How to Ensure a Consistent Brand Experience Across Platforms

Jeffro (00:01.036)
Welcome back to Digital Dominance. Today, I’m joined by Hikari Senju, the founder of Omniki. Hikari and his team are reimagining digital advertising by using AI to generate highly personalized data-driven ads at scale. Now, their platform helps marketers and business owners optimize their ad performance with real-time consumer insights, which I’m excited to find out more about today. So in this episode, we’re going to dive into how generative AI is reshaping the future of advertising and how businesses can take advantage of these advances to stay ahead.

So welcome to the show Hikari.

Hikari Senju (00:32.917)
Thanks, Geoffrey, for having me. really excited to be chatting with you.

Jeffro (00:35.627)
Yeah, definitely. Now, there’s clearly a demand for personalized ads. We’ve seen that. But could you give us some insights as to why that is? Do personalized ads really perform that much better?

Hikari Senju (00:48.061)
Yeah, absolutely. mean, if you think about like the progression of advertising.

We used to live in the world of mass media. know, 1950s, everybody was watching the same television show. think 80 % of Americans viewed most of the episodes of I Love Lucy. And so when you’re living in the same media environment, then these mass ads work because everyone has the same cultural context. But as media has increasingly fragmented over time, in the 90s, you had these cable networks that just spoke to distinct interests like Fox News, and then you have CNN, and then you now have social media

that even hyper personalizes the news feed based on an end of one. Everyone’s living in their own media bubble, in their own kind of distinct media diet, and that means that there’s a lot more work for the advertiser to connect with that end consumer in a personalized way so the end consumer of the content feels like they’re being spoken to and not just spoken at.

Jeffro (01:45.687)
Well, but what’s the difference? Because I know we’re starting to see a lot of this and sometimes it’s easy to spot the fake personalized ads where it’s like, hello, friend or hello, Jeff, you know, and it’s like kind of disjointed or like a different font or something. And seems like it wasn’t a natural actual human interaction. Is that an issue?

Hikari Senju (02:08.979)
Yeah, absolutely. mean, I remember actually.

I think I can share this story. One of my friends and advisors, a company called Teespring, they created a custom audience of people just named Jeffrey, for example. And then they ran maybe a couple thousand people, and then they just said, hey, Jeffrey. And then they got banned, or they got dinged by Facebook for this. So yeah, I think when it’s done poorly, then…

or done in a very awkward way, then that doesn’t benefit anyone. It doesn’t benefit the end consumer, or doesn’t benefit the platform, it doesn’t benefit the brand. It just feels creepy. And so really, this is where you use data to really find the fine line between what’s actually driving higher performance in terms of actual sales versus what’s just, you know, sounds like a good idea, but maybe, you know, not actually, if not very effective, especially if it’s not executed properly.

Jeffro (03:09.3)
So how does Omniki use generative AI to kind of create these personal ads that actually resonate with a specific audience as opposed to like what we talked about where it feeling, you forced?

Hikari Senju (03:20.883)
Yeah, well, today it’s really hard for a marketing team to create personalized ads. We still kind live in the world where design teams are creating this content, the marketing teams run it, but…

They don’t have a whole ton of variation in terms of the content that they’re running, even the biggest companies. And yet the marketer is, 55 % of their success of them hitting their goals is based on the creative, the ad that’s running. If you run a good ad, you’re gonna see an increase in sales. If you run a not a good ad, you’re not gonna see an increase in sales. And so it’s one of these really hard problems. And so…

Partly because it’s hard to just get an understanding of what type of ads are going to run for specific audiences, but then it’s also hard to create all that content as well and do it well. And so the way our tool works is it makes that easy. So customers can, they upload their brand information, they integrate their ad networks, ad platforms. get all the real time data, we go to the performance of the various channels.

customers input the target audience that they’re looking to sell to. And then we’ll generate thousands of variations of ads and ideas that customers can then approve and launch through our platform. And so we make personalized ads really easy in an omnichannel way.

Jeffro (04:40.021)
So are you, do you have your own ad network or are you running on top of like Facebook ads and Google ads?

Hikari Senju (04:45.299)
Yeah, so we’re an OmniChannel platform, so we’re integrated with all the various social platforms, but there’s also big streaming platforms as well for television ads. So, you know, think of the Disney Pluses and the Netflixes. And then there’s gaming platforms. There’s certain types, certain kinds of out of home networks. And so really, when it comes to like OmniChannel, like the advertiser today has many options when it comes to buying media and, you know, kind of being in front of the customer.

in front of their customer. And so we make all that available to them.

