What Running Digital for a Presidential Campaign Taught Me About AI

I'm not here to talk politics.

I know how that sounds coming from someone who spent years in the political arena — who ran digital for Mitt Romney, who has written extensively about COVID policy, who tweets opinions that generate friction. But this piece isn't about any of that. It's about what the hardest marketing job in America taught me about building systems that work under pressure.

And why AI just made those systems available to everyone.

Here's what nobody tells you about running digital for a presidential campaign: it's not a marketing job. It's a systems engineering job that happens to involve marketing. The outputs are ads, emails, texts, and landing pages — but the work is measurement infrastructure, creative pipelines, decision frameworks, and real-time feedback loops running simultaneously under conditions where a 0.3% difference in conversion can swing millions of dollars in a single week.

That environment forged three principles I've used in every role since. Now, after 25 years of operating in direct-response marketing, CMO roles, Fortune 500 consulting, and now AI systems — I can tell you clearly: these three principles are exactly what separates CMOs who get results from AI from CMOs who get reports about AI.

That pressure is surgical. It forces you to build creative pipelines where the bottleneck isn't approval or production — it's strategy. Everything else has to be fast enough that strategy is the only delay worth having.

AI does exactly this for every marketing team right now. Not by making strategy faster — that still requires a human with judgment. But by collapsing everything else: first drafts, format variations, compliance checks, image briefs, landing page copy, A/B test scaffolding. The gap between "idea approved" and "creative in market" shrinks from days to hours.

Last week I generated 50 SMS/MMS creative variants for a client, ran them through a scoring model calibrated to their historical yield data, and surfaced the top three for testing — in under 90 minutes. That workflow used to take my team three days. The three days weren't producing better creative. They were producing slower creative.

Speed-to-market is now table stakes. If your AI adoption strategy still involves week-long review cycles for first drafts, you're using AI as a spell-checker. It's a content pipeline.

This sounds obvious until you're in a room where a senior executive has a strong opinion about a headline, and you have to show them the CTR data that says they're wrong. The data wins because the culture of measurement is established before the debate starts.

Most organizations don't have that culture. They have dashboards that nobody uses to make decisions. They have analytics that confirm what people already believe. They have AI tools generating reports that sit in shared folders unopened.

The unlock with AI isn't more measurement — it's measurement at a level of granularity that previously required a data team. I now run creative forensics that would have taken a junior analyst two weeks: yield by hook type, by character count, by send time, by list segment, by signer identity. I can tell you that "anger/outrage" opening hooks outperform "deadline/urgency" hooks by 15x in my direct-response work — not because I believe it, but because I have 42 tested creatives that prove it.

Ruthless measurement means you have to be willing to be wrong. It means the creative you love gets killed by the data. It means your instincts get tested, repeatedly, against outcomes. AI accelerates that cycle from months to weeks to days. The CMOs who benefit most are the ones who want to be wrong fast, not confirmed slowly.

In direct-response fundraising, I work under strict constraints: merge fields must be exact, character counts must be respected, claims must be documentable, urgency framing must be credible rather than tabloid. Violating any of these doesn't just hurt performance — it creates compliance exposure, donor fatigue, and brand damage that takes months to undo.

AI doesn't replace message discipline. It makes message discipline scalable. When I build a creative prompt system, I encode the constraints — the tone guidelines, the compliance rules, the merge field syntax, the character limits — directly into the system. The AI doesn't just generate copy; it generates copy that operates within the system's parameters.

This is the difference between "using AI to write stuff" and "Campaign-Grade AI." Campaign-grade means the constraints are built in, not added as an afterthought. It means you can generate 50 variants without running each one through a manual compliance review. The system is the discipline.

What This Means for You

Most AI content you'll read talks about what AI can do. This piece is about the operating principles that determine whether it works in practice.

Speed-to-market requires that you've rebuilt your creative pipeline — not just added AI to it. Ruthless measurement requires that you've committed to letting data overrule opinion. Message discipline requires that you've encoded your constraints into your systems, not your SOPs.

These aren't AI things. They're operator things. The campaign trail taught me them under conditions where failure was expensive and public. AI just gave me the infrastructure to run them at ten times the speed for a fraction of the cost.

Here's the frustrating truth for most marketing teams: the reason AI initiatives underdeliver isn't the AI. It's the absence of these three operating principles. You can't make AI campaigns work without a campaign mentality — and most marketing organizations are structured for committees, not campaigns.

I'm not here to talk politics. I'm here because the hardest marketing job in America taught me how to build systems that work when the stakes are real, the timeline is compressed, and there's no room for slow. AI just made those systems available to every marketing team — not just the ones with eight-figure budgets and a war room.

The question isn't whether you'll use AI. It's whether you'll use it like an operator.

What I'm Building

I write about this — the specific workflows, the real numbers, the things that failed — in Hello AI and on X. Not theory. Not predictions. Build logs. The stuff I actually shipped this week, what broke, and what the data said afterward.

If you're a CMO, marketing director, or operator who wants AI that actually moves revenue — not AI that generates impressive slides — that's the conversation I'm built for.