AEM Sites Playbook, Chapter 1: Strategy, value, ownership and adoption
What outcome justifies the work, and who will operate it?
November 10, 2026
Start with one content change
Before funding an AEM initiative, follow one real content change through the organization. Choose an example that crosses the boundaries you need to improve: a product correction, localized campaign, updated policy or offer used by more than one channel.
Record who supplies the information, who edits it, who approves it, which systems receive it and how the team knows the delivered result is correct. Include spreadsheets, tickets, manual copying and work performed by agencies. Those activities explain effort and delay that a platform diagram can miss.
Write the problem in observable terms. Instead of improve content agility, write: regional editors must enter the same product fact in two places, reviewers cannot see the previous approved wording, or an author cannot edit a page that still works publicly. The business owner defines the consequence and acceptable result; practitioners demonstrate the failure; the platform owner identifies which part of the system or process needs to change.
The decision at the end of this chapter is what intervention to investigate or fund, who will operate it and what evidence justifies wider rollout.
Use community cases to distinguish different problems
A headless business case needs an operating reason
In the community discussion about a headless business case, the questioner wanted to use AEM for authoring while another platform handled delivery. Follow-up asked whether that meant hybrid delivery and whether a pure headless requirement was valid. The exact deployment was not stated.
Respondents proposed shared content across channels and independent frontend delivery. The author thanked them, but reported no implementation, savings or measured improvement.
The useful decision is whether reusable content and independent consumers justify maintaining the additional contracts: models, APIs, preview, frontend delivery and support. Adobe's headful/headless guidance permits a mixture of approaches. A team does not need to replace its entire website to investigate a specific cross-channel requirement.
A product import needs ownership before an implementation pattern
In the Salsify-to-Content-Fragment discussion, a Cloud Service team described a page-centric workflow reading product JSON from DAM assets. It wanted to move toward Content Fragments, support updates/re-imports and choose an ingestion approach.
The author compared AEM jobs/App Builder, external API-driven ingestion, an external transformation pipeline and live PIM retrieval. Later follow-up said App Builder was being explored; there was no final author-confirmed implementation.
Before choosing a mechanism, decide which system owns each field, what a repeated import may overwrite and who reconciles failures. Another participant's production-volume account can suggest a test, but it is not a sizing benchmark for this team.
An approval problem can survive a platform redesign
In the workflow comparison question, approvers received a page path but wanted to see what had changed. After a version step was added before the participant step, the author found the new version and current page both contained the changes. The AEM release was not stated.
The final response recommended comparing against the last approved version and recording the next baseline after approval. No subsequent implementation result was confirmed.
This calls for an approval-baseline policy and a representative test before architectural change. Adobe documents content version comparison; the team must still decide which version represents approval and how reviewers find it.
A healthy public page can conceal failed authoring
In the blank Author editor discussion, a page worked on Publish while Edit mode appeared grey/blank. The author later reported a viewport-height configuration obscuring the Author overlay and an Author-specific styling repair. The exact release was not stated.
The author's reported cause differed from the initially accepted JavaScript suggestion. For a business owner, the acceptance gap is concrete: public rendering was insufficient evidence that the team could perform its next content update. Include author task completion in readiness; do not turn this styling cause into a universal blank-editor diagnosis.
Choose the intervention that addresses the observed failure
Use this table to select the next investigation, rather than start with a preferred product feature.
Scroll horizontally if needed.
| Observed problem | First intervention to evaluate | Evidence needed before expanding scope |
|---|---|---|
| Content waits for an absent reviewer | Assign approval coverage, escalation and a meaningful baseline | An ordinary reviewer and substitute can complete approval/rejection using the correct comparison |
| Editors repeatedly copy the same facts | Agree a shared source and explicit field ownership | All intended consumers receive the fact; repeated updates preserve editor-owned fields |
| A public page works but editing fails | Reproduce the author task and isolate the component, styling or configuration involved | The intended author can edit/save; public rendering and related tasks still work |
| Another channel needs structured content | Trial a bounded model/API/consumer contract | Editors can manage/preview content and the consumer handles updates, references and withdrawal |
| Existing code depends on unsupported or incompatible behavior | Assess a supported repair, dependency change or staged modernization | Compatibility and regression evidence for the actual source/target deployment |
| Several of these problems occur together | Separate causes and sequence a coordinated improvement plan | Each workstream has an owner, dependency and acceptance result |
Several interventions can coexist. Fixing approval coverage can be worthwhile while an integration pilot runs. A shared-content model can be evaluated without committing to a new frontend.
