WorkFocus
Marketing Ops Blueprint
Designing marketing operations on Adobe Workfront Planning
Ignite Your Content Supply Chain

Workflows · 6 of 10

ROI starts with a clean taxonomy

Most attribution problems are not analytics problems. They are spelling problems: one campaign tagged three different ways by three different people.

In plain terms

Planning makes sure every campaign is labeled the same way, and builds the tracking code from those labels so nobody types it. The performance numbers stay in your analytics tools, and only the totals come back to the campaign.

Enforce the taxonomy; leave the data outside

Planning’s lasting contribution to ROI is taxonomy enforcement. Source, medium, and campaign type become locked single-select fields, and a formula field builds the tracking string from them. Nobody types it, and that alone removes the most common reason attribution breaks.

The performance data itself does not belong in Planning records at event level. A record type holds at most 25,000 records, and Planning has no time-series handling. Land the data in the warehouse or BI tool, and write summaries back to the Campaign or Tactic.

Taxonomy in Planning, performance data outside, summaries backLocked single-select fields for source, medium, and campaign type feed a formula field that builds the tracking string, so nobody types it. Campaigns launch in ad platforms and the marketing automation platform with that string. Event-grain performance data lands in the warehouse or BI tool, not in Planning, where the 25,000-record ceiling and the lack of time-series handling make it a bad home. A Fusion scenario writes period summaries — spend, impressions, clicks, conversions, revenue, ROAS — back to the Campaign or Tactic.Locked single-selectssource, medium, campaign typeFormula fieldbuilds the tracking stringtyped by nobodyAd platforms, MAPlaunch with the stringWarehouse or BIevent-grain performance dataFusion, nightly: spend, clicks, conversions, revenue, ROASCampaign or Tacticsummaries for the periodranked by summary — not the source-of-truth analyticsnot at event grainEvent data in Planning25,000-record ceiling, no time series
Figure 1. What Planning contributes to attribution: the taxonomy, enforced. Blue is Planning, grey is outside it.
Where each part lives

Press Play, or select any box in the diagram, to walk through it one step at a time.

The fields

Taxonomy and summary checklist

0 of 6 added ·

Setting it up

  1. Lock the taxonomy: single-selects for source, medium, and campaign type, with values agreed before launch.
  2. Add the formula field that builds the tracking string, and check the formula-fields article for the current function list first — Planning formulas use their own syntax.
  3. Land performance data in the warehouse or BI tool, not in Planning.
  4. Schedule a Fusion scenario to write period summaries back to Campaigns or Tactics: spend, impressions, clicks, conversions, revenue, ROAS.
  5. Hand off the Campaigns performance view — or the BI dashboard, if that is where the real numbers live.

Summaries, not the source of truth. A ranking built on Planning’s summary fields is a quick read, not the analysis. Say so whenever you show it, and point people to the BI tool for anything they will make a decision on.

Before each launch

  1. Validate taxonomy completeness on live and upcoming campaigns, and flag missing or non-conforming values.
  2. Generate compliant tracking strings for new launches from the formula field.
  3. Rank by the summary metrics present, and surface the top and bottom performers — with the caveat above.
# Launch check

14 campaigns launching this quarter, strings generated from the formula
  11 complete
  3  blocked: no channel connection, so medium cannot resolve

# Last 30 days, from the nightly summary sync
  Spring giveaway (paid social)    4.2x ROAS
  Newsletter relaunch (email)      3.8x
  Brand retargeting (display)      0.6x, below the channel baseline

With AI in the loop

A scheduled scenario can write a short performance narrative per campaign into a paragraph field. Someone reviews it before it reaches a quarterly business review.

Don’t do this

Event data in Planning

Impressions and clicks by day will hit the 25,000-record ceiling, and Planning has no time-series handling. Keep them in the warehouse.

Tracking strings typed by hand

Every hand-typed string is a future attribution gap. Build it with a formula from locked fields.

Free-text source and medium

A text field will hold “LinkedIn”, “linkedin”, and “LI” by the end of the quarter. Use single-selects.

Planning as the dashboard

Planning holds summaries for context. The real analysis belongs in the BI tool.

Limits worth knowing here

Published Planning limits that affect this build. Adobe revises these, so confirm before you design around one.
WhatLimitWhy it matters for ROI
Records per record type25,000Rules out event- or impression-level data in Planning.
Formula fields per record type20The tracking string, ROAS, and pacing each spend one.
Formula expression length50,000 charactersMore than enough for a tracking string — the constraint is Planning’s own formula syntax.
API requests per minute200Batch the nightly summary sync; do not write one call per metric.

Sources