WEBVTT
00:00:07.040 --> 00:00:21.839
According to a recent report by data intelligence firm Truthset, advertisers will waste a staggering$7.4 billion in the connected TV market in 2026, which will negate roughly 40% of all open programmatic ad spend.
00:00:22.000 --> 00:00:22.480
Why?
00:00:22.719 --> 00:00:26.640
Because the underlying audience data is that unreliable.
00:00:26.879 --> 00:00:33.840
The report highlights that the core method of linking an IP address to a physical home is only accurate 13% of the time.
00:00:33.920 --> 00:00:40.479
And Truth Set CEO commented that if marketers can't rely on the data guiding their decisions, they're flying blind.
00:00:40.719 --> 00:00:47.359
Which is why I'm excited to talk to today's guest, Jonathan Barnes, JB, founder and CEO at Supply Monitor.
00:00:47.600 --> 00:00:48.960
Jonathan, welcome to the show.
00:00:49.200 --> 00:00:50.000
Glad to be here.
00:00:50.240 --> 00:00:51.359
It's going to be a fun one.
00:00:51.520 --> 00:01:03.039
This will probably get spicy because we hear terms like audience, signal quality, we hear numbers like 7.4 billion or 40% of programmatic spend being waste.
00:01:03.280 --> 00:01:05.599
What the heck does it all actually mean?
00:01:05.840 --> 00:01:08.959
You know, it's kind of like you know, beauty is in the eye of the beholder, right?
00:01:09.040 --> 00:01:15.920
And I think anytime you you ask somebody, you know, what is waste, you ask 100 people, you're going to get a hundred different answers, right?
00:01:16.239 --> 00:01:27.920
And um you know, I think the way I like to talk about it is you we talk about impressions, supply chain, you know, all these words that we throw out.
00:01:28.239 --> 00:01:36.879
Really, at the end of the day, when you're buying media programmatically, be it CTV or any other format, you're you're really buying a string of data.
00:01:37.040 --> 00:01:39.439
I mean, it's data coming across the bid stream.
00:01:39.599 --> 00:01:47.359
It's it's and it's got all the information about the impression, all kinds of identifiers tied to it, again, so on and so forth.
00:01:47.519 --> 00:02:05.599
But you know, the mechanics aside, with so many actors involved, intermediaries involved in the bid stream, it can be hard to understand you know how many people, how many hops, if you will, an impression has made before it gets to you, the buyer or the bidder.
00:02:05.840 --> 00:02:15.360
And anytime there's more hops, there could be things that go on that uh create, as I think the intermediaries would probably say, efficiencies there.
00:02:15.520 --> 00:02:27.599
But again, one person's efficiency could be someone else's definition of fraud, I mean, or or lack of quality, or loose ID bridging, or so on and so forth.
00:02:27.919 --> 00:02:55.439
So it's it it's a very complex state of affairs, but the reality is uh, and I think everyone in the industry agrees on the fact that fraud or at least low quality signal, be it CTV or anywhere else, is rampant and it eats up a very significant amount of of your media buy if you're not watching your bid stream data carefully.
00:02:55.759 --> 00:03:01.840
And and this is I I think turning into the perfect follow-up to an earlier conversation we had with with Dr.
00:03:02.000 --> 00:03:05.360
Foo talking about a lot of those similar, similar topics.
00:03:05.439 --> 00:03:08.400
And I think if we boil this down, I love that take.
00:03:08.479 --> 00:03:09.280
You're buying data.
00:03:09.439 --> 00:03:12.159
What we're talking about here are effective ad dollars, right?
00:03:12.400 --> 00:03:21.360
Out of one dollar, how many pennies of that are actually going to reach a consumer so that I, as the advertiser, can maximize the impact, right?
00:03:21.520 --> 00:03:23.759
Advertising ultimately being a means to an end.
00:03:23.919 --> 00:03:25.599
You mentioned ID bridging.
00:03:25.680 --> 00:03:27.039
What is ID bridging?
