AEM Sites Playbook, Chapter 4: Authoring, components, templates and design systems
Can authors create the intended experience with the supported components?
November 10, 2026
Define the author's choices
Use this chapter when creating a component library, improving an editorial experience or accepting an upgrade. Start with the actual templates, component versions, author roles and content examples. Designers and engineers define the available choices; content owners decide which choices authors need for their work.
Maintain a component catalog with its purpose, allowed variants, content fields, responsive behavior, accessibility responsibilities and examples. An author should know when to use a component without understanding its implementation. If teams repeatedly build the same arrangement differently, decide whether that variation serves a genuine local need or reflects missing guidance.
Templates, policies, initial content and shared structures have different effects. Adobe's editable-template guidance distinguishes them; initial-content changes concern newly created pages, while changes to referenced template structures can affect existing pages. Identify that impact before approving a shared change. The multisite template discussion contains competing reuse proposals, not a universal /conf layout.
Test the next edit as well as existing rendering
In the blank-editor discussion, the public page worked while authors could not edit. The original author eventually reported a viewport/overlay styling cause, different from the accepted diagnosis. Treat the ability to perform the next content change as a release requirement.
Build a representative authoring script:
Step 1. Create a page from a supported template and complete its required properties.
Step 2. Add permitted components, enter realistic content and exercise validation.
Step 3. Reorder, copy and remove content; check references and responsive layouts.
Step 4. Open nested dialogs, save, reopen and confirm the stored values.
Step 5. Preview the intended experience and submit a change for review.
Use a normal author identity and supported browsers/devices from the project's requirements. Record the content fixture and component release. An administrator completing a simplified page does not establish that the intended author can do the same work.
Keep customization and author guidance together
When extending Core Components, check the compatibility table and document which behavior you changed. Include dialog labels, defaults, errors and keyboard behavior in the change review. A new option can create support and training work even when it leaves old content untouched.
The nested-multifield upgrade discussion includes reports about focus failures and a private Coral-method override. Use it to select a regression task, not to standardize an unverified UI patch. Reproduce the issue on the target release and obtain a supported remedy before adding a maintenance dependency.
For a hypothetical campaign page, permit the author to choose approved image treatments and headings while preserving required brand structure. Test long translations, missing optional fields and realistic content lengths. A component that works with short demo text may still make a regional editor's task impractical.
Decide which choices belong to authors
Community discussions raise a practical design question: how much authoring structure should brands share? The proposals in the example are not accompanied by a deployed outcome. Make brand independence a deliberate requirement. A shared component need not imply identical page structure, and a new brand need not justify a fork of every component.
Translate design policy into choices an editor can understand:
Scroll horizontally if needed.
| Choice | Likely owner | Test before accepting |
|---|---|---|
| Required brand structure or disclosure | Design/content governance | Existing and newly created pages retain the required structure |
| Allowed component variants | Design-system owner | Variants are understandable and meet supported layout/accessibility requirements |
| Page-specific copy and media | Content owner/author | Fields, validation and references support realistic content |
| Brand-specific template policy | Brand/platform owners | A change has the intended brand scope and no accidental cross-brand effect |
| New capability outside the catalog | Product/design/engineering owners | A genuine requirement, maintenance owner and regression scope exist |
Test both restricted and permitted choices. An interface that hides a required option can block publishing just as effectively as one that permits uncontrolled variation.
Worked scenario of a shared product-page template
This is a hypothetical implementation and set of observations. Two brands share product components. Brand A needs a comparison section; Brand B must omit it. Regional teams need longer translated titles. A common footer disclosure must remain consistent.
The team considers one template with explicit policies, separate brand templates sharing components, and entirely separate component libraries. It compares the smallest option that supports the required differences:
Scroll horizontally if needed.
| Candidate | Main question | Maintenance consequence |
|---|---|---|
| One template with controlled options | Can authors reliably distinguish the intended brand choices? | Central policy ownership and cross-brand regression testing |
| Separate brand templates with shared components | Are the structural differences meaningful and stable? | Separate template ownership, shared component compatibility |
| Separate component libraries | Is behavior genuinely incompatible across brands? | Duplicated fixes, guidance and upgrade work |
The team selects a trial of shared components with separate brand templates. The choice is conditional on testing existing pages and actual author tasks. It records which changes affect template structure and which affect initial content, using the editable-template boundary rather than assuming every later template edit is inherited identically.
Representative authors then perform the following:
Scroll horizontally if needed.
| Task | Hypothetical result | Decision |
|---|---|---|
| Create a Brand A page and add comparison content | Works with an ordinary author role | Pass within the tested configuration |
| Create Brand B content | Comparison option is unavailable as intended | Permitted-choice scope appears correct |
| Enter a long translated title | Title overlaps a control at one required width | Hold component/layout acceptance |
| Edit an existing page after the template change | Existing disclosure remains correct | Retain evidence for this existing-content path |
| Reopen a nested product dialog | Values persist but focus leaves the working area | Hold the affected author task |
The decision is to repair the two demonstrated task failures before broad rollout. Adding a new component fork would not explain either observation. The fixture becomes part of the release set so a later shared change is tested on both brands.
Evaluate an authoring incident without premature redesign
The blank-editor case and multifield focus reports show different author-facing symptoms. Their particular causes and release contexts differ. Record the exact component, content and interaction before choosing a styling change, supported hotfix or replacement. Preserve the author's contrary follow-up when an accepted suggestion is not the reported repair.
For a focused investigation, compare a minimal page with the failing fixture, retain the same user/browser conditions and add content incrementally only where that helps isolate the issue. Record the first divergence and which control or overlay prevents completion. A console message that appears at the same time is a hypothesis until evidence connects it to the task.
Give the component catalog operational meaning
Each catalog entry should include an author-facing purpose, field examples, allowed variants, reference behavior, responsive constraints and a support owner. Show a realistic completed component, a validation failure and the route to request an unsupported variant. Keep the examples aligned with current component versions.
The content owner accepts whether the task can be completed; design and accessibility owners assess the result; engineering accepts the maintained implementation. A screenshot of the ideal page cannot substitute for those separate checks. Record temporary limitations explicitly, particularly where a normal author requires someone else to complete an essential step.
Accept the authoring capability
Use W03 for component ownership and W04 for task results. Content owners accept usefulness; design and engineering owners accept the supported implementation; representative authors confirm the workflow. Send keyboard, screen-reader and other inclusive-use checks to chapter 16. Record unsupported variants and a route for requesting new ones so authors do not solve gaps through uncontrolled markup or duplicated components.
W03 Engineering and customization inventory
Use with chapter 3. Give each dependency or extension a requirement and an owner.
Scroll horizontally if needed.
| Record | Entry |
|---|---|
| Module/component/service and requirement served | [enter] |
| Owner, dependency version and compatible target | [enter] |
| Extension point, configuration and content scope | [enter] |
| Service identity, input contract and secret mechanism by reference | [enter] |
| Request trace and valid/invalid/denied test cases | [enter] |
| Disable/uninstall/rollback approach and regression evidence | [enter] |
W04 Authoring acceptance
Use with chapter 4. Test real tasks using the intended author's authority.
Scroll horizontally if needed.
| Record | Entry |
|---|---|
| Template/component, intended use and permitted variants | [enter] |
| Content fixture including long/localized and missing optional values | [enter] |
| Create, edit, validation, reorder/copy/delete and save/reopen steps | [enter] |
| Responsive preview, references and approval handoff | [enter] |
| Intended author identity and observed task result | [enter] |
| Accessibility/regression record and remedy owner | [enter] |
