Gen AI, Generative Engine Optimization and Personalized Content at Scale

Content marketing through the use of blogs, videos, social media, and email newsletters can feel a little dated. Not so fast: Content marketing is still a viable strategy and is crucial for building brand awareness, generating leads and driving sales. Small businesses need to be aware of the trends that are impacting the creation of an effective content campaign. Panelists will discuss how Gen AI can be a co-creator of content, the move away from SEO to Generative Engine Optimization, and how customer data platforms can create personalized content at scale. 

Transcription:
Transcripts are generated using a combination of speech recognition software and human transcribers, and may contain errors. Please check the corresponding audio for the authoritative record.

Stu Richards (00:08):
Welcome back from lunch. Thank you for joining us. This panel is on a topic near and dear to ours: content, and specifically using AI to optimize content at scale. To kick things off, I want to ask each of our panelists to introduce themselves and their organizations.

Ben Udell (00:31):
All right. Well, I'll kick it off. My name is Ben Udell. I work for Marquee. We do data automation with the core into email, direct mail, marketing strategy, website, and content. In my day job, I get to work with a lot of banks and credit unions on this topic. In addition to that, I do a lot of the GenAI speaking for the industry. I teach at ABA Stonier and GSB Wisconsin. I do a lot of events like this, and I write for The Financial Brand on applicable GenAI in banks and credit unions.

Abbas Merchant (01:02):
Great. Abbas Merchant, I'm with Pravas Consulting. I have over 30 years of experience in marketing across multiple industries. Most recently, the last 20 years were in financial services at US Bank, M&T, and as the CMO at Regions. For the last two years, I've been working with credit unions and community banks to bring the same sophisticated, data-driven strategies to them that were previously only accessible to larger institutions.

Mike Lee (01:34):
All right. My name is Mike Lee. I'm CEO and co-founder of Fundica. Fundica is a funding platform that helps small businesses identify, organize, and apply for relevant government funding. We cover over 10,000 programs, whether federal, state, or municipal, and we license the solution as a white-label solution to financial institutions. This really helps financial institutions acquire deposits, customers, and more firmographics; it really is a marketing tool.

Stu Richards (02:06):
That's great. Thank you. I'm Stu Richards, CEO of Braden. As I mentioned earlier, we do B2B research. Among other things, we survey small business owners on the effect of marketing content on their likelihood to do business with different organizations. What we find is that it depends on the quality of the content. Marketing content generally makes small business owners more likely to do business with you in the first place, more likely to stay a customer, and more likely to purchase additional products and services. We see lots of benefits to content marketing. We conduct research and turn that data into white papers and webcasts, or support our clients' PR, social, or sales campaigns.

(03:02):
Abbas, I wanted to get your take on what kind of benefits you're seeing organizations get from the execution of content to engage small business owners.

Abbas Merchant (03:12):
Yes, Stu. I think content marketing is shifting from just being a marketing function to being a core part of the business growth engine. For community banks and credit unions, it is one of the lowest-cost approaches to scale trust and underscore their core value proposition of expertise and being community-centric. It is critical, but I would say it's underutilized. Unfortunately, in some financial institutions, I find they are thinking of content merely as a marketing function that is not closely integrated into the sales process. The leaders in this space actually integrate content into the sales process. Let me give you a quick example: think of a community bank that turns its expertise in treasury management and SBA into a working capital playbook with relevant content and use cases.

(04:29):
It combines expertise with a cash optimization tool and relevant benchmarks. Creating that ecosystem where content becomes the connecting factor is critical. Using content in this way, we've seen a two-to-three-times increase in engagement versus the traditional approach of email blasting one-off thought leadership pieces.

Ben Udell (04:57):
Can I jump in on that? I get to work with community financials, and one of the numbers I think is fascinating is that 80% of banks and credit unions claim to be applying AI—which is basically the most fake BS number you've ever heard in this industry. I know that because when I speak to banks and credit unions, 50% of them have an AI usage policy, but only 5% have actually provided training. We hear all of these moonshots, but the funny thing is: spend the money on content because it's cheap, easy, and has a quick ROI. People are getting so wrapped around the axle on these big data projects that they're forgetting about this low-hanging fruit.

