Understanding Concepts Over Keywords: Inside Inuvo's AI-Powered Targeting with Eric Tilbury

Summary

Marketecture Podcast Summary: Understanding Concepts Over Keywords: Inside Inuvo's AI-Powered Targeting with Eric Tilbury

Release Date: April 14, 2025
Host: Ari Paparo
Guest: Eric Tilbury, Head of Operations at Inuvo
Listen to the episode

---

1. Introduction to the Episode

In this episode of the Marketecture Podcast, host Ari Paparo engages in an insightful conversation with Eric Tilbury, the Head of Operations at Inuvo. The discussion revolves around Inuvo's innovative approach to advertising targeting, emphasizing the importance of concepts over traditional keyword-based methods.

Notable Quote:
Ari Paparo introduces Eric Tilbury

"I'm joined today by Eric Tilbury, who is the head of operations for Inuvo. Eric, thank you for being here."
[00:00]

2. What is Inuvo and Its Concept Graph

Eric Tilbury begins by explaining Inuvo's core technology—the concept graph. This advanced system processes and understands content at a conceptual level rather than relying solely on keywords or URLs. By feeding the concept graph with specific concepts, Inuvo can uncover related ideas and the relationships between them, enabling more precise audience targeting.

Notable Quote:

"What we have at our core is we have a concept graph. And what that concept graph does is so essentially you feed that concept graph a concept..."
Eric Tilbury, [01:10]

3. The Difference Between Concepts and Keywords

Ari Paparo seeks clarification on how concepts differ from keywords or URLs. Eric provides a clear distinction, highlighting that concepts are understood at the sentence level, capturing the intent and sentiment behind the content. For example, while keywords might identify "Toyota Tacoma" and "lemon," the concept graph interprets the sentence as discussing "truck performance" with a negative sentiment.

Notable Quote:

"A keyword is just a word, right?... But if you're targeting at a keyword level with that sentence, you might understand that sentence as something fruit related... But if you understand that at the sentence level in the form of a concept, you're saying, okay, this is actually talking about truck performance."
Eric Tilbury, [02:21]

4. User Interaction with Inuvo's Platform

Ari delves into the practical aspects of using Inuvo's platform. Users can input specific prompts or URLs to define their target audience. The system then analyzes the input to identify relevant concepts and relationships, facilitating the creation of highly tailored audience models. This enables advertisers to target users based on nuanced interests and behaviors rather than broad keywords.

Notable Quote:

"You could be as flexible or as just like honed in as you want... the model will figure out the relationships in terms of concepts."
Eric Tilbury, [06:07]

5. Technology and Integration with DSPs

The conversation shifts to the technical integration of Inuvo's targeting capabilities with Demand-Side Platforms (DSPs). Currently, Inuvo activates its targeting through Trade Desk and utilizes Microsoft's Supply-Side Platform (SSP), formerly known as Xandr. This integration allows for real-time bidding based on the concept-based audience scores generated by Inuvo.

Notable Quote:

"Right now, if you want to activate... it gets pushed through Microsoft's SSP."
Eric Tilbury, [07:48]

6. Company Background and Growth

Eric provides a background on Inuvo, initially founded in 2007 and later acquiring Net Zero. The company's concept graph was developed in a UCLA machine learning lab, backed by $42 million in funding. Over time, Inuvo evolved into an advertising tool, navigating challenges in gaining market recognition amidst a landscape dominated by ID-based ad tech solutions.

Notable Quote:

"The concept graph that we created was actually developed in a machine learning lab at UCLA... it naturally turned into an advertising tool."
Eric Tilbury, [09:08]

7. Challenges and Competitive Landscape

A significant challenge for Inuvo is attribution—the ability to demonstrate the value of concept-based targeting without relying on user identifiers. Eric emphasizes that convincing advertisers to shift from traditional Cost Per Acquisition (CPA) models to alternative measurement frameworks is a hurdle. Additionally, he addresses the potential threat from major players like Google, asserting that attribution complexities could prevent them from easily supplanting Inuvo's specialized approach.

Notable Quotes:

"The biggest company challenge is attribution... a huge challenge in terms of attribution."
Eric Tilbury, [12:41]

"The biggest difference between a billion dollar ad tech company and a million dollar ad tech company is attribution."
Eric Tilbury, [13:54]

8. Lightning Round Insights

In a brief lightning round, Ari and Eric discuss quick insights about the company:

• Biggest Company Challenge:
  Attribution and demonstrating the value of non-identifier-based opportunities.