00:03:27.199 --> 00:03:30.080
Where does it where does it fit into this to this story?
00:03:30.400 --> 00:03:41.840
So so uh I IDE bridging is more or less it's someone's probabilistic model that says, you know, well, because we have all these different identifiers in the ecosystem, so many different identity graphs.
00:03:42.000 --> 00:03:45.919
So it's basically a process of kind of bridging those together.
00:03:46.080 --> 00:03:58.719
So we think, you know, someone in the process is saying, well, we have a high enough degree of confidence that, you know, Tim Rowe, that you might be this in a UID, you're this in a ramp ID, and we can bridge that together.
00:03:59.039 --> 00:03:59.280
Great.
00:03:59.439 --> 00:04:03.120
Thank you for that explanation around ID bridging and kind of how that ties in.
00:04:03.280 --> 00:04:13.599
Obviously, supply monitor, it sort of tells us tells us in the name a little bit about what you do and the data that you see, but what problem do you solve and who do you solve it for?
00:04:13.919 --> 00:04:20.639
So we we work with buyers to kind of help them understand what they're buying and and and and buy it more effectively.
00:04:20.720 --> 00:04:36.160
We we want to cut out as much of the waste as possible before it even hits their bidder and as as close to real time as possible, optimize out the waste that continues to surface every second in the programmatic ecosystem.
00:04:36.399 --> 00:04:47.360
So the buyers are really our main client because they have a million other things they have to do, other than looking at log file data and and and parsing through all of that to make sense of that.
00:04:47.519 --> 00:04:50.639
So we kind of try and help take that off their plate.
00:04:50.720 --> 00:04:52.240
So it's not optimizing.
00:04:52.319 --> 00:05:00.800
We basically optimize their bidding process so they can focus on the million other things that they have to worry about as a media buyer and media planning team.
00:05:01.040 --> 00:05:02.319
That makes sense.
00:05:02.720 --> 00:05:09.120
How do you so how do you principally maybe how do you cut out the fraud or the waste before the buy?
00:05:09.279 --> 00:05:10.399
What are you looking for?
00:05:10.560 --> 00:05:17.680
How are you identifying maybe questionable sources of supply or the supply that you know to be fraudulent?
00:05:17.759 --> 00:05:19.680
How are you how are you making decisions about that?
00:05:20.079 --> 00:05:21.120
That's a great question.
00:05:21.279 --> 00:05:25.279
And I mean, there's nobody out there, and I'm not going to claim to say we can cut out all the fraud.
00:05:25.360 --> 00:05:26.879
I mean, no one's gonna be able to do that, right?
00:05:26.959 --> 00:05:29.120
But I do believe there's that's a red flag right there.
00:05:29.439 --> 00:05:31.120
Someone says it's the red flag goes up.
00:05:31.439 --> 00:05:32.319
100%, 100%.
00:05:32.800 --> 00:05:34.720
But I do believe there's better ways to do that.
00:05:34.879 --> 00:05:40.160
You know, the chief way that we do that is via you know curation, supply-side decisioning.
00:05:40.319 --> 00:05:49.600
And it's kind of taking that signal upstream and then filtering it at that point before it even hits the buyer's bitter to even decision against.
00:05:49.839 --> 00:05:52.160
So it's it's we it's a couple of different ways that we do it.
00:05:52.240 --> 00:05:58.879
You know, one, we work with uh data partners that provide us various types of signal that we can look at.
00:05:59.040 --> 00:06:11.600
So whether it's IPs that have been hijacked, household IPs that have been hijacked, it's perhaps hymns that may not be tied to any known online or offline purchases.
00:06:11.839 --> 00:06:14.240
So why are why are we activating against those?
00:06:14.560 --> 00:06:16.079
And so on and so forth.
00:06:16.160 --> 00:06:21.360
But there's a couple of examples of the kind of the data that we're looking at in real time to filter against.
00:06:21.519 --> 00:06:24.000
And then the other part is really kind of playing a whack-a-mole game.
