Rethinking Personalization in Digital Ads with Maya Hutchinson of BattlegroundAI | Hello Merge Tag Ep. 39

“We do not need to run the same ad to everyone. It’s 2025 — we don’t need to do that.”

Maya Hutchinson is a data-driven marketer with roots in Democratic politics. She started her career on the digital team of President Obama’s re-election campaign and has continued to work at the intersection of tech and politics ever since.

She helped build the European market for NationBuilder, the first community organizing software supporting candidates, parties and campaigns across the EU. She then helped launch the analytics, polling and paid advertising division of DKC, one of the largest independently owned PR and comms agencies in the US.

She’s worked with brand like Uber, Honest Company, Health Ade and OpenTable and political campaigns like Bloomberg for President, the DNCC and Amy McGrath’s Senate race.

She’s the founder and CEO of BattlegroundAI, a platform built to help progressive agencies, non-profits and political campaigns create high-impact, platform-ready digital ads.

She joined us on the pod to talk personalization in digital ads.

Throughout our conversation, we spoke about the role of AI in supporting the people doing the work, why you should be A/B testing everything, how her toolkit works and a whole lot more.

Listen to the full episode right here, or wherever you stream podcasts.

 

Find all episodes at HelloMergeTag.com or wherever you stream podcasts.

 

Links

Website | Email | LinkedIn

 

Big thanks to our sponsors:

Civic Shout. Learn more about the great work they’re doing to help Democratic campaigns and progressive nonprofits build their lists ethically at civicshout.com/partners.

Campaign Deputy. Built by progressives, for progressives, their platform gives campaigns and organizations the tools to achieve their fundraising goals. Learn more at campaigndeputy.com.

 

Episode Transcript

This transcript was automatically generated and has not been copy-edited. While it can serve as a helpful guide to our conversation, it may contain errors or omissions. For the most accurate representation, we recommend listening to (or watching) the full episode.

Josh Klemons: Maya Hutchinson is a data-driven marketer with roots in Democratic politics. She started her career on the digital team of President Obama’s re-election campaign as and has continued to work at the intersection of tech and politics ever since. She helped build the European Market for Nation Builder, the first community organizing software supporting candidates, parties, and campaigns across the EU. She then helped launch the analytics polling and paid advertising division of DKC, one of the largest independentlyowned PR and comms agencies in the US. Uh she worked with brands like Uber, Honest Company, Health Aid, and Open Table and political campaigns like Bloomberg for president, the DNCC and Amy McGrath Senate race uh before going on to found and become the CEO of Battleground AI, a platform built to help progressive agencies, nonprofits, and political campaigns create high impact platform ready digital ads.

Josh Klemons: So Maya, thank you so much for uh coming on the pod today to talk about your work. Uh why don’t you start by just walking us through what y’all do? Like what does Battleground AI do and how does it all work? I saw the site. I I understand the concept, but uh from your perspective, how do you how would you describe it to someone who’s not familiar
maya hutchinson: Yeah. Well, thank you Josh for having me. This is exciting. It’s great to talk to you. And yeah, let’s dive right in. Uh so Battleground AI was it was built uh because we were trying to really solve a problem that I saw over and over again working in this space for over a decade um which was really trying to create smarter digital content um that’s tied to data uh that is fast that is simple for people to use um so that you can move from information to ad ready assets seamlessly and quickly. Um, you know, the biggest part of advertising is the content.

maya hutchinson: Uh, 70% of the performance is dependent on on what you’re showing and telling people. Um, so we spent a lot of time on on targeting and there’s a lot of other elements to advertising. Um, but really we saw the biggest pain point um as content generation. Um, and really making that smart. Um, and that’s what we’re aiming to do. We were especially in a space that is has more regulations um is is harder to advertise. There’s much more restrictions in the political space and that is really um the the industry in the market that we’re we want to help. Um we know that there’s a big need for folks especially um now that there are more platforms all the time, more placements, higher demand and we need to move faster. Uh and we want to help folks do that. Um, and so that’s why we we launched the tool last year and we’ve continued to develop it. Um, starting with with text um, and we just launched our image generation um, capabilities. Uh, we have a lot more planned um, and and so we’re excited uh, to to be doing that as we head into to 2026.

