TRANSCRIPT 🔗
WEBVTT
00:00:03.200 --> 00:00:09.359
Finding the right audience is also avoiding the wrong audience.
00:00:09.519 --> 00:00:09.759
Yes.
00:00:10.000 --> 00:00:15.919
Providing identity, but not limiting my needs as a publisher to one area or another.
00:00:16.079 --> 00:00:17.519
I don't want to pick.
00:00:17.760 --> 00:00:19.039
I want flexibility.
00:00:19.120 --> 00:00:25.679
If you think about a very simple life, that you say I'm going to buy an ounce of gold.
00:00:26.000 --> 00:00:26.559
Right?
00:00:26.800 --> 00:00:32.000
So it looks gold, it's shiny, but what if it was like diluted 50%?
00:00:32.479 --> 00:00:33.600
Would you still buy it?
00:00:33.840 --> 00:00:34.399
Not at that point.
00:00:34.960 --> 00:00:36.000
Not exactly.
00:00:40.159 --> 00:00:46.640
So when we talk about publishing, when we talk about publishing, we're usually talking about, well, where is the money?
00:00:46.880 --> 00:00:49.359
Or in this case, where is the problem?
00:00:50.159 --> 00:00:54.000
I mean, you've been doing this nearly 20 years at IntentIQ.
00:00:54.399 --> 00:00:56.399
You know, solving the identity problem.
00:00:56.479 --> 00:01:02.640
I mean, is there is there so much to solve and work on and keep you interested above all for nearly 20 years?
00:01:03.119 --> 00:01:03.359
Right.
00:01:03.600 --> 00:01:10.719
You start with one challenge and you see that the industry evolves and the challenges evolve with it.
00:01:10.879 --> 00:01:17.439
So mobile is getting more into being a very big problem for identity.
00:01:18.319 --> 00:01:24.480
Back then, identity has the ability in mobile to provide a single ID across all apps.
00:01:24.560 --> 00:01:25.840
And now it changes.
00:01:26.000 --> 00:01:28.879
And it seems like Android is following Apple as well.
00:01:29.120 --> 00:01:31.359
CTV is becoming a real rush.
00:01:31.519 --> 00:01:33.120
Everyone goes on CTV.
00:01:33.200 --> 00:01:40.480
So now the problem becomes even more challenging, understanding people within the house to their personal devices.
00:01:40.719 --> 00:01:47.680
So as the industry evolved, we found ourselves new problems to solve and new challenges to overcome.
00:01:48.000 --> 00:01:53.760
But then we also get into AI technology that are helping us all together.
00:01:54.079 --> 00:02:01.359
So it's kind of a challenge that grows as we grow, and it's not like one problem to solve and that's it.
00:02:01.599 --> 00:02:02.879
It's always moving.
00:02:03.359 --> 00:02:11.439
We are on the way to being finally and truly omnichannel after speaking about it for well over a decade.
00:02:11.840 --> 00:02:15.919
That's a problem to solve, but how much is there still to solve?
00:02:16.400 --> 00:02:55.759
I think when mentioning omnichannel in terms of identity, it's having a unified solution that identifies the persona on across different devices, but it doesn't stop in terms of okay, I want to sell as a publisher and understand the same persona because then it meets the other side of the buy side, and that's where the advertiser think of a simple use case when the publisher wanna cap frequency across omni-channel, five ads on CTV, then few ads on in app, and then few ads on the web.
00:02:55.840 --> 00:02:58.960
And the person is moving from office to house.
00:02:59.039 --> 00:03:11.680
So our goal is really to understand the relationship between publisher identity side, and as identity, you also meet them on their needs when they're coming from the buyer or the advertiser side.
00:03:11.919 --> 00:03:20.000
So omnichannel from the eyes of the publisher is being able to monetize the user and understand the persona.
00:03:20.479 --> 00:03:29.680
But on the buyer side is understanding that the same person that you met and you recognize on the web is the same person on the CTV.
00:03:30.319 --> 00:03:33.199
And that's always changing and signals always changing.
00:03:33.439 --> 00:03:39.280
So as you pointed out yourself, you don't want to overdo it.
00:03:39.439 --> 00:03:43.759
So you are having consumers moving across the channels.