  "The biggest company challenge is attribution."
  Eric Tilbury, [12:41]

• If Inuvo Were an Animal:
  A wolverine—scrappy, tenacious, and hard to kill.

  "Probably a wolverine. Like very scrappy, hard to kill, and they're very tenacious."
  Eric Tilbury, [14:00]

9. Conclusion

Ari wraps up the episode by thanking Eric Tilbury for his participation, highlighting the innovative strides Inuvo is making in the ad tech space through concept-based targeting. Listeners are encouraged to explore more episodes and interviews available on the Marketecture platform.

Notable Quote:

"Eric Tilbury, thank you so much for joining us."
Ari Paparo, [14:09]

---

Key Takeaways

• Concept-Based Targeting: Inuvo's approach goes beyond keywords, understanding content at a conceptual level to enhance audience targeting accuracy.

• Advanced Technology Integration: Seamless integration with DSPs like Trade Desk and Microsoft's SSP facilitates real-time bidding based on sophisticated audience models.

• Challenges in Attribution: Proving the effectiveness of concept-based methods without traditional identifiers remains a significant hurdle for Inuvo.

• Competitive Edge: Inuvo positions itself uniquely in the market, emphasizing its innovative concept graph and resisting overshadowing by larger ad tech giants through specialized solutions.

For those interested in the future of advertising and AI-driven targeting solutions, this episode provides valuable insights into how Inuvo is shaping the landscape with its cutting-edge technology.

Transcript

[Ari Paparo]

This podcast is brought to you by Adelaide. Media verification and measurement are undergoing major disruption. Legacy players are pivoting to performance. Advertising AI is reshaping brand safety and attention is replacing viewability. Adelaide is leading the shift with au, a new way to assess media quality that scores placements based on their potential to drive attention and outcomes. Before your ads run, think of it like a credit score for media. Finally, a clear view of quality. Before you buy, take the guesswork out of your investment strategy and try Adelaide AU on your next campaign. Welcome to marketecture, where you can get smart fast with in depth interviews of leading technology executives. I am Ari Paparo. I'm joined today by Eric Tilbury, who is the head of operations for Inuvo. Eric, thank you for being here.

[Eric Tilbury]

Thank you for having me.

[Ari Paparo]

It's good to see you off Twitter.

[Eric Tilbury]

Yeah.

[Ari Paparo]

You're taking a break.

[Eric Tilbury]

I exist, all right.

[Ari Paparo]

I exist as well, outside of Twitter. Otherwise known as X. Okay. What does Anuvo do?

[Eric Tilbury]

Yeah, so Anuvo, what we have at our core is we have a concept graph. And what that concept graph does is so essentially you feed that concept graph a concept, right?

[Ari Paparo]

You're saying concept with a C, not content, concept, concept.

[Eric Tilbury]

So you go to a newbo site and what we have is we have a platform. And what you do is you could go into our site and say, okay, this is exactly the audience that I'm trying to target. And what you're doing with that, with that prompt is you're. You're feeding our concept graph concepts. And what that concept graph does is it goes, and it will uncover all of the related concepts to a single concept. And then as you go and seed at those concepts from that prompt, what it does is it looks at the relationships of those concepts and goes, okay, this is what I'm going to try and target. And let me explain the difference of a concept. Yeah.

[Ari Paparo]

What's a concept? What's a concept that's different from a keyword or URL?

[Eric Tilbury]

Yeah, exactly. So, so what we understand the concept as is at the sentence level. So a keyword is just a word, right? So if there's a sentence that says, okay, the Toyota Tacoma is a lemon, right? Like the keywords would be Toyota Tacoma and a lemon, right? But if you're targeting at a keyword level with that sentence, you might understand that sentence as something fruit related, right? So you would serve an ad for adult, right? When you understand that at the set, at the, at the sentence level in the form of a concept, you're saying, okay, this sentence is actually talking about like the performance of a, of a Toyota Tacoma, right? So the concept might be truck performance, right? And then when you call it a lemon, the sentiment of that sentence is actually negative. So you're saying, okay, this is a negative sentiment around truck performance. So as we go and process full URLs, we go and read the full URL, we go and extract all of the concepts from that URL, and then based off of the relationship of the concepts on the page, we can understand what the page is talking about. So when you go and feed our model concepts in the form of a prompt, it's extracting all of the concepts from that prompt and says, okay, this is the model that I need to go and target in real time. And I need to go and figure out where this content is being consumed, how it's being consumed, and then what advertising opportunities in the form of bid requests match this advertiser's model in real time.