Josh Klemons: So your big focus is that you like a campaign comes in, gives you a message, and you’re you essentially built an AI tool that then takes that message and turns it into dozens if not hundreds of versions of copy that you can then test online. Is that my understanding correctly?
maya hutchinson: Yeah. So, we’re a self-service platform. Um, we’re here to help you, but folks can create an account. Um, you want to define those parameters. So, it’s really based on information. You can start with with nothing, but it’s great if you say you did a poll. Um, and you you know, the the first use case of the the product was actually on a schoolboard race. We had done a poll uh I was managing the the paid advertising and it was like how can we take the information um and research that we have uh and turn that into search copy, Facebook copy um how can we use that in in short scripts? How can we take information and turn it into ads and messages that people um are telling us that they care about those topics um and fine-tune those to those specific audiences?

maya hutchinson: really something that you can do very manually, very slowly. It takes a lot of time. Often times we might have that data and then we we hand off that poll to our creative team and then they’re like, “What am I supposed to do with this?” Like, “Thank you for all of these cross tabs.” Um, and so, uh, the goal, um, is really to take that and then turn that into, um, personalized content and messaging to meet those audiences where they are. Um, so much of the time we’re thinking about ads as um, you know, we’re making them from our perspective, right? And our backgrounds and all of us come in with a lot of different biases. And that’s okay for many different, you know, you’re creating a website maybe or organic social content. But with ads, you really are talking to a very specific set of people and you want to make sure that the information that you have in there is personalized and specific to that group. Um, and so the more and more we can do that, uh, there was some research actually a few years ago around uh, perception of ads and and people over 70% of people are expecting that ad to be very personalized to them.

maya hutchinson: And we can do that much more effectively now. And especially when it comes to political advertising, how can we get these issues to really deeply resonate with our specific audiences?
Josh Klemons: define personalization here because like with a meta ad you can get you can get specific by age, gender, location but like there’s less and less you can do with targeting. So I guess I’m curious two front like are you seeing a that folks are like buying data from L2 target smart somebody like that to be able to get more microtargeted or are you talking about building ads that are just like age and gender based like what do you mean when you say that I
maya hutchinson: Yeah. So, you can start with um just, you know, start with it is a sliding scale, right? So, if you’re a small campaign um or a small organization and you’re launching a campaign, most of the time folks will just, you know, we’re creating one ad. We’re running this to everybody. Um start by Yeah, exactly. breaking that out by um issue specific.

maya hutchinson: So, um maybe you have three or four different issues. um you customize each of those messages to each of your target audiences. And yeah, you can do that based it it varies a little bit based on the platform, right? Um because you always really want to think about how you’re crafting these to every individual platform and also individual audiences. Um and so based on your targeting, but yeah, start with um those demographic information. Um start with interests and behaviors. Uh there’s a lot that we can do in terms of you know we we don’t just need to focus on voter data although it’s very important. There’s a whole host of other information about people um that is important uh that we can think about how do we craft these um to placements of where they are um to time to to insights interests um yeah all the demographic information start to really think about how we can personalize those um and then personalize the messaging um different variations you can make you know hundreds of of copy variations you should have many many more visual variations than we could ever think possible.

maya hutchinson: Um so that then you can really start to see what’s resonating with people. Um one message, one one one asset does not um does not speak to everybody. Uh we we just can’t market like that anymore. Um and so the more and more personalized you can get and the faster you can do that uh the the more strategic advantage your campaign’s going to have.
Josh Klemons: Yeah. So, I’m curious about that. Um, so I’m I love AB testing. Um, I’m a huge huge advocate that if you’re not AB testing, you’re just throwing money away. I mean what are you even doing when it come especially when it comes to like meta ads more than anything but really all the platforms but um hundreds of versions is like I mean for a small down ballot
maya hutchinson: Yeah.
Josh Klemons: campaign like you can’t really a like you literally can’t AB test that many ads at once anymore like Meta actually has caps on it. So like how I guess what I’m asking is like if you pump out 200 versions of the ad how are you seeing down ballot races and smaller organizations like what do they do with all that information?

Josh Klemons: like that feels like overwhelming in a whole new way. So, how how are you recommending folks approach that output? Yes, you went from like an overwhelming poll to now you’ve got like messaging, but that messaging is like a whole new problem to like sift through. So, how do how do you recommend like a small down ballot race handles that?
maya hutchinson: Yeah. So, let’s emphasize there’s a sliding scale.
Josh Klemons: Okay.
maya hutchinson: um for for folks that potentially like lots of lots of local races who who never run an ad at all, right? So, it’s like they never ran ads, we need to start doing that. Um the smallest races need to be able to do that. Um and that could be a couple variations. That doesn’t mean you need hundreds for for just a small race. Um but but one of the large case studies um that we did last cycle um or one of our consultants conducted last cycle uh was looking at about it was about 400 ads across three states um and you know hundred of different candidates um and really being able and this was from an IE um but really being able to increase that personalization both for um the candidate candidates, the location, um the audience that they were talking to.

maya hutchinson: Being able to create many more variations of that ad, um so that they weren’t going to just run. Okay, these are just, you know, Virginia, uh North Carolina, Michigan ads. That’s it. Like they’re just by the state. No, like we can do a lot more now. Um yes, that meant more variations and that meant a little bit. Right now, you know, still there’s like a manual component of launching those. Um, but I think we can we’re we’re going to easily be able to move past the manual element of that, too. Being able to rapidly set up those campaigns, always making sure that humans are are very clearly reviewing and in the loop on that. But you can start to automate a lot more of that flow. Um, AI is a important element in the analysis and creation, but it’s also making development um much faster. It’s making connections and API integrations a little bit more seamless. Um things are being able to talk to each other a lot more um than they were before.

