How Southwest Marketing Mapped AI Bets Across 8 Functions
A 70,000-person airline's marketing team ran Double Diamond design thinking across 8 functions — codifying five signals that flag where AI actually pays off.
70,000-Person Scale
AI signals mapped, 8 teams
Nicola Smith
Senior AI Programs Advisor
Southwest Airlines
Connect ↓
The Challenge
Southwest Airlines' marketing team relied on largely manual go-to-market campaign processes, including repetitive copy-paste actions between tools and documents, inconsistent templatization, and limited visibility into where friction was slowing execution. The AI Programs team needed a structured method to identify which parts of the campaign process were genuine candidates for AI or automation — without defaulting to applying AI everywhere indiscriminately. The challenge was surfacing the right problems before jumping to solutions.
What They Built
Nicola's team applied the Double Diamond Design Thinking Framework to conduct end-to-end discovery across Southwest's marketing function. One-on-one interviews and cross-functional workshops with representatives from social media, paid media, brand, creative, strategy, analytics, digital, and technology teams surfaced pain points and validated findings. The team codified a set of 'signals' — indicators that a process area is ripe for AI: available data or historic assets, repetitive tasks, existing templatization, reusable technical infrastructure, and high human error rates. These signals formed a reusable assessment framework now being applied to future AI opportunity identification across the enterprise.
Nicola's team began by establishing a disciplined separation between problem identification and solution design — a step that prevented the common failure mode of applying AI to the wrong processes. Using the Double Diamond Design Thinking Framework, the team conducted one-on-one interviews and cross-functional workshops with representatives from social media, paid media, brand, creative, strategy, analytics, digital, and technology teams across Southwest Airlines.
Interviews were designed to surface where work was slow, repetitive, error-prone, or dependent on manual copy-paste actions between systems. Workshop sessions validated and pressure-tested findings across functional boundaries. From these sessions, the team codified five signals that reliably indicate where a process is ripe for AI: availability of data or historic assets, repetitive task structures, existing templatization, reusable technical infrastructure, and zones of high human error. The resulting framework now serves as a reusable assessment tool for future AI opportunity identification across the Southwest enterprise, giving the AI Programs team a structured, repeatable method rather than a one-off analysis.
AI Role
In this engagement, AI itself is not yet deployed operationally — the AI role is prospective. The team used structured human research methods (interviews and workshops) to identify where AI could be most effectively applied, codifying a set of signals that indicate process readiness for AI or automation. The output is a prioritised set of AI opportunities, not an active AI system.
Impact
Structured Signal Framework for AI Opportunity Identification
The team codified five 'signals' that indicate where AI could add value in a process: available data/assets, repetitive tasks, templatized deliverables, reusable technical infrastructure, and high human-error zones. This framework is now reusable across future Southwest AI initiatives.
Cross-Functional Alignment Across a 70,000-Person Enterprise
By conducting user research with representatives across marketing, tech ops, ground ops, digital, analytics, and brand, the team built ground-level advocacy before any tool was built — converting research participants into future adoption champions.
Validated Design Thinking as the Repeatable AI Assessment Process
The Double Diamond framework proved effective for AI use-case discovery, giving Southwest a repeatable, mature process for evaluating AI opportunities — replacing ad hoc 'AI hammer looking for a nail' approaches common in enterprise AI teams.
Implementation Complexity
The implementation is a structured research and facilitation engagement — interviews, workshops, and framework development — requiring no software development, technical integration, or AI model deployment. The primary inputs are time, cross-functional access, and facilitation expertise.
Best Fit For
Enterprise organizations with large, distributed workforces that need a structured, research-grounded approach to AI adoption — particularly where leadership wants to move beyond 'AI for everything' toward strategic prioritization. Well-suited to organizations in transportation, hospitality, retail, or any sector with complex internal marketing and operational workflows.
