TL;DR
The marketing leader’s job is changing faster than the annual planning cycle can track. If I were still in the seat at a mid-size credit union or community bank, three projects would be on my list this quarter. An internal AI assistant the team actually uses. An attribution stack that stops pretending the funnel is intact. A content engine built to publish for AI first and humans second. Ninety days each. All achievable with a small senior team and a moderate budget.
I spent thirteen years owning the marketing function at a mid-size financial institution. Brand, demand gen, digital, content, channel, analytics. The whole machine. I left at the end of last year to start an advisory firm, and the question I get most often from old peers is the same one I would be asking myself if I were still inside.
“If you were back in the seat today, what would you build first?”
It is a fair question. So here is the honest answer. Three projects. Roughly ninety days each. All achievable in a regulated environment. None of them theoretical.
Project one: an internal AI assistant the marketing team actually uses
Not a chatbot on the website. Not a co-pilot in a CRM. An internal assistant that knows your brand voice, your product set, your compliance redlines, and your style guide. Something the team uses for first-draft member email copy, blog ideation, paid search variations, social captions, and the seventy other small writing tasks that eat a marketing department’s week.
This is not exotic. It is a structured prompt library, a small library of approved examples, a couple of guardrails, and a thirty-minute weekly review of what the team produced.
Most of the heavy lift is documenting the voice and the redlines in a way that an LLM can actually use. That part is invisible work. It is also the part that decides whether the assistant feels like a productivity gift or a babysitter.
Why this one first? Because it produces measurable hours back to the team in week two. That gives the function the political cover to do the bigger work in projects two and three. AI inside marketing is a culture change before it is a technology change. The team has to feel it work before they trust the bigger conversations.
Project two: attribution that stops pretending the funnel is intact
The funnel is broken. Not partially. Structurally.
Members and customers are getting their first impression of your institution inside an AI chat. They are doing rate comparison inside an AI search. They are not clicking through to your website to start a journey you can measure. The journey is happening upstream of you, and your existing attribution stack is sitting downstream watching tumbleweeds.
The fix is not a new vendor. It is a new posture. You stop trying to attribute every member acquisition to a measurable touch and you start measuring whether your brand is showing up at the right moments in the right ways.
That looks like a quarterly AI visibility audit. It looks like a measurement framework that separates demand creation from demand capture, where demand creation is something you invest in and measure with brand and category metrics, and demand capture is the only place clean attribution still works.
Most marketing leaders are still buying attribution tools that promise to model the broken thing. Walk away. Build the new layer that measures what is actually happening, and let the broken thing keep its smaller share of the picture.
Project three: a content engine that publishes for AI first
This one will feel uncomfortable. Most marketing leaders have a decade of muscle memory built around content for humans, optimized for search, then distributed for social. The model is flipping.
AI tools are now the first reader of a meaningful share of your content. They are deciding what gets recommended, what gets summarized, and what gets cited. The page that ranks third in Google barely matters anymore. The page an AI tool cites when a member asks a question is the page that wins.
What that means in practice: your content has to be specific, named, and structured in a way that AI tools can lift cleanly. Generalities lose. Vague brand voice loses. The page that says “we offer competitive rates” loses to the page that says “the rate on our 36 month CD is X percent, and here are the three other terms we are also offering.”
Naming the product, citing the number, giving the structured fact. That is the new SEO.
A content engine built this way is not a different content team. It is the same content team with a different brief. The brief is: be the most useful page in your category for a question an AI is likely to be asked. Win that test and your traffic starts to come from AI citations the way it used to come from organic search. This is just as true for a community bank as it is for a credit union.
What I would not do
This part matters. The list of things I would not chase, even with budget burning a hole in the plan.
I would not buy a marketing-specific AI vendor that is essentially a wrapper on a foundation model with a higher price tag. There are dozens in the market right now. Most will be acquired or absorbed in the next twelve months. The work you do inside them does not travel with you.
I would not run a brand refresh. Brand refreshes are the move marketing leaders default to when they want to be seen doing something. Right now, your brand is being tested at the AI surface layer, not the visual surface layer. Spend the brand refresh budget on the AI visibility audit.
I would not staff up. Adding head count in a year when the productivity story is “use AI to do more with the same team” sends the wrong signal to a CFO and dilutes the budget you actually need for the projects above. Hire to replace, not to expand.
The bottom line
If I were back in the seat, I would pick the three projects above, sequence them in that order, and protect the time of two or three senior people to actually run them. Ninety days each. Measurable outputs. Real cultural change.
The marketing leaders who do this work in 2026 are going to look meaningfully different from the ones who do not by the end of 2027. The gap is not closing. It is widening. The budget conversation in twelve months is going to favor the ones who can show what they shipped, not what they planned.
The seat is uncomfortable right now. It also has more leverage than it has had in years. Use it.
Kevin Farley writes about AI visibility, AI readiness, and strategic growth for financial services. Read more on the blog.