maya hutchinson: And this is a great advantage especially if you’re working in ads um because there is a lot of manual um work in the space and it’s rote right. We’re doing the same thing over and over which is why um this new technology is really uh a perfect application for the sector.
Josh Klemons: And so are you talking about actually building ads through your platform? Because I I thought it was more about content creation, but you’re actually turning around and like placing ads for folks. Like walk us through that.
maya hutchinson: So um that a few integrations are are in the works. Um well I just haven’t really publicly shared but yeah we will have we have an API um just some breaking news breaking yeah first to know um yeah we’re we built an API um so that not only um
Josh Klemons: Is this a breaking? No, I’m just kidding. Okay, we got a scoop here.
maya hutchinson: one folks who are already have ad platforms um can integrate our our technology directly in there so that they can you know you’re already placing an ad so why not be able to um intelligently create some content um uh quickly for that seamlessly in that process.

maya hutchinson: Um, and then on the other hand, excuse me, while we’re generating the plat the content on the platform, we are, you know, we’re focused on on advertising. And so the more integrated our platform can be with where people are placing those ads and being able to really um have a full life cycle of um of connection between the content that you generate uh being able to run that as an ad and then pull that data back in to inform your creative moving forward. like that full life cycle is really what our goal is in in constructing the platform. A very deeply intelligent um system so that you aren’t just guessing. You know, you see a a report, you’re running a bunch of ads, uh you see a report, oh these are really high performing ads. Well, I should make something like similar. I probably should make more um in this vein for this type of folks. like we can also actually start to do that in a smarter way um by integrating that system. Um and so that’s really our our goal um down the line is to have much more integration with those advertising platforms.

maya hutchinson: So, if you’re sending content from our image generation platform to a DSP um or you’re sending um a YouTube ad or search ad or your meta ads um that you can do that from in the platform
Josh Klemons: Interesting. Yeah, I didn’t see that on your site. It sounds like you haven’t announced it yet. So exciting. And um I also didn’t see image generation. I mean, I saw like the textbased um images, but are you doing more than that? Are you starting like So explain what I saw on the site and then what y’all are working on with the image generation because that’s obviously a big difference.
maya hutchinson: Yeah. Yeah. Totally. So, we launched Programmagic, I guess it was a couple weeks ago or almost a month now. Um, and really with with two big goals, which is like we want to be able to um leverage this new technology in a very clear, applicable way for our sector um and make that fast and easy to see that the results, right?

maya hutchinson: this isn’t just like we do not have free form chat on our platform. We are not a chatbased tool. Um I actually I think that ends up being a lot more of a waste of time for people and and so um we are really trying to create a structured process for folks when it comes to messaging when it comes to text to image generation. So yeah, right now if you use programmatic, the outputs are all um instantly sized for uh programmatic and display and social sizes as well are in there. Um so you get seven sized assets instantly. Um and they are uh text and and various um colorbased results. So we really want to make sure that as we develop the product, it’s in line with the evolution of the underlying large language models as well. Um when we first launched images were barely functional. Now they are getting much better and that ability to start to integrate very um you know intentionally uh and consistent outputs um on more and more image generation.

maya hutchinson: So that’s V1 that we just launched. Folks can try that out for free.
Josh Klemons: so that’s what I saw at programmatic is like the one.
maya hutchinson: It’s Yeah.
Josh Klemons: Okay.
maya hutchinson: And then so that’s a V1. Um and then we’ll be continuing to roll out improvements to that. those will be in our sort of the internal tool. Um you’ll continue to be able to um generate and pull in um more visual assets um saving templates um being able to generate much more elevated um asset creation um and also more personalized as that is then integrated into our underlying platform. Um so images uh and then you know as that evolves right the biggest thing everyone came back with immediately was video um right you know this isn’t as easy as as selling a a product right we this takes a lot more to make sure that we can create and help folks in our space get to you know 80% right we can help you get something quicker, faster, more personalized, and then you can take it and and you can continue to modify it.

maya hutchinson: Um, we’ve seen that in our own research, uh, that predominantly the content that performs the best, I don’t think to anyone’s surprise, is when it is the human and the AI combined working together. Um, that is where that sweet spot is. So, our goal is really to help folks get there.
Josh Klemons: And that was a study done by Miles Bugby. Yeah. Um he’s a he’s a friend of the pod and yeah I saw that like so his takeaway was that um uh he said that AI generated ad copy edited by humans perform the best compared to just human generated
maya hutchinson: Yeah, that was a that and that was a 400 ad. Yeah.
Josh Klemons: or AI generated. Was that a fair is that the correct takeaway from his uh study?
maya hutchinson: Yeah. And I think we’ve seen that in a number of of research studies um you know in different sectors too. I think AI can get you more ideas. It can get you more variations, but everything needs a human lens.