Jeffro (05:19.761)
But at the same time, aren’t you still limited by the capabilities of these different ad platforms? Because I imagine some may give you more flexibility than others in terms of serving a specific ad to a specific person.

Hikari Senju (05:31.219)
Yeah, so the targeting options are definitely varied across the different platforms and the data you get is also very varied across the different platforms.

But what isn’t different is that you still need to create good content. Every single one of these platforms, you need to upload it creative to run. And in fact, you really shouldn’t be delegating that to the platform because you want to orchestrate this seamless customer experience where a customer is getting a unified brand experience, kind of surround sound. So when they see the ad on, say, Instagram, it’s consistent with the ad they’re seeing on YouTube or Twitch or on

on Pinterest and so that orchestration isn’t going to happen if you just delegate it to like, metigenerate the ads for me or Google generate the ads for me, you’re going to just get different customer experiences. so orchestrating that is really where we come in, terms of running these Omni Channel campaigns.

Jeffro (06:30.347)
How specific can you get? Because obviously if you’re identifying the same person on different platforms, and that’s one thing, and you’re giving them an ad maybe based on what buyer stage you think they’re in, are we close to the point where you could start inserting a name into that for that person, or is it not quite there?

Hikari Senju (06:50.037)
Yeah, so the stage that we’re broadly in is, I would say, more audience targeting. So there is an element where you can’t target an individual on, say,

Jeffro (06:54.688)
Okay.

Hikari Senju (07:05.026)
You can target like an audience, like a lookalike group or an interest group or a demographic. And you can overlap these to create your own custom audience. But you can do that kind of targeting across different platforms. You can target, say, a similar zip code or the same zip code in a similar kind of broad array of demographic information going into target audience across these various platforms. And that’s honestly usually actually sometimes superior in terms of instead of just targeting one person. Because there’s a lot more data when you kind of cluster these, the consumers in

in similar groups, right? So I might not have enough information, Jeff, about you as an individual, but if I cluster you with other people in the same zip code, with a similar education, with similar kinds of work experience and similar kinds of hobbies, then I may be able to get an insight from somebody else and then be able to kind of just assume that maybe those things may apply to you.

That audience-based targeting on omnichannel basis is really where we’re at, I think, today when it comes to state of art. But who knows? Maybe at some point you can target just one person across all different platforms. That’s really up to regulation and things like this.

Jeffro (08:17.162)
Well, I imagine this kind of makes it harder to accurately measure the ROI of certain campaigns and things because, you know, we’ve got traditional split testing A or B, this one does better, done, right? But like you said, if we’re grouping people into these, you know, audiences, you’re kind of assuming that you got that right, that they’re in the right group in the first place. And then maybe you can test something within there, but how do you really know what’s working or not working?

Hikari Senju (08:40.147)
Well, we get the data from the performance of the ad. So for every ad creative, we then see, we’re talking to this audience, how many clicks they were, how many sales it drove, how much revenue it potentially drove, the return on ad spend of that advertising creative. We see that information on a per creative basis. And so we can see if an ad is being effective. So is this ad performing, has a higher return on ad spend or a worse return on ad spend compared to other creative? And then you can optimize that way.

Jeffro (08:58.261)
Okay

Hikari Senju (09:10.023)
running campaign targeting. So an audience are running thousands of variations of creative across all different platforms. We see which channels performing the highest, delivering the highest return on ad spend. We see which channels aren’t delivering the highest return on ad spend. We see which creative is driving the highest return on ad spend. We see which creatives aren’t. And then from there we can optimize against that audience.

Jeffro (09:26.761)
Yeah, and I imagine this is where the deep learning comes in, right? Because with these huge data sets, you need some filter to kind of pull out all this data and make sense of it.

