How I would modernize campaign operations in 90 days
How I take ownership of a campaign operation, from request intake to launch: what I learn, build, and measure in the first 90 days, and how I set up the AI and automation roadmap that follows.
- Method
- DMAIC: define, measure, analyze, improve, control
- Core metrics
- Cycle time, throughput, rework rate
- Systems
- Work management, marketing automation, and event platforms
- Scope
- Global teams across regions and time zones
Approach
I would structure the first 90 days on DMAIC. Each phase ends with a readout, and no change goes live without a baseline to measure it against. The first weeks answer three questions the rest of the plan depends on: who builds in each tool today, how requests arrive, and what automation is already in place.
| Days | Phase | Focus |
|---|---|---|
| 1 to 30 | Define and measure | Map the current state and set baselines |
| 31 to 60 | Analyze and improve | Find root causes, then design and pilot fixes |
| 61 to 90 | Improve and control | Roll out globally, hold the gains, and build the automation roadmap |
Goals at day 90
- One intake path for all campaign requests, capturing the metadata builders and personalization teams need.
- Published SLAs by campaign tier, with routing rules for each region and clean handoffs across time zones.
- A live operations dashboard tracking cycle time, throughput, and rework rate.
- A governed library of program templates and build standards across the work management, marketing automation, and event platforms.
- A prioritized AI and automation roadmap, with the first automation in production.
Listen, map, and baseline
I would listen before changing anything. The first month builds the map of how campaigns actually move today and sets the baseline every later decision depends on.
Actions
- Meet with requesters, builders, partner operations teams, personalization and journey owners, enablement, and regional marketing leads.
- Shadow three to five campaigns from request to launch across different types: email program, webinar, event, and nurture update.
- Map the current workflow, including every intake channel, handoff, approval, QA step, and time-zone gap.
- Audit the work management workspace and inventory existing program and event templates with the build team.
- Write operational definitions for each metric, then pull historical data to set baselines by campaign type and region.
- Log friction points by source: missing information, late approvals, rework loops, overnight handoff delays, and manual build steps.
Deliverables
- Current-state workflow map with handoffs, time zones, and wait states marked
- Stakeholder map of owners, approvers, and builders by region and function
- Metrics baseline: median and 90th percentile cycle time, throughput, and rework rate
- Ranked friction log and template inventory
Day 30 readout: current state, baseline numbers, top five friction points, and proposed targets for day 90.
Find causes and pilot fixes
I would trace the top friction points to their causes, then design the new intake and routing model with the people who will use it, and test it with one region before rolling it out globally.
Actions
- Run root-cause analysis on the top five friction points, verified with data.
- Design a single intake form with conditional fields by campaign type, covering goal, audience, region, launch date, assets, program template, and data needs for personalization.
- Route a request to the build team only once it is complete, so handoffs across time zones never stall on a missing field.
- Define campaign tiers by complexity, with an SLA for each tier stated in business days.
- Write routing rules by campaign type, region, and builder capacity.
- Standardize QA checklists for email programs and events, with a required reason code for every rework.
- Pilot the model with one regional team for three to four weeks, and build the first version of the dashboard.
Deliverables
- Root-cause summary with the fix for each cause
- Intake form, handoff standard, SLA framework, and routing rules
- QA checklists and build standards for naming, tokens, and versions
- Pilot results against baseline, and dashboard version one
Day 60 readout: root causes, pilot results, changes made from feedback, and the global rollout plan.
Roll out and hold the gains
I would roll the model out to all regions with enablement partners, set controls so the gains hold, and turn measured friction into the automation roadmap.
Actions
- Roll out intake, SLAs, and routing globally, staggered by region.
- Train each region at local times, with recorded walkthroughs and a one-page quick reference.
- Publish process documentation in one place, with an owner and review date for each page.
- Retire legacy intake channels on a set date.
- Launch the governed template library with the build team.
- Set a control plan: each metric gets an owner, a threshold, a review cadence, and a response when it drifts.
- Score every manual production task by time cost, volume, and risk, then put the first automation into production.
Deliverables
- Global rollout across all regions
- Training program and central documentation
- Template library and control plan
- Ranked AI and automation roadmap, with the first automation live and measured
- Day 90 report against baseline, with next-quarter priorities
Metrics
| Metric | Definition |
|---|---|
| Cycle time | Median and 90th percentile days from intake to launch, by campaign type and region |
| Throughput | Campaigns launched per week, by region |
| Rework rate | Share of builds returned from QA, with reason codes |
| SLA attainment | Share of campaigns launched within the SLA for their tier |
| Intake completeness | Share of requests submitted with all required fields on the first pass |
| Handoff wait time | Hours between submission and build start |
| Adoption | Share of requests coming through the new intake path |
I track the 90th percentile alongside the median because the slowest campaigns show where the process breaks. Day 90 targets are set at the day 30 readout, once the baseline exists.
First AI and automation candidates
The final ranking comes from the friction log. These are the likely first candidates:
| Candidate | What it removes |
|---|---|
| Intake brief completion | AI drafts a structured brief from the requester's notes and flags missing fields before submission |
| Automated routing | Assigns requests by type, region, and capacity with no manual triage |
| Program setup from intake | Writes intake data into marketing automation program tokens and event platform fields |
| QA pre-check | AI checks copy, links, tokens, and naming against build standards before human QA |
| Status and SLA alerts | Notifies requesters and owners when work is at risk of missing its SLA |
| Weekly ops summary | Reports cycle time, throughput, rework, and blocked work by region |
The rule for all of them: standardize the process first, measure it, then automate. A person approves anything a customer sees.
Risks
| Risk | Response |
|---|---|
| Regional teams keep using old channels | Give regional leads early input on the form, then retire legacy channels on a set date |
| Incomplete requests stall handoffs | Block routing until required fields are complete, and set a daily overlap window for questions |
| Intake form becomes too long | Use conditional fields and test completion time in the pilot |
| Baseline data is incomplete | Use a four-week manual sample if system history is thin |
| SLAs set without builder buy-in | Draft the tiers with the build team before publishing |
| Automation built before the process is stable | Automate only steps that are standard and measured |
This plan draws on the intake and capacity systems I built at Everpure and Upwork.