Compare alternatives against mandatory requirements
Write the requirements before scoring options. Separate mandatory conditions from preferences. Preserving approved local content, enforcing access and completing essential author tasks can be mandatory; a preferred implementation technology is usually an option to assess.
Scroll horizontally if needed.
| Option | Appropriate when | Additional responsibility to account for | Reason to pause |
|---|---|---|---|
| Repair the current process/components | The observed failure is localized and the wider architecture meets requirements | Maintain the fix, training and regression checks | The repair cannot meet a mandatory requirement on a supported configuration |
| Introduce shared structured content or controlled synchronization | Repeated facts or cross-channel reuse create identifiable work/errors | Model/field ownership, consumer contracts, reconciliation and publication | Re-import overwrites local edits or consumers cannot handle the contract |
| Add a bounded headless consumer | A specific channel needs independently rendered content | Frontend operation, preview, delivery, access and instrumentation | No team owns the consumer or its failure response |
| Replatform or migrate more broadly | Supported operation or required capabilities justify the wider change | Content/code migration, compatibility, training, cutover and ongoing support | A narrower option has not been compared, or readiness evidence is missing |
Use the actual contract for license and entitlement assumptions. Include content remediation, translation, accessibility work, integration, author training and ongoing support in the estimate. Identify the owner of each estimate and the uncertainty it contains.
Keep active labor, waiting time, cash cost and delivery capacity distinct. Reducing eight hours of approval waiting does not release eight staff-hours. Reusing staff time does not automatically reduce payroll. Do not count the same time as both cost reduction and additional production capacity without explaining the treatment.
Assign decision rights and operational ownership
Give each decision one accountable role, while recording the contributors and substitutes it needs. The same person may hold several roles in a small team, but the responsibilities still need to be explicit.
Scroll horizontally if needed.
| Decision or responsibility | Accountable role | Evidence they accept |
|---|---|---|
| Desired customer outcome and priority | Business/content owner | A specific task, baseline, deadline and acceptance criteria |
| Architecture and supported operation | Platform owner/architect | Alternatives, compatibility, trial results and support responsibilities |
| Source and overwrite rules for each field | Named business data/content owner | A field map and repeated-update/local-edit tests |
| Approval, rejection and absence coverage | Editorial operations owner | Correct baseline, reviewer/substitute completion and escalation |
| Integration retries and reconciliation | Named integration/service owner | Failure/replay tests, failed-record visibility and an operator procedure |
| Author onboarding and support | Content operations owner | Representative task completion and an owned support route |
| Pilot funding and rollout decision | Sponsor with business/technical sign-off | Benefit assumptions, mandatory results, residual conditions and operating cost |
Do this before the implementation partner leaves. Templates, models, redirects, translations, measurement and support procedures each need an ongoing owner. An unresolved handoff should appear in the rollout decision, not disappear inside a project-complete status.
Worked example of a product-content improvement decision
The scenario, numbers and pilot results below are hypothetical illustrations. No pilot was run for this manuscript, and these are not community benchmarks or Adovo customer results.
Establish the problem and boundary
A team uses AEM as a Cloud Service for a website in four locales. A separately operated partner portal also needs product facts. The product-information system owns factual attributes; regional editors own localized marketing descriptions.
People currently copy facts into the two channel implementations. Approval coverage also causes delays. The sponsor asks whether moving the website to headless delivery will solve both problems.
The platform owner separates the decisions: investigate controlled reuse of product facts, improve approval coverage and retain the existing website rendering during the trial. A new website frontend must earn its own justification.
Count one change as one requested product update through its required locale/channel destinations. Keep that definition constant when comparing effort; do not multiply the monthly volume again by the number of locales or channels.
Scroll horizontally if needed.
| Measure | Illustrative baseline | Proposed pilot criterion |
|---|---|---|
| Individual product-content changes | 32 per month | Use this volume only for a conditional capacity estimate |
| Average active effort per change | 30 minutes | At most 20 minutes for the defined task |
| Median approval waiting time | 8 business-hours | At most 4 business-hours |
| Editor-owned local fields | Preservation not reliably demonstrated | Preserved in every required replay/update test |
| Rejected changes | Need an explicit delivery test | Must not become customer-visible through the tested publication paths |
| Additional ongoing operation | Not established | Estimate and assign an owner before rollout |
Choose eight representative trial changes, including localized copy, a repeated input and a missing/changed reference. This small trial tests the design and work process; it does not establish capacity at production scale.
