Content Architecture Workbook: turn topics into a decision network

1. Start with decisions rather than word lists

Collect concrete situations in which a potential customer must make a decision: recognise a problem, compare approaches, justify a solution internally or prepare implementation. Write the question in that person’s words. Add what they should know afterwards and which action that knowledge enables. A topic without a recognisable job stays in the idea backlog for now.

The fictional company Übergabewerk offers consulting and software for handing projects over to operations. One starting question is: “Which documents are missing before operations takes responsibility?” This creates a testable information task. “The future of digitalisation” is too broad: it identifies neither the intended reader nor the problem the page resolves.

2. Inventory existing content around one primary question

Record existing URLs, their main question, update needs and available evidence. Do not assign quality based on length alone. A short completed example can solve a task better than a lengthy definition. Also assess whether a topic deserves a separate page or would work better as a section within another page.

In its fictional inventory, Übergabewerk finds three articles about handover checklists. Each defines the same term and none provides a usable list. The problem is an unfinished reader task rather than insufficient publishing volume. The team selects one authoritative main treatment and plans to consolidate useful details there or on clearly differentiated pages. Technical URL changes require separate preparation.

3. Give the pillar a clear navigation job

A pillar explains the overall problem, identifies the relevant decisions and helps readers find a starting point. It need not repeat every detailed answer. Define its boundaries: which terms does a newcomer need, which decision paths exist and where does someone with a specific task find the right detail? Build the outline around those requirements rather than an artificial word target.

The fictional pillar “Handing projects over to operations” outlines responsibilities, documents, acceptance and ongoing operations. It connects to the checklist, meeting agenda and tool-selection guide. A reader preparing a meeting should reach that task directly. Link wording and introductory context therefore need to explain the destination’s practical value accurately.

4. Separate pages by task and required evidence

Two similarly worded search terms do not automatically require two pages. Create a separate page when the reader, decision, deliverable or required evidence materially differs. If those elements remain the same, improve the existing treatment instead. This prevents multiple pages from competing to perform one editorial job and saves readers from moving between repetitions.

In the example, a checklist is a different deliverable from a meeting workflow, so both receive separate pages. “Project handover checklist” and “checklist for handing over projects” describe the same task and are combined. This is an editorial rationale; observed search behaviour and customer questions may later justify a different division.

5. Design links as justified next steps

An internal link answers a follow-up question. For every important connection, record where the reader is, what is missing and why this particular destination helps. A service link fits when self-service reaches a limit or specialist implementation becomes relevant. It does not need to appear in every paragraph.

Übergabewerk connects the checklist to the meeting agenda when several departments must resolve outstanding points together. Its consulting link explains how a facilitated handover workshop can clarify responsibilities and missing evidence. Linking indiscriminately to every service would not offer comparable value. Check the reverse direction too: a detailed page should show how it belongs within the pillar.

6. Write a brief that makes the answer testable

Every brief identifies the primary question, reader, current alternative, required concepts, supporting evidence and a testable deliverable. State explicitly what the page will not cover. The author can then choose useful examples and graphics independently. Without these boundaries, briefs easily become lists of arbitrary subheadings.

For the fictional meeting agenda, acceptance means a project lead can prepare a session, identify necessary participants, record decisions and assign unresolved issues. A general essay about collaboration does not satisfy that task. The brief requires a completed example agenda, a blank template and a statement explaining which assumptions need local adjustment.

7. Build evidence and clear answers into the content

Answer the main question early, then explain conditions, limitations and alternatives. Separate your professional judgement from substantiated facts. For checkable external claims, point to the actual source and preserve its context. An explainable example is more valuable than an unsupported success figure that merely imitates authority.

In the fictional case, Übergabewerk labels every completed template as an illustrative example. A table explains which documents to review and which decision depends on them. Descriptive headings, definitions and useful link wording improve human navigation and give the material a clearer structure. None of this guarantees selection or accurate representation by a search engine or language model.

8. Choose media according to the task

Use a diagram when a sequence or dependency is hard to explain in paragraphs. Use a table for comparisons, a completed template for application and a screenshot only when the actual interface matters. Decorative images can provide breathing space but cannot replace explanation. Important information conveyed in a graphic also needs an understandable textual equivalent.

The fictional architecture gives the pillar an overview of handover phases. The checklist shows completed inspection items; the meeting page shows an agenda with decision ownership. Distinct media emerge from distinct tasks. Repeating one symbolic graphic everywhere would obscure those differences. Include descriptive filenames and appropriate image descriptions in editorial acceptance.

9. Prioritise gaps by usefulness and delivery capacity

Assess a gap through three questions: how important is the decision to suitable prospects, is there enough reliable material to answer well, and can the team maintain the content? High assumed demand without expertise may justify research, but it is not yet a sensible near-term publication assignment.

In the example, Übergabewerk prioritises the checklist because support and consulting can already provide concrete inspection items. A broad software comparison waits for defensible criteria and current information. Record the rationale in the plan. An unexplained priority is easily displaced by the next spontaneous topic request; a documented decision can be consciously revisited.

10. Organise updates and consolidation

Assign a subject owner and a review trigger to each page. A changed process, recurring questions or an outdated source can justify review. Adding a newer date is not an update. Record which statement changed and whether linked pages need revision as well.

Before consolidation, the fictional team checks that useful information will not disappear. It then plans internal links, navigation changes and technical redirects. A URL change is a separate implementation task with verification rather than an incidental editorial action. Keep retired material visible in the inventory so old links and previous decisions remain traceable.

11. Run the architecture workshop on a bounded scope

In the first session, address one decision path rather than the entire topic universe. An adaptable agenda allocates fifteen minutes to reader and task, twenty to existing material, twenty to the proposed page map, fifteen to linking rationale and twenty to briefs and responsibilities. These timings are working assumptions, not requirements.

Ask sales to bring an actual conversation sequence and subject experts an existing working document. The workshop ends with a page map rather than twenty aspirational headlines. Every proposed page has a main question, necessary evidence and a next step. Assign research for unresolved concepts before turning them into production assignments.

12. Accept the architecture through a task walkthrough

Select a typical entry question and walk through the planned path with someone uninvolved. They should explain which page helps first, which follow-up question remains and which link continues the journey. If several pages promise the same task or an essential step is missing, correct the architecture before production.

The plan is concrete enough when every page has a distinct job, important transitions are justified and evidence is available or recorded as a gap. After publication, observe questions, actual use and relevant enquiries. Individual rankings are only one possible signal. The central editorial test remains whether visitors can handle their task more clearly and completely.

YOUR WORKSHEET

Editable page and linking plan

Entries illustrate the explicitly fictional company Übergabewerk. Replace them with your own material and add specific evidence, existing URLs and owners. Every row must have a distinct reader task.