Abbas Merchant (05:41):
Actually, I'll build upon that and say that with the technology and tools available today, you can build highly relevant and impactful content at scale using generative AI with a human in the loop.

(05:57):
We'll talk about it.

Stu Richards (05:58):
Mike, I wanted to get your take on how you see fintechs, banks, and credit unions using content effectively and what kind of business outcomes they're getting.

Mike Lee (06:07):
It's very important that they establish some level of trust. As our last speaker said, banks and fintechs have a strong level of trust to look at. After that, content needs to be very personalized—very individualized to the group—and that can be done now with the tools available. Finally, it needs to be actionable and helpful. You need to be helping them with what they need help with at that time.

Stu Richards (06:36):
Great. Ben, question for you. What would you say separates a good content marketing program from a great content marketing program?

Ben Udell (06:48):
I think about that as content, channel, and cadence. Poor performing ones don't do any of that, the okay ones do bits and pieces, but the really strong ones figure out a strategy where they can do all three well. A good example is in the onboarding space. If you don't lock down a new relationship in that first 90 to 180 days, you're going to struggle with future sales because 10 other FIs are pounding them for business. The lower performing ones might worry about emailing too often while competitors fill the inbox. Or they don't have the time to put together direct mail, email, financial literacy, and fraud support.

(07:47):
The power with GenAI is that I can take an onboarding campaign and, instead of it taking someone a week, I can get a first draft of everything in three hours. The banks and credit unions that are struggling aren't using it that way. They don't understand that they need to work with it like an intern to knock out first drafts. They're too freaked out about compliance and hallucinations. As Abbas said, use the human in the loop; that's going to protect you from the evils in the news.

Stu Richards (08:23):
Absolutely. Mike, same question for you. What are you seeing between the good and the great in engaging small business?

Mike Lee (08:33):
The really good ones don't look like they're trying to sell. They're not "marketing"; they're actually trying to establish a relationship by providing information, solutions, or tools the group is looking for. It's a genuine, personalized relationship. They are really connecting and turning it into a conversation as opposed to a blast of text. It's creative and not easy to do, but if it's well done, it's extremely valuable.

Stu Richards (09:12):
Okay. Love a show of hands on who is using content marketing. By content, we typically mean advisory content—not product information—but emails, articles, or case studies to educate small business owners. Okay, that's great; that's most of you. Tactical question: our research shows email newsletters work well at the awareness stage. At the research stage, they like research reports, but they also like video on your site or YouTube. Ben and Mike, what formats are you seeing work well, and what other tactics are effective?

Mike Lee (10:20):
It really depends on what kind of information you're trying to relay because you have to make sure it's consumable to the user at the right time. Certain information goes well through interactive short video, but for other information, video isn't the right way. It depends on the info, who you're speaking to, and when you're speaking to them.

Ben Udell (10:47):
That's a great answer. Here's something I would encourage everybody to do: go into ChatGPT, Gemini, or Copilot and act like a client. Say, "I need to open a business account in Madison, Wisconsin. Who should I open it with?" There's a good chance it won't be any of you. Because of how it pulls in content, it might recommend tiny credit unions that nobody's ever heard of. We're still figuring out the best way to use GEO (Generative Engine Optimization), but a lot of that is driven by content. They want good quality content that can be read and understood to feed their models.

(11:44):
To stand out, you have to put a lot of good quality content out there. People aren't just searching for "a small business account"; they are searching for specific scenarios, like "I'm a consultant that needs an account because I run a DBA, but I have a full-time job." My bank really struggled with answering that for me.

(12:14):
We are at an inflection point. The institutions putting out significant, high-quality content—even if it's just public on your site—will stand out in search. One last piece: if you haven't played around with ChatGPT's new engine, SearchGPT or the Atlas model.

Abbas Merchant (12:47):
Atlas is coming.