00:06:24.079 --> 00:06:39.360
It's getting looking at log file data, looking at data within a pixel that we use that that's refreshed uh closer to real time, that we can understand what's going on with the qual the quality and the health of that supply.
00:06:39.439 --> 00:06:49.120
Is are there geolocation anomalies that we see from that pixel tells us rather than what's actually shown in the bench stream is one example of that.
00:06:49.600 --> 00:06:51.439
What could that potentially indicate?
00:06:51.759 --> 00:06:54.560
So uh if there was a geolocation discrepancy.
00:06:54.879 --> 00:07:18.560
Yeah, so it could be someone potentially really you know viewing a CTV ad in Singapore, but it could be spoofed to appear that someone in New York City is watching it, and yeah, just modern in Google Analytics, there's a ton of scraping bot traffic coming from overseas.
00:07:18.720 --> 00:07:23.199
I'm sure that's showing up inside of sort of what you're describing here.
00:07:23.519 --> 00:07:34.079
How has AI, not from the the functionality standpoint, but just how has AI maybe been a contributor to this problem as time goes on?
00:07:34.160 --> 00:07:34.879
Is it a factor?
00:07:35.040 --> 00:07:40.240
Is is AI exacerbating the problem, or where where do you see the bad actors originating?
00:07:40.560 --> 00:07:44.480
So AI is kind of a bit of a double-edged sword, in my opinion.
00:07:44.639 --> 00:07:56.399
Um, it's been amazing to help us fight the fraud and the garbage, but it's also allowed the fraudsters to be more efficient with with their tactics as well.
00:07:56.560 --> 00:08:16.800
So the then the net net is organizations that are actively working with companies out there to to mitigate these issues, to actively clean up their supply paths, clean up their supply stream, are are able to leverage AI and be very effective, I think, at what they're at what they're doing.
00:08:17.040 --> 00:08:29.199
The folks that are not are seeing a lot even more waste because again, the the the fraud is you know, I'm going back to that whack-a-mole game, it's it's something you have to constantly be keeping up with.
00:08:29.519 --> 00:08:37.440
So so having some of those AI detection mechanisms is like having having a guard on duty 24-7-365.
00:08:37.600 --> 00:08:45.919
Hey, we're there's always going to be new threats popping up, but we're always monitoring, it's an always-on monitoring system that that probably gives us coverage we didn't have.
00:08:46.240 --> 00:08:46.639
100%.
00:08:47.120 --> 00:08:51.679
I mean, the the reality is so much information lives in silos right now.
00:08:51.759 --> 00:09:17.279
And the the biggest thing that we've seen for us and just anybody in the ecosystem is AI has allowed the iteration of kind of building code to kind of build bridges across those silos and analyze data and bring disparate data points together to look at and analyze and then action against exponentially faster than you could, I mean, even a year ago.
00:09:17.519 --> 00:09:20.000
And it's pretty amazing to see how that works.
00:09:20.080 --> 00:09:21.360
I mean, you mentioned GA4.
00:09:21.600 --> 00:09:32.559
I mean, a lot of times you know, optimizations are done by programmatic media buyers based upon the truth that they see in their DSP report, perhaps, right?
00:09:32.799 --> 00:09:33.759
That's great.
00:09:34.000 --> 00:09:38.960
But GA4 or Adobe Analytics, whatever analytics platform might be telling a different story.
00:09:39.200 --> 00:09:42.720
Or if you're using another analytics tool like Dr.
00:09:42.879 --> 00:09:52.879
Foo's Foo Analytics, you know, you're working with like Jounce Media, so on and so forth, these are still disparate platforms telling you different pieces of information.
00:09:53.120 --> 00:10:07.440
And with AI, I mean, you you can you can pretty quickly spin up a tool that can put these pieces of information together, parse it together, and then give you a holistic view so you can action against uh all the information more quickly.
00:10:07.759 --> 00:10:12.720
Organizations that are doing that are really kind of winning the day and staying ahead of this as best you possibly can.