Hikari Senju (09:35.861)
Yeah, so there’s deep learning and gen AI throughout the entire stack. And really, this strategy from day one for us has been kind of automating, you creating this workflow and then automating and kind of bringing each part of the workflow to state of art as the technology is advanced. first, know, pure point, getting the good insights, getting data in from all the different platforms in real time, making that system as robust as possible. So we’re getting the freshest data in a robust way.

extracting insights from each creative. So we initially start with computer vision, but then now using multimodal AI to kind of explain the image in various dimensions and then build the predictive models against those insights, say, okay, so this target audience will more likely resonate with this kind of messaging or this kind of imagery, or we should try these other ideas because it’s been shown to, this predictive model has shown that it might work.

to then using deep learning to then actually generate the content. Fine tuning the model of the customer’s brand, fine tuning the image generation models of the customer’s product photos.

assembling the various assets together in a design and style that works for that platform that we’re the ad on, and it also works for the kind of aesthetic that the customer may like. That’s done through fine-tuning, that’s done through retrieval, augmented generation, to find ads and designs that are similar to ads that performed well in the past, and then launching the content, issuing the content across the various platforms. So every part of that stack is its own distinct

problem with its own kind of AI that’s integrated into it. And then the end vision for us is really trying to automate that entire flow as much as possible.

Jeffro (11:20.387)
So yeah, eventually you’ll be able to click a couple buttons and then sit back and watch everything work magically. That’s the dream, right? So I wanted…

Hikari Senju (11:27.559)
Exactly. Make advertising as easy as clicking approve is what we say.

Jeffro (11:32.141)
Yeah, I mean, that’s, everyone would love that, right? Cause right now it is a huge process. Like you got to stay on top of it and analyze and optimize and tweak and test. you know, there’s so much that goes into it as you know, but I.

Hikari Senju (11:43.926)
And people might not be really the best at it either, right? Some people might be really good at it, but the average person who also has to juggle many other jobs, maybe you’re the CEO of company, maybe you’re the marketer, you’re doing other things, marketing, to also have to do that is, you know.

Jeffro (11:48.134)
Bye.

Hikari Senju (12:00.007)
is not, you might not be the best person for that. So it actually may be better to actually have an AI do that on your behalf. And then the AI can do it 24 seven, the AI works on weekends. We’ve had problems with campaigns in the past where like, ads are running over the weekend and it’s like, I misspent money and no one’s working. right. So like you don’t have those kinds of issues with if it’s AI powered, if it’s AI trend.

Jeffro (12:08.743)
Exactly.

Jeffro (12:25.733)
Yeah, because it can, yeah, like you said, it can work while you sleep. So it’s much faster and more efficient. I wanted to ask you about real time data. How does that factor into AI driven ad creation? currently, is this a manual process that you have to say, OK, I’m ready, create another new ad? Or does it just do it based on what it’s seeing?

Hikari Senju (12:47.477)
Yeah, so today there is a human in the loop. So the human has to approve the content before it’s launched. So we get the data in real time. We’re ingesting data constantly from all the various app platforms for the customer. We’re getting new insights from the customer all the time. But it’s up to the customer to then approve the ads so that they’re launched. So they’re in the driver’s seat there. That said, think ultimately, you know,

a full self-driving system is gonna be better performing than a co-pilot system because the full self-driving system can be more reactive, can optimize the ads of weekends.

And maybe I’ll also be able to do things that a human might not be like, it might not make sense for a human to launch this kind of app. An AI, as long as it’s aligned with the brand, AI might have some unique insight as to why they’re running that creative, and it might actually outperform. And so therefore, agentic system probably ultimately will perform better. But for now, it’s very much a co-pilot tool.

Jeffro (13:57.306)
What about if you’ve already got a bunch of ads that are approved for different stages? And how reactive is this system across channels? So let’s say if I’m over on Facebook, I see an ad. Maybe I click on it, but I don’t buy. A few minutes later, I go over to Instagram. Is it going to show me the next ad in the sequence or a retargeting ad? Or am I going to see the same one?

Hikari Senju (14:19.753)
Well, if you go from Facebook to Instagram, it will show you that sequence because it’s the same ad platform. you can definitely do that on Facebook and Instagram and threads. So definitely, you can run that kind of ad there. I think the question is, can I run an ad on Facebook and then show the next sequence, for example, of YouTube?

say Amazon, Twitch, or LinkedIn, Microsoft. And so that’s really where that coordination and orchestration key becomes tougher and therefore an area that we can really help the customer in terms of the way they create their audiences and at least making sure that the messaging isn’t contradictory or like weird or orthogonal to each other.

