From content velocity to campaign confidence

Readiness was reconstructed, not shown.

Operators opened Workfront for tasks, AEM for assets, and review for approvals, then rebuilt a mental model of launch risk. Often in a spreadsheet or Slack thread the night before go-live.

Late surprises clustered after “approved.”

A master offer or hero change after approval invalidated localized or already-routed experiences, with no clear impact map. The team found the fallout during activation prep, not at change time.

Legal and brand would not accept AI as the decider.

Stakeholders were open to AI summaries and anomaly flags. Compliance review and regional overrides had to stay human-gated. That constraint became the principle “AI advises; policy governs; humans decide” before any UI existed.

Context died at tool boundaries.

When ops routed a blocker into Workfront or AEM, the campaign, market, channel, and reason it was launch-critical often had to be explained again. Carrying context across tools became a requirement.

“I can tell you every asset’s status. I still can’t tell you if we should launch on Thursday.”

Paraphrased. That question is the one the Command Center has to answer first.

Exception-first readiness beats a completion percentage.

Shifting the operator’s mental model from “how complete are we” to “what blocks launch, and what depends on it” should cut time to a launch decision under pre-launch pressure. The concept test below checked this.

Inventory dashboard vs dependency-aware readiness

I chose exception-first readiness over complete inventory. See direction 01 in How the model evolved.

AI copilot as primary UI vs structured decision surfaces

I chose structured decision surfaces, with AI as an advisor for summaries and anomaly explanations. See direction 02 in How the model evolved.

Full dependency graph vs progressive disclosure

I chose progressive disclosure: Campaign → Market → Channel → Experience → Asset/Version. See direction 03 in How the model evolved.

One-click “update all” vs impact-gated apply

Scenario: the master headline changes from “Save $100” to “Save 25%.” A single update-all would be fast and dangerous. Regional offers, prior approvals, and already-activated experiences carry different risks.

I chose Change Impact as its own screen: classify safe, needs review, local override, activated, and unknown, then apply only what policy and people allow. Slower, but nothing breaks silently.

Automated readiness score vs visible ambiguity

A single AI readiness score would feel decisive. It would also create false confidence when data is stale or incomplete.

I chose readiness by dimension, with data freshness and unknown states shown. Automate deterministic, low-risk actions; escalate ambiguity. The scoring model stays a cross-functional definition, so I didn’t invent an algorithm for this case.

“Launch is in three days. Are we ready?”

5 / 5 unaided · 18 s average · confidence 4.8 / 5

No errors. One participant paused to compare the 99.7% asset figure with the 78% launch readiness score before answering.

“Usually I’d have to ping three regional leads on Slack to know if we’re actually good for Thursday. Seeing 78% right next to the 3 blockers immediately tells me we’re not ready without me having to guess.”

“Why is France · TikTok blocked, and what is affected?”

5 / 5 unaided · 42 s average · confidence 4.6 / 5

One participant opened campaign settings first, then found the blocked cell in the matrix.

“The side drawer giving me the root issue, owner, and the exact 8 affected experiences in one view is huge. I don’t have to open Workfront in another tab to figure out who owns the legal hold.”

“The global offer changed. Which experiences need review?”

4 / 5 unaided · 1 min 15 s average · confidence 4.2 / 5

One participant needed a prompt to find the Change Impact entry point. Two hesitated over the difference between “Local override” and “Previously approved.” This was the weakest task, and it shaped the changes below.

“Normally a copy change like this terrifies me because someone hits ‘apply all’ and wipes out custom French pricing.”

“Apply the change without overwriting regional work.”

5 / 5 safe · 35 s average · confidence 4.9 / 5

No unsafe applies. The impact gate kept regional overrides out of the bulk update for every participant.

Asset completion and launch readiness read too alike

In Task 1, the 99.7% asset figure drew attention before the 78% launch readiness score.

I changed the hierarchy. Launch readiness leads as the large card with the critical blockers beside it. Asset readiness became a secondary figure that reads “1 asset pending.”

The impact classes needed definitions

In Task 3, participants paused to work out how “Local override” differs from “Previously approved.”

Next iteration: inline definitions and an icon for each class, so operators know why an experience is flagged before any bulk action. Not yet in the prototype.

Reviewers had to see why it was urgent

Participants wanted proof that the regional legal reviewer in Workfront would understand the urgency without a follow-up message.

I changed the handoff panel to lead with “Context preserved”: campaign, market, channel, owner, and due date travel with every request, followed by why it was routed.

Treating readiness as a derived property of the campaign, not a progress bar. Once the screens followed the operator’s questions in order, the scope stayed small: five screens inside an ecosystem I deliberately didn’t redesign.

Whether operators trust a readiness answer when data freshness is low. Whether critical-only mode hides something they need. Whether legal reviewers find the routed context complete enough to act without asking again.

The readiness scoring definition. It belongs to operations, legal, and engineering together, not to design alone. My next step would be a draft model reviewed with all three before building further.