00:10:12.960 --> 00:10:19.759
That sounds like a good follow-up conversation for us is all about AI in advertising, ad fraud.
00:10:20.000 --> 00:10:23.679
That's that's probably a whole series right there that we'll have to come back to.
00:10:24.000 --> 00:10:24.559
Absolutely.
00:10:24.799 --> 00:10:33.120
I would love JB to spend a few minutes double-click into specifically what you see as it relates to connected TV.
00:10:33.279 --> 00:10:34.879
What are the best brands?
00:10:34.960 --> 00:10:35.840
What are they doing?
00:10:36.080 --> 00:10:44.720
What are maybe some of the pitfalls to avoid success in CTV as it relates to what we're talking about here today, audience signal quality.
00:10:44.879 --> 00:10:45.919
What does that look like?
00:10:46.159 --> 00:10:51.360
Yeah, so connected TV, in my mind, it's similar to the mobile app universe.
00:10:51.519 --> 00:10:58.559
When you're buying from an app-based environment, it works a little bit differently than when you're buying from a browser-based environment.
00:10:58.879 --> 00:11:09.679
So there's you kind of have a bit of like a hybrid, almost like a like a walled garden, you know, stuff going on because you just can't, you don't have the visibility into what's going on there.
00:11:09.919 --> 00:11:15.519
So, you know, my two cents on that for anybody buying is to do as much as you possibly can direct.
00:11:15.679 --> 00:11:28.080
If you can either do a direct buy, programmatic direct, to kind of skip those potential third-party intermediary hops, that's going to be obviously the most efficient way to do things.
00:11:28.240 --> 00:11:44.080
But the whole reason, in my opinion, you know, programmatic came into existence is no media team, even at the largest of whole codes, has enough people you can throw at and managing a million different direct relationships and understand how this all we've come full circle.
00:11:44.320 --> 00:11:44.480
Right.
00:11:44.559 --> 00:11:51.200
So eventually, you know, again, even the biggest media teams reach a wall where it's just like, okay, we got to find scale more efficiently.
00:11:51.279 --> 00:11:53.600
And that's kind of what programmatic does.
00:11:54.000 --> 00:12:13.759
And I think that when it comes to connected TV, you know, I think it's you know, step one is just really, you know, working with the platform or platforms that you work with to understand what guardrails they have in place to protect from the known fraud that goes on in the CTV space.
00:12:13.919 --> 00:12:15.360
I mean, just have that conversation.
00:12:15.440 --> 00:12:18.639
I'm I'm so I'm always surprised by two things.
00:12:18.720 --> 00:12:22.559
You know, media teams again of all sizes don't ask these questions.
00:12:22.720 --> 00:12:28.720
So they they don't know kind of what the latest and greatest is of their DSP is doing.
00:12:28.879 --> 00:12:37.919
And and two, quite often, SSPs are involved providing traffic for CTV, creating these you know, multiple hops that are going on.
00:12:38.240 --> 00:12:51.039
And buyers aren't reaching out to those SSPs, whether they have a direct relationship with them or not, at least reaching out and understanding EO that coming up, hey, OpenX, hey, index magnite, so on and so forth.
00:12:51.279 --> 00:12:57.759
What are you doing to ensure that your supply chain is is healthy in terms of connected TV?
00:12:57.840 --> 00:13:00.000
Well, how do you vet resellers that show up?
00:13:00.080 --> 00:13:02.639
And in your opinion, what value do they really add?
00:13:02.799 --> 00:13:13.679
You know, those are questions that you could be asking, uh, again, not just your DSP, but anybody that you have visibility into in the process of buying CTV.
00:13:14.080 --> 00:13:18.960
And I think the last thing I'd say is I'm a big believer in curation.
00:13:19.039 --> 00:13:24.159
I'm a big believer in self-side decisioning, kind of however you want to call it these days.
00:13:24.240 --> 00:13:32.399
Um, and that doesn't necessarily mean you have to do it yourself, but just having control over the supply that's being shaped, that's being sent to you.