00:03:44.000 --> 00:03:49.199
You want to show them the ads that are relevant and ones that they will appreciate.
00:03:49.280 --> 00:03:53.280
But again, capping frequency, that is very hard.
00:03:53.599 --> 00:03:53.840
Yep.
00:03:54.719 --> 00:04:01.439
So I think even understanding that there is a problem on the buy side, it's not that clear.
00:04:01.919 --> 00:04:11.360
Because they're putting the budget, they're targeting the audience, and targeting audience is basically setting a list of identifiers.
00:04:11.759 --> 00:04:16.399
That's my audience, male or female, interest.
00:04:16.879 --> 00:04:38.079
So when we look at it, we understand that when we create the matching between publisher identity and their goal to improve and yield performance, we're also thinking about the buyers and their ability to cap their budgets or even a simple thing like doing ad suppression.
00:04:38.879 --> 00:04:41.279
I'm seeing the user on different apps.
00:04:41.439 --> 00:04:44.079
You the advertiser already bought the user.
00:04:44.240 --> 00:04:47.120
Without identity, you cannot do ad suppression.
00:04:47.439 --> 00:04:48.000
Yes.
00:04:48.480 --> 00:04:53.839
So instead of spending the money where you should, you're just spreading it across the board without identity.
00:04:54.160 --> 00:04:59.920
Finding the right audience, it's also avoiding the wrong audience.
00:05:00.160 --> 00:05:00.399
Yes.
00:05:00.560 --> 00:05:01.519
That's what you don't want.
00:05:01.680 --> 00:05:03.600
You don't want duplicated reach.
00:05:03.920 --> 00:05:14.399
Right, because if you don't know that you already saw that user on a different app, and that's what happened on iOS, like every app has its own ID, but still there is the challenge of privacy.
00:05:14.560 --> 00:05:20.959
How you keep privacy in an era where you want to have identity and understand as a persona.
00:05:21.120 -->
Load full transcript
00:05:27.839
So there are many challenges, and everyone that deals with it understands that they need to always be on top of regulation.
00:05:28.000 --> 00:05:32.560
For instance, in the US, there are several types of regulation, just in the US.
00:05:32.639 --> 00:05:32.879
Yes.
00:05:33.040 --> 00:05:34.160
And then you have Europe.
00:05:34.560 --> 00:05:43.759
They need to think of a long term, like they need the experience of the user in their website relevant, not just selling more, because eventually it will come back to them.
00:05:44.079 --> 00:05:45.839
Identity, privacy.
00:05:46.000 --> 00:05:48.639
We've been talking about it for a while.
00:05:48.959 --> 00:05:51.360
Why so central now?
00:05:51.519 --> 00:05:55.920
Is it the omnichannel is the driver, or what is what is change?
00:05:57.040 --> 00:05:58.160
You have to be compliant.
00:05:58.319 --> 00:06:15.120
At the end of the day, there are many moving parts that require a holistic solution because you might be finding a user belonging to one audience based on IDA, and then there is another audience which is on ID B.
00:06:15.360 --> 00:06:22.639
So the publisher needs to accommodate multiple identifiers, sometimes over 20, in a single auction.
00:06:22.959 --> 00:06:28.639
And we found ourselves as one of the value is reducing the hustle of operation.
00:06:28.800 --> 00:06:38.720
If you think of a publisher, if they need also to manage other than content, the different relationship with different identity, because every SSP has their own ID.
00:06:38.879 --> 00:06:40.480
Every DSP has their own ID.
00:06:41.040 --> 00:06:51.040
So I think what we found is that you need to have a very reliable house of identity that could provide you the solution that would fit your needs.
00:06:51.360 --> 00:06:57.199
Providing identity, but not limiting my needs as a publisher to one area or another.
00:06:57.439 --> 00:06:58.879
I don't want to peak.
00:06:59.120 --> 00:07:00.480
I want a flexibility.
00:07:00.879 --> 00:07:04.399
And I think this is very important in a technology event or of identity as well.
00:07:04.879 --> 00:07:20.000
That's also changing, as you pointed out, that that the different focus and the different questions publishers need to be asking themselves and the different stress on them dealing with the identity as we go across more and more devices.
00:07:20.240 --> 00:07:24.000
So how would you actually then describe intent IQ?