Jeffro (15:13.904)
So as this gets better, what do you see as kind of the biggest opportunities for marketers in terms of creating these campaigns that are driven by AI?

Hikari Senju (15:25.301)
Well, I think brands are going to become more important than ever, right? Especially as the competition for attention becomes fiercer. Because now anybody can create a good movie and everybody will be able to create a good app because anybody can just generate code now and you can generate videos. And even though I do that with images, in the near future, that’s going to apply for images, video, and apps, and code, and whole entire experiences. And so that means the value of brands are to be more important because it’s only

going to get noisier for the consumer. Consumers are going to be inundated with good content, which means now which service do I trust? Which service do I pay more attention to? It’s going to be the bigger brand. so managing that brand well, building that brand well, orchestrating that brand well, we say that OmniHee is the control hub for your brand.

with the Genitive Experience Management tool for your brand. So I would say in practical terms, it’s really three things. One is getting your brand assets lined up and your brand guidelines and your brand rules really clear so that you can fine tune models on it very well and you can scale it out very well. I would say having that analytics platform in place very well as well so you can actually see how consumers are interacting with your generative experiences and optimizing based on those insights and you have visibility.

And then really embracing these new workflow tools or these GenTech systems to better drive efficiencies and productivity and better performance in your marketing work.

Jeffro (17:02.501)
So I’m assuming you guys also integrate with Google Analytics, like you mentioned, having that set up. Do you integrate with other tracking platforms, too, like Woopra or any of the others that are out there?

Hikari Senju (17:12.493)
Yeah, what we do is we work with the customer and we see what pixels and tracking systems they use, and then we integrate with those. We build integrations with those. So Google Analytics, Google Tag Ministers is popular one. Metapixels is another. Every ad platform has its own tracking system for optimized conversions.

Yeah, and then on top of that, we can do it with customer CRM platform so we can get a longer tail view of that and the return on ad spend, the lifetime value of a particular lead. And so those are some of the elements.

Jeffro (17:46.585)
What about, I mean, know going forward, like there’s more strict privacy laws and data regulation and things like that. Is that going to affect the ability of your platform to make these decisions and have that real time data if a lot of that is getting restricted?

Hikari Senju (18:05.045)
Well, I we’re living in a very restricted environment already, especially when to content. So you have these very vague copyright laws today regarding AI-generated content that makes it very difficult actually for marketing teams to fully embrace Gen.AI because it’s not clear who owns the copyright to say an AI-generated images or an AI-generated empowered image or video. And so once there’s more clarity there, that would actually, I think, increase adoption of this technology. I think we’re still not seeing a whole ton of Gen.AI

content by outside of what you see trending on say social, occasionally on social media is because of that ambiguity regarding copyright and these legal concerns. If you accidentally generate an image of say a famous character, who’s liable for it? You run it, who’s liable for it, right? So.

I think that clarity will actually, 100x increase the performance of the adoption of the ads. I’m sorry, 100x increase the adoption of this technology. And then separately when it comes to data, think that’s always kind of a moving goal post in terms of like who owns the data.

Jeffro (19:09.828)
Okay, because I’m sure…

Hikari Senju (19:30.299)
and data privacy and concerns, and I think that’s for regulation, but advertising is an industry as old as really civilization in some ways. so advertising businesses will always be seeking ways of connecting with customers and notifying customers of their products and services. so, yeah.

Jeffro (19:46.071)
Yeah, because I know there’s always as stuff like this happens and you’re to get a more AI driven advertising, there’s going to be ethical considerations that people have to keep in mind. now this by collecting information from these different places, you’re building a profile on a person, which maybe they didn’t want you to know, right? Or maybe they only wanted this company to know about this and this one to know about that. And so now they’re like, wait, hang on. And so any, any considerations like that that you’ve seen or thinking about?

Hikari Senju (20:12.469)
I think what’s gonna end up happening is more of this personalization is gonna be happening at the edge. So instead of like this kind of centralized data where all the data is collected centrally, where a model processes, I think more likely, a lot of the personalization will be happening at the edge. like on your device, the model will be deployed to where the data is instead of the data going to where the model is as these models become more powerful. And so that’s another way privacy and security is maintained because,

Jeffro (20:23.107)
Mm-hmm.

Hikari Senju (20:42.355)
All the data isn’t just being collectible because if you’re a hacker, there’s nothing monetizing that, instead of hacking all your edge devices, I’m just going to hack this one place where all your data goes to.