00:13:32.559 --> 00:13:48.480
So whether it's working with a third-party uh curation service or going deeper with an SSP that you you currently buy a lot of connective TV supply from and getting into the platforms that they they offer to to cure an inventory.
00:13:48.639 --> 00:13:53.279
So you have control, more control over that and transparency into what's going on.
00:13:53.519 --> 00:13:53.919
Very good.
00:13:54.080 --> 00:13:55.360
Let's recap those.
00:13:55.600 --> 00:13:58.399
For the inventory, you've got to have go direct.
00:13:58.559 --> 00:14:00.480
Traffic programmatically, go direct.
00:14:01.200 --> 00:14:03.600
Second, talk to your partners, ask good questions.
00:14:03.759 --> 00:14:05.600
You outline three good questions there.
00:14:05.679 --> 00:14:08.559
We'll make sure that they are easy to find in the show notes.
00:14:08.720 --> 00:14:12.000
And the last piece, thoughtful, intentional curation.
00:14:12.080 --> 00:14:15.279
It's not just a buzzword, it's a practice and it's ongoing.
00:14:15.440 --> 00:14:17.440
Not just a set it and forget it thing either, right?
00:14:17.679 --> 00:14:18.639
No, absolutely not.
00:14:18.799 --> 00:14:24.720
I mean, I said earlier, you know, these threats to your to your media buyer are always evolving.
00:14:24.799 --> 00:14:30.879
And again, as we've always already covered, AI is just making that move exponentially faster.
00:14:31.120 --> 00:14:37.919
So, you know, leverage AI tools again, perhaps whether it's a third-party team or I mean, just the stuff that you can build.
00:14:38.080 --> 00:14:40.480
I mean, I I'm not a technical person personally, my background.
00:14:40.559 --> 00:15:01.200
I've always had a team developers, yeah, that, but I I actually write a lot of code now using Claude code that again might that I can hand off to a product developer now, and they at least tell me that, hey, this is all good or mostly good now, which even six months ago wasn't true, a year ago wasn't true.
00:15:01.440 --> 00:15:09.519
So I I would challenge media buyers that are watching to spend some time with that, with, with, with these AI tools yourself.
00:15:09.759 --> 00:15:28.559
I think you'll be surprised what kind of what you can do yourself that you can hand off to uh a product team that you work with or an engineering team and kind of say, hey, like I was kind of tooling around here, and if we could kind of create something like this and spit out some code for me, I mean, can we use this?
00:15:28.720 --> 00:15:43.279
If you could get, you know, everyone's strapped for engineering resources, but if you can help get an engineering resource even halfway home before they even have to start from scratch from building out a custom tool for you, that's a huge win.
00:15:43.360 --> 00:15:47.440
And and I'd urge anybody to tinker around and see what you can do.
00:15:47.600 --> 00:16:07.440
I think you can be surprised at what you might be able to do to customize some kind of solution to make your job easier, really across the board, but even within analyzing again these siloed data sets that you have, bringing them together so you can then make quicker decisions on improving your supply quality.
00:16:07.759 --> 00:16:25.440
It really might be the biggest opportunity, broadly speaking, for the industry, for teams to just iterate faster using AI to create a proof of concept, whether it's all the way there, it's the start of a code base, or it's just a mock-up made with nano banana, right?
00:16:25.519 --> 00:16:37.679
Just having a point of reference because it's it's really the translation layer of what's in my brain and getting what's in my brain into your brain as quickly as possible so that we can do something useful with it.
00:16:37.759 --> 00:16:40.159
I think that's a great, great lesson for everyone.
00:16:40.480 --> 00:16:41.120
Absolutely.
00:16:41.279 --> 00:16:42.000
Absolutely.
00:16:42.320 --> 00:16:44.480
You teased in the lead up a little bit.
00:16:44.559 --> 00:16:51.039
Uh, we can't announce any specific partnerships, but I know you've got some big partnerships lined up for later this year.
00:16:51.200 --> 00:16:56.320
Anything you can tease out any cool use cases uh that are worth mentioning here today?