maya hutchinson: Um, you that you know the the creativity, the ingenuity of humans is not what AI is supposed to be doing. Like it can it has an endless energy capacity and it can start to generate lots of iterations for you which which help you get you know that that idea. Um, which is why in our platform on the copy side and and certainly um in other elements moving forward like you can make those edits, right? In chat GBT, you’re not going to go in and edit your responses, right? But in our platform, you can um you know, it’s meant to be uh that that connection between the two the two worlds.
Josh Klemons: Okay. And I didn’t see within programmatic like can you can you like upload your brand bible so it’s using your fonts and your hex codes or uh it’s more just standardized
maya hutchinson: Yeah. So, in the public facing like programmatagic. Battleground.ai, you can customize your hex codes. You can upload an asset um or like a logo um as well. So, you can and and adjust your font type.

maya hutchinson: um in the you know when it’s in our platform um you’ll be able to just much more further customize that save assets um and manage that from a a much more personalized um standpoint. But we really wanted to get um the first version out to as many folks as possible. Um start to get feedback exactly like that like we need to save our brand assets, we need to make these edits. Um but that we wanted to get folks to really see um what is possible um that we also don’t need to be afraid of using this new technology especially in the political spa space. We need to be able to craft it to work for us. Um nobody is going to do that except for the people in our space right like nobody else is going to try to make AI work for for political or um you know nonprofit organizations. We have to be the ones that are are pushing that forward and making sure that that it does work, that we can use it effectively um and safely and and have guard rails on it.

Josh Klemons: So like obviously this is data driven first like you talked about like you’re a data driven person. So, are you finding are there themes, are there terms, are there ideas that are resonating that you see working across like everybody’s ads right now like collectively that we could all be learning from or is it really um individualized so much so that there’s like it’s hard to to identify trends based like you have a very interesting perspective into a lot of ad programs. I run a ton of ad programs and I kind of know what works like within my clients. Um but and I you know I I spend time in the ads manager. I see what’s working for other folks but like you have a very different angle. you’re involved. You know, you’ve built a tool that’s helping folks actually develop things. So, I’m curious if you’ve got like any interesting insights worth uh worth that we should all be learning from.
maya hutchinson: I think my biggest one is that people should be running more ads in they should be consistently running them um even if they’re at lower spend levels.

Josh Klemons: Okay.
maya hutchinson: Like I think the biggest thing um especially in the political space is like people need to see you all the time like seven times at least a week um I think is the most recent uh research but like you know of course that you know depending on the source but you need to kind of be always on um which means you you constantly need to have iterations and and fresh versions of content. Um, and so I think the folks who are uh doing that in a really thoughtful way, like they’re, hey, we don’t need to spend a ton of money all the time, but we need to have something always on for folks and we need to keep it very relevant. Um, I think where I’ve started to see a lot of opportunity, um, is is this content that’s hyper relevant. Um, you know, I think especially in our space, we do a lot of like monitoring of the news, right? We’re we’re always like paying attention to or getting, you know, uh, reports every day of like here are the new headlines, but like that that’s great in an email format, but like can we keep our content as relevant to the people we’re talking to as possible?

maya hutchinson: Um, I think when we can do when folks are doing that, um, they’re not just letting something run and run and run. um they’re thinking and especially in this this climate like something is happening all the time and so it doesn’t mean it everything but like think about all those points that are coming out um that you can make fresh and relevant content around um and we can do that a lot quicker and a lot more intelligently now um with AI that you’re really um not even just it’s not even just personalization it’s like um relevant like constant relevancy to your audience. Um I think that’s really where I see opportunity and I see um folks starting to think in that way.
Josh Klemons: Uh I recently gave a session at Netroots. I I did one at Netroots, but I also did a virtual session before Netroots about um how we’re think about we’re thinking about meta ads wrong.
maya hutchinson: Amazing.
Josh Klemons: And it was essentially like me taking a blog post I wrote several years ago and developing it into a session um three rules for downbell candidates specifically around meta ads.

Josh Klemons: And my three rules are start early, test a ton of content, and don’t make yourself the hero of the story, make your audience the hero of the story. And I saw your blog post, which is how like I I reached out based on your blog post, three critical ways to improve digital ad performance. You had different take like different rules, but like similar sort of concept and some overlap. Uh yours were frequency, personalization, and keeping ads fresh, which again, not exactly the same as my three, but like definitely overlap a lot. So, you just sort of walked us through it, but like for me, a lot of it is about like I want my ads my ads quote unquote like because it’s hard to even call them that the way I’m running them. Like I’m doing one giant AB test for months pro like yes there might be fundraising. Yes, there might be lead acquisition but for me I just want to raise the hell out of somebody’s name ID and I have won numerous elections doing that. like, you know, like I’ve seen massive success doing this.