00:07:24.560 --> 00:07:34.000
So we evolved from starting on a device graph that we were uh selling to uh DSP as a service.
00:07:34.480 --> 00:07:45.519
And then we come to realize that not every entity that needs identity can ingest this large volume of data or to ingress the data.
00:07:45.680 --> 00:07:50.879
So what we we found is you need to build solution on top of identity.
00:07:51.199 --> 00:08:09.839
So for instance, in our case, we have a product called Beat Enhancement, and Bid Enhancement is all about helping to get the product
of identity very easy to use instead of hosting huge volume of data and hosting huge volume of processing units.
00:08:10.240 --> 00:08:18.879
they just need the solution that helped them at the right moment, Like we call it a millisecond before the bid leaves the auction.
00:08:19.360 --> 00:08:29.199
And that's where we found ourselves providing a solution on top of identity, the mechanism, accuracy, and everything that we we can uh discuss.
00:08:29.680 --> 00:08:44.000
But providing the solution, whether it's an attribution, whether it's onboarding, or for publisher, a service that connects in real time, making the technology very accessible.
00:08:44.240 --> 00:08:49.120
If like looking back five, six years, not every publisher can take the solution.
00:08:49.360 --> 00:08:58.399
He needs to host a big solution, and now you're taking like the last mile of identity, just matching real time and provide me the IDs.
00:08:58.559 --> 00:09:07.360
Today a publisher needs a set of IDs, and tomorrow they have a need for uh, for instance, ID that are based on deterministic login data.
00:09:07.519 --> 00:09:12.240
Okay, I have my ID as a company, as an identity with my own demand.
00:09:13.200 --> 00:09:19.120
The publisher doesn't need to pick or commit to one because there is no single bullet of ID that works.
00:09:19.519 --> 00:09:29.840
That's where we fit ourselves in a solution that is very flexible and it's like an identity agnostic with 60 different IDs in one solution.
00:09:30.240 --> 00:09:35.600
We talked about what matters more also in an omnichannel world.
00:09:35.759 --> 00:09:47.200
So a recent identity infrastructure 2.0 report um point out that a loan just hashed email to postal linkages, which is very basic.
00:09:47.919 --> 00:09:53.679
Averaged only 51% accuracy, which means 49% inaccurate.
00:09:53.840 --> 00:09:57.120
I mean, that is a startling number for me right now these days.
00:09:57.279 --> 00:09:59.200
I mean, how would you explain it?
00:09:59.279 --> 00:10:00.320
What do you make of that?
00:10:00.720 --> 00:10:08.080
But if you think about a very simple life that you say I'm going to buy an ounce of gold, right?
00:10:08.639 --> 00:10:13.919
So it looks gold, it's shiny, but what if it was like diluted 50%?
00:10:14.320 --> 00:10:15.440
Would you still buy it?
00:10:15.759 --> 00:10:16.320
Not at that moment.
00:10:16.879 --> 00:10:17.600
Not exactly.
00:10:17.759 --> 00:10:30.559
So it's like when you think about what you've just mentioned, there is a real problem of matching real deterministic hashed email to a person, not even speaking on a house for like 50, 49%.
00:10:31.360 --> 00:10:36.399
Think about the challenge to connect it to a person on an in-up moving outdoor.
00:10:36.960 --> 00:10:39.519
So the accuracy becomes really a challenge.
00:10:39.679 --> 00:10:49.360
We keep monitoring, and we see if you don't do the monitoring, you are not aware of the fact that it's like taking half of your dollar and throwing it away.
00:10:49.679 --> 00:10:54.639
So, how is it possible that publishers are really confusing?
00:10:54.720 --> 00:10:57.279
You know, I can match more because we can do more.
00:10:57.519 --> 00:10:58.720
We'll get to AI in a moment.
00:10:58.799 --> 00:11:00.720
You know, scale, it's all about scale.
00:11:00.960 --> 00:11:05.279
When actually it's more like, have I matched correctly?
00:11:06.080 --> 00:11:06.559
Exactly.
00:11:06.720 --> 00:11:14.960
So as a publisher, they have a lot of pressure to gain performance and numbers and management want to see growth.
00:11:15.279 --> 00:11:23.519
And as the the the media evolve and more lost of signals, they want to push for coverage.