00:16:56.559 --> 00:17:07.119
Yeah, I think that uh again, there's a lot of good companies out there that that have been working on the invalid traffic, fraudulent traffic, you know, traffic validation, coming back from some different ways.
00:17:07.440 --> 00:17:19.759
But we are really, in addition to that, tackling what I call invalid signal, um, which again is getting fraud, perhaps looking at if not fraudulent signal, low quality signal.
00:17:19.920 --> 00:17:21.519
There's a lot of great data providers.
00:17:21.599 --> 00:17:28.720
It's you know, it's funny, it's yeah, I kind of go to some of these data because I have a strong background in data as a co-founder of a data company.
00:17:28.880 --> 00:17:32.720
So reached out to a lot of my old contacts and kind of say, Hey, what does your trash look like?
00:17:32.799 --> 00:17:33.359
I want to look at it.
00:17:33.440 --> 00:17:40.240
I want to go through your trash and kind of sort that together because this is because you said one man's trash, another man's treasure, right?
00:17:40.400 --> 00:17:46.960
So it's so we can go through that and and because a lot of these data points are are are showing up in the bid stream.
00:17:47.039 --> 00:17:57.039
And and so it's just if it's if there's data companies out there that are saying that this data is not of quality, then why should we be activating against it or buying media against it?
00:17:57.200 --> 00:18:00.160
And and there's a lot of cool stuff going on with that right now.
00:18:00.240 --> 00:18:02.319
And then we're kind of ramping up this year.
00:18:02.480 --> 00:18:16.720
I mean, I know we've got a big midterm election coming up, and and and and political advertisers, especially, are our targets for fraud because they don't have sales outcomes they can activate to.
00:18:16.799 --> 00:18:18.640
They spend a lot of money in a short amount of time.
00:18:18.720 --> 00:18:24.079
So it's like you can hold over money and get or get a make good to go put it towards the next election, right?
00:18:24.240 --> 00:18:25.599
They've got to spend that money.
00:18:25.759 --> 00:18:36.000
So doing some cool things uh in that space to kind of help them make sure that their their media buys are going to affect the media.
00:18:36.160 --> 00:18:40.720
Cause I know we see all the money that gets thrown into political campaigns, but really kind of go to the top ones.
00:18:40.799 --> 00:18:46.400
I mean, there's a lot of good candidates down ballot that want to spend on connective TV that don't have a lot of money to spend.
00:18:46.480 --> 00:18:53.200
So I want to help those folks make sure their message is getting out to the right people and the right audience with the little bits of money that they they have.
00:18:53.359 --> 00:18:54.400
So that's great.
00:18:54.640 --> 00:18:59.839
JB, if folks want to learn more about you, supply monitor, follow you on socials, where should they go?
00:19:00.079 --> 00:19:03.440
Yeah, uh so uh supply monitor.ai is our website.
00:19:03.519 --> 00:19:16.640
Uh, you can reach me at jb at supply monitor.ai and uh Jonathan L Barnes, J-O-H-N A-T-H-A-N, L Barnes on Twitter.
00:19:16.799 --> 00:19:19.759
Um X, sorry, I I'll I don't think I'll ever change.
00:19:19.839 --> 00:19:21.200
It's I it's hard it's hard.
00:19:21.599 --> 00:19:22.799
I'll have this die hard.
00:19:22.880 --> 00:19:26.880
We'll we'll make sure to link to all that so it's easy for everyone to find in the show notes.
00:19:27.119 --> 00:19:32.400
JB, I can't thank you enough for prompting this conversation and for the insightful dialogue today.
00:19:32.640 --> 00:19:32.960
Great.
00:19:33.039 --> 00:19:33.839
Thanks for having me.
00:19:34.000 --> 00:19:34.559
Thanks so much.
00:19:34.720 --> 00:19:38.640
If you found this conversation to be helpful, please share it with a colleague or a client.
00:19:38.799 --> 00:19:41.920
Start a conversation today, and we'll see you next time.