Make ownership executable
Scroll horizontally if needed.
| Information | Owner | Rule to test |
|---|---|---|
| Product identifier and factual attributes | Product-information owner | Apply approved source updates using a stable identifier |
| Localized marketing description | Regional content owner | An ordinary source re-import must not replace local editorial content |
| Review decision and approved baseline | Editorial operations owner | Reviewers can compare the proposal with the prior approved content |
| Delivery/reconciliation state | Integration and channel owners | Detect failed/older updates and verify the intended revision in each channel |
The technical trial may use Content Fragments and an API consumer if supported by the target requirements. Adobe distinguishes management APIs from delivery APIs; proving one operation does not establish the other. Confirm the interface, credentials, publication behavior and consumer contract before committing to an implementation.
Define acceptance before observing results
The team agrees to test whether:
Step 1. An approved product fact reaches the intended website and portal output.
Step 2. Replaying the same input creates no duplicate and preserves regional descriptions.
Step 3. An older update or failed operation can be identified and reconciled.
Step 4. Rejected content remains out of the tested customer-facing publication paths.
Step 5. An ordinary author and the substitute reviewer complete the work without administrator intervention.
Step 6. References and locale variants remain correct after a change and withdrawal.
Step 7. Active effort and approval waiting meet their separately defined criteria.
The content owner supplies the expected content; engineers capture the results; channel owners verify their actual endpoints. Assign someone to deal with failures after the trial team disbands.
Make the decision from the results
Suppose the illustrative trial produces these results:
Scroll horizontally if needed.
| Check | Hypothetical result | Decision consequence |
|---|---|---|
| Active effort | Average falls from 30 to 22 minutes | Improved, but misses the 20-minute criterion |
| Approval waiting | Median falls from 8 to 5 business-hours | Improved, but misses the 4-hour criterion |
| Local-field preservation | One replay replaces a regional description | Mandatory condition fails; hold wider rollout |
| Rejected content | Remains unpublished in the tested paths | This check passes within the tested scope |
| Additional support | Planning assumption of 2 staff-hours per month | Include in the capacity estimate; validate during operation |
At the assumed 32 changes per month, eight fewer active minutes per change would release 256 minutes, or 4 hours 16 minutes. Subtracting the assumed two monthly support hours leaves 2 hours 16 minutes of potential net capacity. This is a conditional estimate, not realized cost savings. Approval waiting is a separate measure and is not added to it.
The sponsor's decision is to fund a bounded repair/retest of field preservation and approval coverage, while holding wider rollout and the proposed website replatform. The team retains evidence of improvement without declaring the pilot successful against requirements it failed.
Turn the decision into an adoption and review plan
Use ordinary roles
Ask authors and reviewers to complete the changed task.
Own the failures
Give every unresolved condition an owner and review date.
Check live operation
Revisit the benefit and support assumptions before expansion.
Train authors and reviewers on the actual changed task. Ask them to create/correct content, identify the approved baseline, reject a proposal and handle a failed input. Record task failures and support questions; use them to improve the model, interface or guidance.
Complete Worksheet W01 with the problem, baseline, alternatives, field/decision ownership, pilot evidence, operating assumptions and funding/hold decision. Give every unresolved condition an owner and review date. Check the proposed design against the actual deployment; a Cloud Service API trial does not establish a 6.5 or LTS implementation.
The chapter is complete when the business owner can explain the outcome and chosen scope, the technical owner can show the evidence and operating responsibilities, and practitioners can perform the task and handle its failures. Revisit the benefit assumptions after representative live operation before expanding the investment.
W01 Strategy and adoption
Use with chapter 1. Agree what the investment should change before estimating a solution.
Scroll horizontally if needed.
| Record | Entry |
|---|---|
| Customer task and business outcome | [enter] |
| Current process, baseline and recurring failure | [enter] |
| Alternatives including a narrower repair | [enter] |
| Implementation, integration, content, training and operation assumptions | [enter] |
| Content/platform/support ownership and substitutes | [enter] |
| Representative author tasks, training and adoption check | [enter] |
| Benefit measure, observation method and review owner | [enter] |