Josh Klemons: Just making sure that by the time that person knocks on their door or shows up to that like event, they’re already like know you well. Um, and a lot of that is like very personalized in the sense that it’s the campaign responding to the real world and then adding that in on the back end to like creating an AB tested um campaign budget optimized account. You’re saying that they should like just be like putting tens or dozens or hundreds of things in at once uh and seeing what happens. Like walk me through what you mean when you say um walk us through your basic premise, frequency, personalization, and keeping ads fresh. Like uh you’ve sort of already walked it through it, but give us the give us the high level again.
maya hutchinson: Yeah, maybe we should write a joint blog post because I feel like it’s actually four or five elements because I think yeah, I would also say like early um which should probably also be the first one there to
Josh Klemons: Yeah, I’d be interested in that. Yeah.
maya hutchinson: your point like start as early as humanly possible with your
Josh Klemons: Do folks push back on that? Because I get like funny looks when I tell folks like somebody brings me on to do digital for them. I’m like great, let’s get your ads manager up and running. And they’re like now? I’m like yeah, doesn’t matter when. Like I don’t care if your election’s in 3 weeks or a year. Like you should be running meta ads at a low level because it’s the cheapest and most effective way to raise your name ID in your district and to stay relevant to your audience. And I am like I am amazed how often I have to like push that because yeah, so few people like sort of they that’s not how people think about ads in general in our party. Republicans are much better at it. Democrats like to wait until the end and put all of their money into, you know, persuasion ads in the final three weeks.
maya hutchinson: Josh, you literally like I should just have that clip and and play it over and over for people.

Josh Klemons: It It’ll be live on YouTube.
maya hutchinson: Um uh yeah, I’ll just play it over and over to myself to to users.
Josh Klemons: You can use it as you see fit.
maya hutchinson: Yeah, I mean I I I’ll dive into to the other points too, but I really can’t emphasize that enough. It’s not and and it it’s the same with a lot of other sectors too because having worked across many different industries like folks it’s it’s not even yes because we don’t have a hard deadline when you’re selling products but like same thing people are like I want to run these ads so someone buys this thing and I want to run this ad so someone votes for me and you’re like okay but that has to start like a year plus you know no one’s going to buy your product no one’s going to vote for you if they don’t know about you they haven’t heard your name before. People forget so fast. Content moves so quickly. You not only need to be early, but I think to the point of frequency, um, you need to be appearing for people over and over and over.

maya hutchinson: Like one time is never going to be enough. And these your audiences for most of these races, like it is not a huge amount of people, right? If we’re talking a local race, even a congressional race, like it is not a massive amount of people you talk to. You talk to them over and over and over again and you need to meet them on platforms where they are. You need to be on meta. Like I know a lot of people are like the ad manager is hard. It’s annoying. The verification it’s like we will there’s a blog post on our site Josh will help you like Miles will help you like we can help you do that.
Josh Klemons: True. Exactly.
maya hutchinson: Um you need to be there and I think that to that point is like yeah same goes for like different podcasts or news outlets like you need to be and show up where people are. You need to be on YouTube. and those are lowlevel um we we worked with a a small state ledge race last cycle and uh they didn’t have a lot of budget um but we were able to get them up on on meta on YouTube

Josh Klemons: Sure.
maya hutchinson: on search um I know searches falling off in places of chat GBT but like you still need to be there um voters are going to Google who you are like show up be in search it’s so cheap anyone can do it. You don’t have to have raise any funds to to run some search ads. Um so so that’s what I mean I think with yeah frequency and and showing up consistently over and over multiple platforms as many as you think you know realistically can manage far out. Um, but those three are and then yeah, be on news sites, be on local display is is a a very costefficient platform. Like show up on those local sites, show up on news that people are reading on mobile. Um, you just you’ve got to be there. It’s the same with, you know, trying to get candidates to be on Instagram and Tik Tok. Like the ads need to be on the places. Obviously, I wish we could run Tic Tac Tik Tok ads, but you know, uh yeah, we could talk about the restrictions on our time.