[Ari Paparo]

Okay, that makes sense.

[Eric Tilbury]

It was actually very validating for me to hear all of the conversations that you guys were having on architecture was because this content and how do you match brand opportunities in real time to a brand? Right? And I was about to just raise my hand and be like, that's what we're doing right now. That's what we've been doing the whole time. Right?

[Ari Paparo]

So I'm a big believer that in moving from keywords to sort of AI generated understanding of the page I've written about in my newsletter. So it sounds like that's what you're getting at, but explain to me how the end user creates a concept. Like, I log in, I have a campaign coming up for an advertiser. I understand the advertiser. What am I typing into your UI and how much detail and how are you helping?

[Eric Tilbury]

Yeah, so you would go to, you go to our site and then it really depends on what kind of campaign are you trying to set up. Like, are you trying to set up a campaign that's just very much so lower in the funnel? Like, hey, you know, I so, so say your iPhone, for example, and you're saying, I want to, I want to capture anyone that's very much so ingesting content around, upgrading their phone, right? And then you could go to our, our system and simply type in that prompt. You could, you could give it, hey, I want to, I want, I'm interested in users that want to upgrade to whatever version of the iPhone they're on right now, right?

[Ari Paparo]

Literally. Would you literally type into the ui, I want to Find users who are interested in upgrading their iPhone. Is that as conversational as it gets?

[Eric Tilbury]

What you can also do is you can feed it URLs from the web that would match that audience. And what you would do, we would process that URL and then we would see what concepts are actually within that URL and those concepts would be fed to the model. So whatever is not ingested from the, from the original prompt that you just manually write out, it's always good to also have a URL and say, okay, this URL matches this model. And then you could be as flexible or as just like honed in as you want. What we get a lot of times from partners is they'll go to our portal and then they create different funnels and different audiences just to push users down that path to conversion. So for example, you get a lot of people that come in and they make models based off of lifestyles just because off the shelf third party audiences, they're hard to match up with really what you're trying to target. You could get as detailed as I want to target gym goers that are interested, interested in, you know, the supplement Creatine and you know, for whatever reason they like Domino's Pizza as well. Right. And then you make a model based off of that and the model will figure out the relationships in terms of concepts. And then at the time of BitterQuest, right, it will look for those concepts. It does a, you know, it gives it a score of the BitterQuest and then if that score is high enough it will say, okay, this, this is good for you to go and serve.

[Ari Paparo]

I can imagine having a funnel where you start at the top of the funnel, like I don't know what a phone is and then you move down to like, I know what a phone is, I want an Apple phone all the way down to I'm ready to upgrade. So how do we execute this? So you, you've mentioned targeting a bunch of times. So is this a data add on to dsps or do you have your own bidder?

[Eric Tilbury]

Yeah, so, so right now if you want to activate. So it's the buzzword curation, right. So it gets activated through curation. You would go to, go to our site, you would say, I want to activate through Trade desk, here's my seat. Right. And then it gets pushed through Microsoft's ssp. So it's just right to your seat.

[Ari Paparo]

Formerly known as Xander, right?

[Eric Tilbury]

Yeah, yeah, Xander. Xander's a. So I mean, I don't know where they're at with their name yet.

[Ari Paparo]

It's confusing. It's confusing. So is Microsoft the only SSP you currently work with?

[Eric Tilbury]

Yeah, right now, as of right now, it's just Microsoft's SSP is one of the limitations.

[Ari Paparo]

Getting like URL level scoring because I know that's a hot topic right now.

[Eric Tilbury]

No, so they actually have a tool and it's the reason why we actually chose Microsoft. They have a tool which is their real time solution that it goes back to when they were at Nexus. Right. Which gives us actually the full URL understanding. And for us to function we actually need that full URL understanding. So you know, using that tool is just like a slam dunk for us.