Ben Udell (12:49):
That's going to be a new search tool to find financial services, and they are going to use the ChatGPT version, not Google, which throws everybody for a loop.

Stu Richards (12:59):
You raise a good point about optimizing for different personas. How do financial institutions do a good job of optimizing content by industry or stage of growth? Abbas, do you want to take that?

Abbas Merchant (13:32):
Absolutely. Segmentation has to go well beyond industry and size. We identify the role of the audience—are you talking to the founder or the CFO? Different content resonates with them. We also look at the life stage: are they a startup, thriving, or in transition? You can do that level of segmentation using third-party data. The next level of depth comes from understanding the mindset and moments, which are critical. We use three data streams for this. The first is behavioral: what are they searching for, and what content are they engaging with? That tells us their intent.

(14:40):
The second part is market intelligence. We've created a database of thousands of financial institutions' content, and we can benchmark your content against that to identify white space and gaps. The third piece is sentiment analysis. We look at social media engagement and review sites. When you stitch together behavioral attributes, market intelligence, and sentiment analysis, you get a holistic picture. You stop guessing what topic to publish next and start answering the questions customers are already asking.

Stu Richards (16:05):
Thank you. Mike, any thoughts?

Mike Lee (16:07):
The challenge, especially for smaller institutions, is doing it efficiently. You want the right solution for everyone, but you have to be efficient. You can use tools where content will actually adapt to who's coming in. Or you can build a lot of different content for many different segments so it's ready to be found. It still comes back to being relevant and hyper-personalized.

Ben Udell (16:50):
Institutionalizing this is hard, but you see leaders getting smart about it. For example, younger business bankers who can't understand every industry are asking generative AI, "I'm working with this construction company. These are the financials. What ratios look funny, and what questions should I be asking?" Those are transferable questions. They use AI to accelerate productivity in learning and communication. When a business owner talks to four different banks and one banker is asking great questions and is on top of it, that's a huge competitive point.

Stu Richards (18:27):
What kind of mistakes do you see organizations making in their content marketing strategy?

Ben Udell (18:40):
A couple of things. First, they just don't have the time to do it, so the content isn't great. It's not a value-add; it's a regurgitation. Is it better to have something weak out there or nothing at all? The other side is not putting content out there that resonates with a specific niche. Especially in the credit union space, the attitude of "we serve everybody" makes it hard to take care of everyone's specific needs. You have to put a stake in the ground in a certain area.

Stu Richards (19:45):
Mike, how are you using data to optimize content strategy?

Mike Lee (19:56):
When using Fundica within the bank's platform, we allow businesses to identify where they're at, their revenue, number of employees, and zip code. We figure out their profile and their intent—whether they want to export, hire, or buy equipment. That's a win for the financial institution because they see what the user is looking for. We provide a comprehensive list of government funding like grants and tax credits. The bank captures that data and can then decide how to reach out, perhaps routing a lead to an advisor.

Abbas Merchant (21:14):
That is a great example of capturing data about what they're interested in and turning that into leads.

Mike Lee (21:26):
Exactly. For Fundica, we're really trying to understand our different types of users and build an engaging, useful story for each one.

Stu Richards (21:51):
Ben, how are you seeing AI tools used effectively for content market strategy and creation?

Ben Udell (22:15):
We are still very much in the second inning of AI advancement in banking. Out of 8,500 banks and credit unions, most have only maybe turned on Copilot, but they haven't turned it on for everyone. Other industries are significantly improving productivity and content quality. I see people starting to get smart about providing access. CRM projects are usually top-down from the CEO. With generative AI, the CEO has to enable access, but they might not understand how someone in deposit ops will save four hours this week. Leadership needs to reward and recognize great use cases.

(24:07):
If one person saves four hours and 10 others do too, you've saved an FTE in productivity. Leadership needs to encourage safe, responsible use cases. People are getting wound up about PII (Personally Identifiable Information). If you don't want to use private data, then don't use private data. They don't trust their people to use the tools responsibly, even though they've trusted them with PII for their entire careers. Give them the tools to get rolling with it.