Josh Klemons: Maybe when MAGA owns it, they’ll let us do, you know, political ads.
maya hutchinson: Um yeah, yeah, yeah, yeah.
Josh Klemons: Just like Twitter, you know, the worst person in the world bought it, but now we get to run political ads. I don’t, but we can if we want to. It’s no longer valuable, but whatever.
maya hutchinson: I I have I have started getting some Dems running some running some some Twitter X ads. Uh I I that’s the thing. It’s like you don’t don’t handic or like handcuff yourself before um before you you even like get out of the gate. Don’t write off different platforms um and be like, “Oh, I’m not going to do that.” Like, we still need to to reach people. Um we still need to talk to people um even if you don’t like the platform yourself. Uh and and so I think that um yeah um personalization I think to your point earlier like it doesn’t have to be crazy overwhelming. You don’t have to run 400 ads. Like Miles is an incredible um uh digital advertising strategist.

maya hutchinson: Like not every campaign is going to be like that. Um you can start with with some personalization just just small personalization. um you know in a way that’s like hey I want to talk to to probably um this group um you know older folks more specifically about about Medicare um and my plan for Medicare like think think just small like what’s a way that like I can make these messages more personalized for different audiences um and it could be you know one or two groups and and start with just like some um some low-level personalizations like it doesn’t have to be crazy overwhelming Um I think if you have a bigger um you know budget and you and you have you know a lot uh more ambitious goals too we can you can scale um you can start to really like expand that personalization level um expand it across platforms and and across audience segments but it doesn’t have to start like that. Um but I think any level of personalization like we do not need to run the same ad to everyone.

maya hutchinson: this is it is 2025. Like we don’t need to do that. Um we we don’t need to make the same type of ad. Like I think that’s also on the personalization front like we do not need to make one big campaign ad and have it always look exactly the same and have it be the thing that everyone’s like, “Oh, I already know this is a political ad.” Like from the way it opens. It’s like no, your your content should match the platform. It should match the way people are consuming content on that platform. Like that’s how you should think about it. Um, if I’m seeing this this ad on Instagram, it should look like something I would see on Instagram. Um, I think we need to to get into get out of the like we made one ad for TV and now we just need to mash it up into a bunch of different places. like we got that also that needs to stop on the personalization front. So personalization is a big bucket I think.
maya hutchinson: Um
Josh Klemons: Yeah, I always say like if you have a TV ad, great. Let’s put on social, but that is not what I want for social. What I want is something that looked like it was made by somebody’s friend because that’s what people will actually stop and watch on social. And yes, the viral moments, like those viral videos, like of course, thank goodness for like big launch viral moments, but like that doesn’t sustain a campaign for a year. Like now you got to pull out your phone and just walk and talk or you know sit in your desk chair and talk you know and like you know like introduce yourself to people in a way that looks
maya hutchinson: Yeah.
Josh Klemons: like it’s from their friends. I do a lot of Tik Tok work and the more produced my videos are the less they perform. It’s so frustrating and such a gift because I can make a Tik Tok video every day without having to be like you know it takes me five minutes to make a Tik Tok video that like does great and then I like will spend hours like working on one and it like bombs.

Josh Klemons: Like it bombs. It’s like hilarious to me. Every time I spend time on something, it bombs.
maya hutchinson: Nice.
Josh Klemons: Like Tik Tok hates it. Instagram reals hates it. Like people just want things that feel like they’re friends.
maya hutchinson: Yeah, they want things that feel like other content that they’re consuming. And so, you know, if you’re a consistent watcher of YouTube, like their ads look a certain way, they’re those moments where they don’t have ads and it’s just, you know, blue sky and an ocean or whatever. um you know there’s there’s ways that those platforms are are unique in how they display and deliver ads to people and and what their content looks like. So make it fit. I think personalization yeah comes down to message it comes down to a placement. Um be thoughtful when you’re thinking about that even with small budgets like yeah don’t overlook that that this isn’t you know this doesn’t have to just be like yeah a one one sizefits-all. Um and then in terms of um just like content updating um I mean I know you know ad platforms will say this because in people are like well it’s in their own interest to like you know make sure that content new content does better right than the older content.

maya hutchinson: And the first thing, you know, if you work with someone um at Meta or at at Google, they’re like, “Well, you need to have like, you know, be updating every two weeks.” And and and you’re like, “Well, do I really need to do that?” And I would say like probably even more frequently. Um and don’t be afraid to like turn off things that don’t work and and and and to the point earlier like relevancy. Like when something is immediately relevant, run the ad for like three or four days. like if it’s just a a moment in time and you’re like I need to capitalize on this um just run it for for a little bit um and then you know go back to your your previous always on campaign but but continually like review the content um and and have sort of a couple core metrics um you know whether you’re running you we’re talking about name ID like you’re running just an awareness campaign and you’re optimizing thing pick a solid metric um for that campaign and be constantly reviewing them, constantly checking that content.

maya hutchinson: I mean, I don’t have to tell you. It’s like you’re in that ad manager like the the person managing these ads is in there like one, two, three times a day probably checking on that content or if they have a nice like dashboard to report on them. Um but you’re in there, you’re seeing what’s working. like don’t be afraid to um to update to to to turn things off to like keep that really fresh and relevant. Um I think it it is a lot of work and I AI has not you know solved all of the all of the hard parts about this. Um that really takes human like review too very very deeply. like that’s that’s important knowledge and and when you’ve worked in ads you know for a long time you you start to really understand how to to do that in a way. Um but yeah I think that that is something that um even at the smallest level uh you can start to do pretty effectively.
Josh Klemons: So, I have never run an ad on meta that was AI generated.

