LDExplicitly describe eligible, visible page information.Choose a supported type and keep markup consistent with the page; rich results are not guaranteed.
RedirectsCarry visitors from an intentionally changed old path to the new one.Record migrations and avoid redirecting unrelated pages just to catch traffic.

06Run a source, render and path check

Do three passes on the published URL. First, inspect the returned HTML or a text-only fetch. The title, description, canonical and essential project explanation should be available without relying on a dramatic entrance sequence. Google can render JavaScript, but its own guidance still recommends server-side or pre-rendering for speed and because not every bot runs JavaScript. [2] This is a resilience check, not a promise of indexing or AI citation.

  • Source pass: confirm a 200 response, unique title and description, one intended canonical, the useful answer, approved credits, and the chosen index state.
  • Render pass: use the published page on a narrow viewport, at increased text size, with keyboard navigation and reduced motion. The explanation and next path should survive.
  • Path pass: reach the project through a normal link from the work index. Google’s documentation says links are generally crawlable when an anchor element has an href; scripted click targets are not a reliable substitute. [3]
  • Media pass: confirm informative images use HTML image elements with useful descriptions. Google documents that it can find images in img src attributes and does not index CSS background images. [4]
  • Markup pass: validate any JSON-LD, then compare every field with visible page content. Google describes structured data as explicit classification that can enable eligible rich results; it does not guarantee them. [5]
  • Change pass: if the path moved, test the old URL, the redirect destination, the canonical and the sitemap entry together.

07Use AI where it reduces preparation, not accountability

AI can prepare a first draft from approved project notes: extract candidate facts, identify missing fields, propose an image-description draft, and list claims needing a source. Keep the source note beside each sentence. A person who knows the work should approve credits, permissions, factual claims and the final wording. Do not let a model infer an outcome from an image or turn private production notes into public copy.

Draft instruction to adapt: Using only the approved project notes and asset manifest, prepare the eight project-record fields above. For every factual sentence, include its source field. Put unresolved facts, permissions, credits and contradictions in a separate missing-information list. Do not invent a client, result, quote, date, technology, collaborator or approval. Do not publish, alter URLs or add structured data.

08Pilot three pages before changing the whole archive

Choose three representative pages: one image-led project, one process-heavy case, and one motion or installation page. Record the current technical checks, time needed to understand the work, missing evidence and maintenance owner. Apply the worksheet, then repeat the same checks. Three pages are a diagnostic sample, not proof of search or business impact.

Swipe or scroll to compare

MeasureRecord before and after
Technical completeness200 status, canonical, title, description, index state, sitemap presence, crawlable index link, image treatment and valid markup.
Comprehension checkWhether a willing reader can state what the project is, the constraint, one decision and the creator from the page alone.
Production costWriting, asset preparation, review, correction and ongoing maintenance minutes.
Discovery evidenceFirst-party impressions, visits and qualified enquiries only when those systems are connected; keep citation observations separate.
DecisionKeep, simplify or stop; reason, owner and next review date.

Stop or simplify if the record exposes unapproved material, makes the project less accurate, duplicates existing pages, or costs more to maintain than the decision it supports. Continue when the page becomes clearer, retains its character, and has an owner who can keep the record true. The next action is one project: fill the worksheet, publish only approved evidence, and run the three checks.

09When a second URL appears, compare before you remove it

A Search Console exclusion is a classification, not by itself a broken project page. Google describes redirects and canonical annotations as strong signals, sitemap inclusion as a weaker one, and still makes its own canonical choice. [6] Use the following diagnostic before deleting a representation or changing an index setting.

Swipe or scroll to compare

RecordInspectDecision
Preferred project URLFinal status, returned canonical, page identity, and normal internal link.Keep the public route that carries the approved work.
Alternate URLCompare content, purpose and canonical destination with the preferred page.A duplicate can intentionally point to the preferred page; a distinct project needs its own diagnosis.
Source versus rendered pageCompare canonical and essential text before and after scripts execute.Investigate conflicting identities or missing facts; do not remove motion just because it exists.
Google-selected canonicalRecord the inspected URL, selected canonical, inspection time and available indexing state.If unavailable, write unknown. A local fetch cannot establish Google’s choice.

Try this on one affected pair and one unaffected project. Preserve both observations before making a change, then recheck the same addresses. This is our proposed diagnostic, not a performed portfolio test. Use the linked crawler-log audit for access evidence; access, canonical selection and answer-engine citation are different observations.

FOLLOW THE EVIDENCE

Sources & checks

  1. Google Search Central: Image sitemaps ↗

    [7] Image location and crawl-access guidance, discovery scope, and deprecated image caption, title, license and geolocation sitemap tags.

    CHECKED 2026-11-18
  2. W3C WAI: An alt decision tree ↗

    [8] Text alternatives depend on informative, functional, decorative or complex-image context; complex information needs a nearby text equivalent.

    CHECKED 2026-11-18
  3. Google Search Central: General structured data guidelines ↗

    [9] Structured data must represent visible relevant content; correct markup does not guarantee rich-result selection.

    CHECKED 2026-11-18
  4. Framer Help: Guide to SEO features and tools ↗

    [1] Official Framer documentation, updated 15 September 2026, for metadata, CMS detail pages, automatic sitemap and robots files, semantic elements, alt text, video posters, JSON-LD and redirects.

    CHECKED 2026-10-14
  5. Google Search Central: Understand JavaScript SEO basics ↗

    [2] Official crawling, rendering and indexing guidance; recommendation for server-side or pre-rendering and caution that not all bots run JavaScript.

    CHECKED 2026-11-11
  6. Google Search Central: Link best practices ↗

    [3] Official guidance on crawlable anchor elements, href values and meaningful link text.

    CHECKED 2026-10-14
  7. Google Search Central: Image SEO best practices ↗

    [4] HTML img discovery, responsive src fallback, contextual descriptions and relevant representative preview images; preferred-image selection remains automated.

    CHECKED 2026-11-18
  8. Google Search Central: Introduction to structured data ↗

    [5] Official explanation of structured data as explicit page classification that can enable eligible rich results; no result is guaranteed.

    CHECKED 2026-10-14
  9. Google Search Central: Specify a canonical URL ↗

    [6] Canonical signals and Google selection; URL-pair worksheet is an editorial recommendation, not indexing evidence.

    CHECKED 2026-11-11

Sources stay attached so you can check the argument. Meaningful revisions keep this address.

FROM A GOOD READ TO A USEFUL NEXT STEP

Put one idea
to work.

Pick the problem on your desk. Try the method on the page, or take the whole kit.

The method is public. The kit needs one email address.

FIELD KIT 01 / EDITABLE WORKBOOK / HTML + EXPORTS

The discovery field audit

You have a page that should answer a customer’s question, and no way to tell whether an AI answer can find it or cite it.

Leave with one defensible fix for one page, not a site-wide to-do list.

  • A crawl and page checklist
  • A citation observation log you can repeat
  • An editable audit with CSV export

One page, one sitting. Run it again after your fix.

Get the kitNot available yet

Private downloads are not available yet. No email is collected while this form is unavailable.

The download appears on this page. Nothing is emailed to you, and the optional box stays off unless you tick it.