Section 8 of 9

Don’t do this

Every mistake on this page has been made by a capable team for a reasonable-sounding reason. The reasons they fail are less obvious than the mistakes.

In plain terms

This is the list to read before you start and again before you launch. Each card says what not to do and why; the link under it goes to the part of the guide that explains the better way.

Structure

Rebuilding Workflow inside Planning

Tasks, assignments, hours, proofing cycles, and resource leveling belong in Workflow. Copies in Planning create two sources of truth and no audit trail. If someone asks for a Tasks record type, ask what Workflow is not giving them first.

Time periods as record types

Campaigns 2026 and Campaigns 2027 break every year-over-year comparison. Use a date or a fiscal-period field.

Five-deep campaign hierarchies

Three tiers covers most organizations; four when they ask. Deeper is fragile to maintain, and the timeline becomes unreadable.

A record type that is really a view

“Calendar” and “Q3 Dashboard” are ways of looking at records. Make them views.

Duplicate primary values

People pick connected records by searching the primary field. Duplicates make the picker unusable — compose a unique value or add a code.

One date field

Timeline and calendar views need two date fields, and records missing one silently disappear.

Read more in Pick a shape and How Planning works.

Platform assumptions

Assuming Fusion

It is bundled at Workflow Ultimate or sold as an add-on. Without it, native automations and Workflow business rules are the whole automation budget.

Assuming AEM Assets or GenStudio

Both need their own license and Admin Console onboarding. Ask before designing an asset flow around them.

Gated features without the package

Cross-workspace connections and global record types are gated above Select. A multi-workspace design for a Select customer wastes a session.

Quoting limits from memory

Package limits, tier gates, and endpoint paths change. Verify against Adobe’s documentation when you build.

Assuming Marketo

Marketo Engage is common at enterprise scale; HubSpot dominates below it; Eloqua, Salesforce Marketing Cloud, Braze, and Klaviyo all have real share. Ask.

Assuming Slack

Microsoft Teams, Outlook, and SharePoint are roughly half of enterprise. Ask.

Data and scale

Pushing past the ceilings

25,000 records per record type, 500 fields and 30 connection fields per record type, 20 formula fields, 200 API requests per minute. Contacts, impression-level data, and large asset catalogs belong in the CRM, the warehouse, and the DAM.

Planning as a DAM

No file field, no versions, no renditions. Connect to AEM Assets, Workfront Documents, or the DAM you already run.

Scattering taxonomies

Brands, regions, products, and audiences repeated per team drift immediately. Define them once — as global record types where the package allows, or in one workspace.

Splitting by region by reflex

Regional workspaces fragment campaigns that ladder up to the same strategy. One Marketing workspace is the default.

Read more in the Quick reference, including the limits calculator.

Scope and defaults

A B2B software frame

Industry diversity is the rule — consumer goods, financial services, healthcare, media, higher education, nonprofits. Ask about audience shape first.

ABM or demand-gen scaffolding by default

Both are niches. Account-based tracking is rare below enterprise; lead pipelines are mostly a marketing automation job.

Localization by default

Rare outside enterprise multi-market teams. Add locale-aware fields when asked.

Budget and PO tracking by default

Enterprise-heavy, and Workfront financial management may already cover it.

Views, forms, and automations nobody asked for

A workspace with 40 views nobody owns is worse than one with three. Scaffold the structure, then hand off the steps that matter.

Skipping scope

Five minutes of questions beats a wrong-shape rebuild. Record types cannot move between workspaces.

Read more in Scope it first and Build layers.

AI

Promising an AI field

Planning has no field type that computes itself with AI. Enrichment runs through AI Assistant, Planning Designer, Fusion AI and MCP modules, or an outside model writing back through the API.

Suggestions in the decision field

Write AI output to Suggested fields. When suggestions and decisions share a field, nobody can tell what the human decided.

An AI that can approve

A compliance pre-flight check writes flags. If it can set the approval field, the review is no longer a human decision.

AI before the data is clean

AI assistance comes last in the rollout — after the structure is stable and the data is worth reading.

Sources

Adobe Workfront Planning documentation — terminology, request forms, automations, object limitations: Experience League article index
Planning API reference, versions 1 and 2: developer.adobe.com/wf-planning
Package contents and entitlements: Adobe Workfront product description