Josh Klemons: So, I’ve never clicked that little button that says this is an AI generated ad. Does that have an effect on performance in your experience or it’s like just an awareness and it it doesn’t help or hurt?
maya hutchinson: So, we haven’t run a ton of research with images yet, but um work with the AAPC on our AI committee and and obviously this is a big topic um uh for everybody in the space, which is like what are the impacts of an AI disclaimer? um for the most part um in the research that has been conducted to date. Um it it it doesn’t really do people don’t really pay attention to it. Um but there is obviously a fear that it will degrade performance or and somehow negatively impact that. My sort of general thesis is that pretty soon everything that you create, I mean, anything you’re developing in Figma or Canva or like there’s AI built into now almost every platform you’re using and it’s not something that you’re like turning on and off like that. That’s the underlying core creation.

maya hutchinson: Um, and so, you know, I think more and more that that doesn’t really impact how people view and and comprehend the ad, but I do think that the quality and the, you know, the output of that AI, um, that’s where it will impact it. And and I think you know for for video uh straight out of the gate AI generated video I just do not think is in a place that people for the most part people can tell like it is it is still in in a emerging um place. that doesn’t mean that we shouldn’t um continue to work with it and and see how it can be beneficial to us and and where and how we apply it. Um but it’s not there yet. And I think we can’t out of the gate just you know without any review or edits from from folks like that that quality is not good enough. um and and people will see that and and you you know um and so I I don’t think there’s a disclaimer on anything.
Josh Klemons: Mhm.

maya hutchinson: Um is and and it’s also just like I don’t even you know politics has been dealing with disclaimers to whatever you know if those help or or hurt for a long time and I think we’re going to start to have more research on that. I know the AAPC is thinking a lot about that of like creating a thoughtful research on performance and and what what impact that has and if there’s even additional disclaimers, you know, in some certain states. Um, but I think candidly that like it’s not really ever going to be a disclaimer that gets anyone to like, you know, not do, you know, not have deep fakes or do sort of like malicious things with the tools either. Disclaimers are, you know, I think they’ll they’ll probably not be relevant in in the very near term. What’s important is that you’re using the tool in a very intentional and thoughtful way and like you’re able to produce things that are accurate, are informative, are high quality, are um personalized. Uh and that’s what’s going to be important.
Josh Klemons: Um, on that note, like so Meta has their own AI tools and they push them harder and harder all the time and I can’t say no thank you enough because a lot of times they’re like missing the mark. Um although I have heard from folks like they’re testing more and more and seeing interesting results with them. So it’s like something on my radar. Um have you done tests like your AI tools against theirs or is that something that’s like in your works? Like I’d be curious to know if like Meta has their own tools built right in like where do where does that fit into sort of like your thinking on AI and ads?
maya hutchinson: Yeah. Uh I have not tested it against um any of the platforms and Google has the same thing too.
Josh Klemons: Sure.
maya hutchinson: Um I think where uh you know those are platform specific. So, um I think there’s there’s kind of three things. Um but like the first is that um while helpful to some degree um with copy variations, um Facebook doesn’t have all of your context for the data um on your audiences um your polling, your research, your talking points.
maya hutchinson: Like it has metadata. It has it has your meta data um uh which is Yeah.
Josh Klemons: Right. I got you. Yeah. Your data from a meta. Yeah.
maya hutchinson: which is which is one view, right? Um but like yeah, wouldn’t have your customer data or any of that. Um and and so it’s one view. And so yeah, I think an interesting test would definitely be to to compare them. But um our platform is is all, you know, encompasses all of those digital platforms um and your data and context. So it’s the idea is that that’s getting smarter over time. Um, and that’s applied to, you know, your entire campaign, not just one platform. Um, and so while I think, you know, we should be testing and and leveraging tools in every space to get better performance, um, additionally, you know, essentially the the core of Battleground is that, you know, all of your ads live in there. So they’re not just going to be sitting in your meta ad library or my case they’re in a thousand spreadsheets and and lots of Google drives like that content lives in battleground so that um you can start to learn and if you do start to you know you need another you have another account or another client and you just start a whole other account.
maya hutchinson: Hey what what was working? Hey, what what can I see um that I could take and and apply to to future um campaigns or how do I you know, how do I know that this is like continuing to build and grow over time? Um that’s really the powerful part of AI is like yeah, of course, some accounts you’ll have that same ad manager forever, but most of the time you don’t. Um and you want to be able to to start to get smarter. um not only with performance but like AI in general like it starts to get smarter as it knows what you’re working on what you’re creating how things are performing like that’s the goal of that sort of centralized database of ads um regardless of platform and so same with with YouTube like one off platforms are great um but the biggest thing that I and many we didn’t have is is a place where all of those live. Um, we we have to have a place that that that lives and breathes um specifically about ads.
Josh Klemons: Um, yeah.