[Ari Paparo]

So I said, I skipped over a little bit about your company. So tell me about, you know, size, your company, when it was founded, any funding you've taken to date.

[Eric Tilbury]

Yeah, so the original, the original company. So Anubo bought a company called Net Zero that was founded in 2007. And to. The concept graph that we created was actually developed in a machine learning lab at UCLA. And then that, that funding, there was actually $42 million worth of funding that, that stood that up into the advertising space because when they created it, they actually didn't know what they wanted to do with it. They created this concept graph and they were like, like how do we make money off of it? And then, you know, it naturally turned into an advertising tool. The long, the long evolution of the product and then just like the getting our name out there and telling our story, it was actually hard because a lot of the market was very much so into odd. Like very much so. I build IDs, I target IDs, I track IDs, right. And it solutions that were very much so making decisions off of different identifiers is making, building its audiences off of content. Right. Like it was hard for us to, to really get our name out there and be a, like a staple on every media plan. Because I feel like we were just so far ahead in terms of what needed to be done to target the right users at the right time. But for example, this solution that we have because we're working with concepts, we can make decisions and opportunities. Where there's IDs, we make a lot of decisions and there's opportunities. Without IDs, when you go and look at it in the form of a user tracked cpa, right. It makes it hard to. For your average media plan or your average media buyer to say, oh, like I added a new bow on the plan. You know, I have to understand that my CPA the way that we look at it is different.

[Ari Paparo]

Right.

[Eric Tilbury]

So it's hard for the media planner at the agency side to go to their team and be like, we should look at something in the form of like alternative measurement, like to see if we're actually moving the needle or not.

[Ari Paparo]

Yeah, I gotcha. So you raised $42 million to do. To do this, to be. That's a lot of money for curation.

[Eric Tilbury]

Yeah. Oh, so it was actually a long.

[Ari Paparo]

It's a long, complicated story.

[Eric Tilbury]

It's a managed service like only solution. There's like, it's, it's been since what, 2007? Was that, that first round of funding? Let me interrupt you.

[Ari Paparo]

So the current iteration, how many employees do you have? How big are you?

[Eric Tilbury]

Oh, yeah, the current. So Anuvo right now is about 100 employees. Oh, that's a lot.

[Ari Paparo]

Do you have a business outside of advertising or is that all advertising?

[Eric Tilbury]

Yeah, no. So there's actually two sides of the business. There's a publishing side and there's the ad tech side. So internally we call one side. Bonfire we call one side. It's the intent key side. And the intent key sides really were that our ad tech solution sits.

[Ari Paparo]

Okay, so it's a mature business. It's only half ad tech.

[Eric Tilbury]

Yeah, we're actually public.

[Ari Paparo]

Oh, I didn't know that either. Okay, cool. Let's do a quick lightning round. So what is your biggest company challenge?

[Eric Tilbury]

The biggest company challenge is attribution. So it's what I scream about all the time and it's convincing people that the value is actually in the opportunities that don't contain identifiers, where what Meredith Dash at Marketexture is talking about, you know, 10% of ad tech is all fighting. Well, 10% of the opportunities ad tech is all fighting over. When there's 90% of this just quality audience that's sitting there and being able to activate that in real time is, you know, a huge challenge in terms of attribution.

[Ari Paparo]

Sure. Why won't Google and the other giants ultimately do this and kill you?

[Eric Tilbury]

I think the again, attribution. Right. If they do, it's going to be hard to convince the average buyer to get away from that cpa. And ultimately I even did a tweet about it the other day. The biggest difference between a billion dollar ad tech company and a million dollar ad tech company company is attribution.

[Ari Paparo]

All right, and if Inova was an animal, what animal would it be?

[Eric Tilbury]

Probably a wolverine. Like very scrappy, hard to kill, and they very tenacious, hard to kill.

[Ari Paparo]

I'd love to hear it. All right. Well, Eric Tilbury, thank you so much for joining us.

[Eric Tilbury]

Yeah, thank Foreign.

[Ari Paparo]

Thank you for listening to the marketecture podcast. New episodes come out every Friday and an insightful vendor interview is published each Monday. You can subscribe to our library of hundreds of executive interviews at marketecture tv. You can also sign up for free for our weekly newsletter with my original strategic insights on the week's [e-mail]. and if you're feeling social, we operate a vibrant Slack community that you can apply to [e-mail].