Mike Lee (25:06):
As Ben says, there's a love-hate relationship. Risk groups worry about trouble while others say it's the future. In time, the two will marry better.

Ben Udell (25:22):
Alcohol is the cause of and solution to all of life's problems—that is generative AI. It will cause problems if you're writing slop, but it will solve problems if you ask it to think critically or provide grammatical support.

Mike Lee (25:50):
Software ate the world, and now AI is eating software. In five or ten years, asking "Do you use AI?" will be like asking "Do you use Excel?"

Ben Udell (26:03):
Imagine telling a business banker they can't use Excel anymore. That's what's happening with AI. If you could do something in one hour instead of eight, and you're not allowed to, you're going to be ticked off.

Stu Richards (26:19):
Abbas, what are the best use cases for AI in content execution?

Abbas Merchant (26:27):
The use cases are many. When I talk to CMOs, they focus on efficiency, which is good, but they miss the capability to deliver high-impact content. The real value is in personalization and research analytics. We use secure LLMs trained with brand and compliance guidelines to generate content optimized for SEO and GEO. That's how you get from data to a highly relevant piece of content.

Stu Richards (27:53):
Ben, how is the cautious adoption of AI impeding content marketing?

Ben Udell (28:09):
In most organizations, you have high performers and you have lazy people. Both are figuring out how to use AI. High performers want to work four hours instead of eight, and lazy people want to do the same. Both have powerful real-world applications. 95% of community financials don't have a "Head of AI" or any ownership of the topic. It's a fractured environment. Some people will do awesome things, and then the bank across the street will hire them away, give them the AI title, and pay them 30% more. My previous bank lost me because they wouldn't give me the title, so I left for Marquee and it's been awesome.

Stu Richards (29:54):
Mike, how can organizations implement guardrails to ensure they aren't cranking out "slop" but high-quality copy?

Mike Lee (30:21):
A lot of that comes down to doing your homework. You should run an LLM in your own environment. Prompt engineering is where it's at—organizing, using, and testing properly. You put guardrails in to make sure certain things never happen. Software development is getting easier, but prompt engineering has gotten harder.

Stu Richards (31:03):
Ben mentioned GEO. Abbas, how do you see GEO factoring into content marketing?

Abbas Merchant (31:14):
Imagine a time when your website becomes a content warehouse with very little direct traffic because AI search platforms become the front end. To be visible in generative engines, you need three things. First, findability: many banks are blocking AI traffic. You won't get visibility if you block them. Second, authority and expertise: every piece of content should have a bio, a citable source, and a credible author. Third, credibility: build partnerships with trade associations and chambers of commerce to share your content. AI likes co-citation.

Stu Richards (33:36):
Mike, I'd like your take on that as well.

Mike Lee (33:40):
Content has to be reliable, personable, and actionable. We're never going to know exactly how it works. It's not like the old days of Google ranking tags; now it uses complex, unexplainable algorithms.

Abbas Merchant (34:31):
It is much more complex, but we've put together an eight-point checklist I'd be happy to send to you.

Mike Lee (34:41):
Putting out fluff doesn't work anymore. Get rid of the long, painful fluff. You need stuff that makes sense and connects.

Stu Richards (35:03):
No more fluff. Any questions for our expert panel?

Ben Udell (35:16):
Hot tip for everyone: if you aren't using GenAI for content yet, go into Microsoft Word and hit the dictation button. Talk to it. Create a long dictation of your board report or sales plan. Take that and go into ChatGPT and say, "This is content I'm writing for small business customers in Madison, Wisconsin. Write this for that audience." It will give you a draft. Most of us aren't hired to type; this will do it as fast as you can talk. Edit it like an intern wrote it for you. This will save significant time and energy.

Stu Richards (36:44):
What are your top tools?

Ben Udell (36:49):
I love Otter.ai. I work with Microsoft Copilot and ChatGPT is my BFF.

Abbas Merchant (36:59):
ChatGPT, Copilot, and Perplexity is great for research.

Mike Lee (37:13):
Same for me—the foundation models. In our software stack, we have more specific ones as well.