Jeffro (20:45.131)
Yeah.

Jeffro (20:53.837)
Yeah, it’s a gold mine.

Hikari Senju (20:55.861)
It’s a goldmine, exactly. But instead, if the data is dispersed across all the edge devices, and then the models are actually just being deployed at these edge devices and utilizing that data in a secure way to say, personalize content or generate these personalized experiences, that’s probably a more secure system as well. I think there’s some technological ways that that’s been, know, technological ways some of these challenges are also being solved.

Jeffro (21:22.115)
Got it. All right, well, let me pull it back. Can you share with us a real world example of a business that have successfully used your platform to kind of improve and enhance their ad performance and reach using this approach?

Hikari Senju (21:33.789)
Yeah. Yeah, so there’s this company called Omiana, and they are a vegan beauty business. And when they signed up, they connected the brand assets, they connected the various ad platforms, and then ultimately, we used AI to get insights about what types of ads are performed well. So we realized, for example, the phrase vegan in the title of the ad drove click-through rates by 20 % up.

talking and focusing on their ingredients and how their ingredients are cruelty free is, know, driving higher impressions. And really, again, focusing on the health elements of the product.

drove high performance. And so this ultimately resulted in a 3.5 increase in return on investment for them and a 2x increase in year over year sales. Kevin, their CEO, said that the exceptional quality and scale of our digital assets have been instrumental in broadening our customer base and driving substantial sales growth, and destroying the strength of our partners in the digital marketing sphere.

Jeffro (22:41.943)
That’s awesome. mean, just even identifying, you know, specific words in the title that are driving an increased click rate is super helpful because that can save you a lot of money on your ad spend.

Hikari Senju (22:53.321)
Bye.

Jeffro (22:54.271)
Is there a certain level that a business should be at in terms of ad spend before this makes sense? Because if you’re only spending a little bit, you might not have enough data to actually take advantage of this.

Hikari Senju (23:06.837)
So the average cost per impression is, so average cost for say thousand impressions is probably somewhere around 10 bucks. So to get to a statistical significant result, you actually don’t need to spend a whole ton of money to get to run like say a test that delivers some threshold of statistical significance. So.

Generally, that said though, I think the more data you have, the better insights you can deliver, the more tests you can run. And so for that, I would say it’s around at least $10,000 or $5,000 a month in ad spend is where we come in historically. But that number is also going down further as we further just make our technology better and more accessible to our customers.

Jeffro (23:52.045)
Well, I mean, yeah, that’s getting within reach for a lot of small businesses, especially small and medium as they’re growing. So that’s good to know that that’s there because they can hop on and as they grow, they get this ability to grow faster with the personalized ads and everything. But I think this is also a good time for us to wrap up. Thanks for joining me today, Hikari. It’s always fascinating to hear from people who are on the ground actually working with AI instead of just some of these headlines that are extreme one way or the other because there are a lot of promising

benefits and I appreciate you kind of walking through so many of these details here. For those of you at home listening, check out OmniKey if it sounds like a good fit for you and your business. And I’ve got one last question for you, Hikari. What prompted you to start OmniKey?

Hikari Senju (24:37.181)
Yeah, so I’m an entrepreneur. I’ve been an entrepreneur basically my entire career. I have a deep passion for helping entrepreneurs and the mission was to democratize growth. so, you know, I think a wealthier role is a role where anybody with a good idea can go to market, connect with customers. So historically I think there’s been a barrier, strategy in terms of advertising.

running, you know, have to have a large budget to be able to advertise certain channels or be able to hire decent agencies. And so

If we can use AI to automate that and have anybody with a good product or good idea be able to connect with their customers, that means a world with more products, more interesting products, a more vibrant world with faster growth and productivity and faster growth in these human experiences and a flourishing of creativity. And so that’s really been the motivating factor. It’s the idea of democratizing growth. It’s really kind of the mission that our team is aligned around. And yeah, it continues to be the motivating factor for us.

Jeffro (25:39.395)
Well, that’s awesome. It sounds like you built a great platform that’s allowing people to do that. thanks again for being here, Hikari. Thanks to all of you guys for listening. If you thought this was valuable, please leave a review for the show. Take care, and we’ll see you next time.

Hikari Senju (25:51.829)
Thank you, Jeffrey.