Josh Klemons: No, super interesting. Uh, I’m curious if you have thoughts other than Battleground AI, other AI tools in the industry. Is there anyone else that you’re excited about the work they’re doing? I assume you’re keeping your eye on the AI tools out there. There’s more and more of them all the time. So, yeah. if there are any that uh you’d recommend folks check out.
maya hutchinson: Yeah. Um I mean industrywise or like I I look across Yeah.
Josh Klemons: Your choice.
maya hutchinson: I mean I look across so many different industries and I think I would encourage folks to do that too which is like look to a lot of different sectors to see what tools they’re using and what is really like moving things forward. Um whether that’s like experimenting with a tool like Lovable or Replet, like you know, you want to see how you could like fast code something. Um start to like move across different spaces.
Josh Klemons: Mhm.
maya hutchinson: Um there’s uh you know, a host of different um tools that are that are doing that.

maya hutchinson: Um uh in in our space um there’s definitely an interesting tool um uh that I’ve talked a lot with um they’re called Trollwall AI. They monitor um comments and um conversation online um for for political campaigns for nonprofits um for for any organization. Um, so they’ll, you know, are constantly monitoring your social accounts and and really, um, intelligently, um, reviewing that. Um, I think, you know, I’m always like I’m the first to always test and experiment with new tools. So, I’m like literally constantly on the lookout. Um and and I do think also um uh in terms of like a model AI models um you know change agent and the the team there who are building um sort of a a progressive AI model like that’s just a really powerful way to innovate in the space and to take you know a tool that is you know so global and and exactly to what I was talking about before like if we don’t take you know the tool and apply it to our use cases is we’re either just like letting other people have control or you know just sitting on the sidelines and being like I’m going to wait I’m just going to wait until AI is like you know it’s so good like it’s you know it’s as good as people and you’re like that is not at all what you want to do and and AI is not going to be people like

maya hutchinson: it is not people it is it is a technology that we’re building and you want to be involved with it um as much as possible in every facet. That does not mean that like a lot of things I you know especially I do are not AI but um but thinking about where it can be helpful and especially if you’re a small team like we’re a very small team um you know we have to scale content production or customer service or you know email triggers um even scheduling like how can you automate a lot of that those manual tasks and maybe that won’t be all of them but it’ll be one or two um like to free yourself up and and to accomplish a lot more um in less time. I think where we can do that.
Josh Klemons: Yeah, sure. How can folks keep in touch moving forward? The website, I assume, but like which like send folks to Yeah.
maya hutchinson: Oh, you can email me. Uh Maya, you can text me my phone number.
Josh Klemons: What’s your email?

Josh Klemons: What’s your website? What’s your uh social channels?
maya hutchinson: Um um I think my my phone number is in my email signature so you can uh uh yeah you can email me at battlegroundai.com may battlegroundi.com uh you can send me know on LinkedIn or any other platform.
Josh Klemons: You’re welcome to. I wouldn’t recommend it. Okay, right.
maya hutchinson: I’m I’m usually I try to be pretty fast in responding. Um uh yeah I I love feedback. um folks want to try the tool and and give thoughts um we are here to help and and really try to be um you know more than just a platform like we we all have worked in this space um so we want to you know support people who are are are working to innovate um especially in in advertising you don’t need anything yeah it’s a any anyone can yeah anyone can generate generate content.
Josh Klemons: and I’ll say with programmatic you don’t even need a login much less a credit card to get started. Right. Right. You just go on there and play around with the tools.

maya hutchinson: Um, you can generate up to five sets of of ads a day. Um, and and that is completely available um for folks and uh in our internal tool. We’ll have a lot more updates to program magic, but we really want folks to um experiment and see uh you know that will continue to evolve and the outputs will continue to improve just like with all of this technology um faster than than you think. Uh, and so yeah, I I encourage folks to experiment and try it and and write us and tell us what you think and what you need. Um, because I think that’s the most fun part about um, well, from our perspective and from building a tool is is really to solve um, and help people. Um, but from their perspective, I hope as we we think about the plans for next cycle, like how do we do things better? How can we help um you know scale programs or or help folks think through ways um to apply this um new technology?
Josh Klemons: I’ll have links to the site and your socials and all that fun stuff in the show notes.

Josh Klemons: Thank you so much for coming on and talking. Super super interesting. So, write the blog post.
maya hutchinson: Thanks so much Josh.