00:11:23.759 --> 00:11:38.000
If as a publisher you are ignoring the fact that maybe 50% is wrong, then the experience of the ads that are being sold on your site would come back to you as a boomerang.
00:11:38.399 --> 00:11:43.919
Trying to maintain 92 to 95% accuracy, it's not an easy task.
00:11:44.320 --> 00:11:46.000
You're measured on your ad experience.
00:11:46.159 --> 00:11:48.960
That's part of your performance, that's part of your brand performance.
00:11:49.039 --> 00:11:53.600
If you are a brand involved in that as well, what is the expense?
00:11:53.679 --> 00:12:04.480
You know, what what is the number that matters most if it's not about the expense of keeping the customer, losing the customer?
00:12:04.559 --> 00:12:07.759
I mean, what is at stake if you're not matching correctly?
00:12:08.159 --> 00:12:16.159
First, on the advertiser buy side, they're actually getting less and paying more.
00:12:16.879 --> 00:12:19.519
So their performance because they bought the diluted gold.
00:12:19.840 --> 00:12:23.679
They they're like spread and pray, and that that's wrong.
00:12:23.919 --> 00:12:32.720
So publisher in the short term may gain value by saying, I don't care as long as the number grows, but eventually it's like what happened on MFA.
00:12:32.960 --> 00:12:36.720
MFA three years ago was like a jungle.
00:12:36.879 --> 00:12:42.960
Everyone was buying user from social and then selling ads, yeah, and now it stopped.
00:12:43.200 --> 00:12:51.759
So think what happened if all this idealist traffic on publisher will suddenly say, okay, let's take a look.
00:12:52.000 --> 00:12:55.039
Is this the real person that you are selling or not?
00:12:55.200 --> 00:12:56.960
And then the party will stop.
00:12:57.279 --> 00:12:58.639
It's like a music chair.
00:12:58.720 --> 00:13:04.159
Everyone is now playing in that, but once the movies the music stops, you're losing.
00:13:04.879 --> 00:13:14.559
So I think we as an identity vendor putting this accuracy and the understanding that we need to measure ourselves deterministically very important.
00:13:15.039 --> 00:13:16.720
Don't trust, measure.
00:13:16.960 --> 00:13:17.919
That's the takeaway.
00:13:18.240 --> 00:13:20.879
So let's stay with the buyer for the moment, right?
00:13:21.440 -->
00:13:30.320
Buyers out there, three identity companies are pitching and they're saying, hey, we have scale, we have reach, we have accuracy.
00:13:31.200 --> 00:13:33.200
What helps make the right decision?
00:13:33.279 --> 00:13:44.960
I mean, is there a non-negotiable here for someone who's really just starting to understand and wants to make certain that they have to prove what they offer?
00:13:45.440 --> 00:13:51.200
What they have, they have precious data that their customer brought them when they are buying.
00:13:51.279 --> 00:14:12.399
So if you take a true set and you want to evaluate three vendors, and you take a small set that you know for sure, and you can provide uh like a hidden test saying, okay, I'm giving you 5%, but in it's part of like 100% of data, and I want to see how you're gonna match this unique data.
00:14:12.480 --> 00:14:14.879
Like it's your own first-party data.
00:14:16.399 --> 00:14:41.120
It's be foolish not to evaluate based on your and it's very simple, just an offline test could demonstrate whether this vendor is saying what they are, and if this 5% that you know for sure is matched, whether it's uh a zip code, like you said, to a hashed email or mobile ID to other IDs because the the user purchased, they gave you all the information, evaluate before you decide.
00:14:41.679 --> 00:14:44.000
So you know this so well, you're up for the challenge.
00:14:44.240 --> 00:14:44.879
Yeah.
00:14:45.360 --> 00:14:51.519
What about buyers when they're thinking about, you know, because scale, scale is attractive.
00:14:51.759 --> 00:14:54.159
And they're saying scale versus accuracy.
00:14:54.240 --> 00:14:59.440
You know, I can live, I I talk with many publishers also that come on mobile group.
00:14:59.600 --> 00:15:00.879
What do you say to them?
00:15:01.360 --> 00:15:07.919
So I think this is a word that could come from the mouth of publisher because there are less the uh that's true.
00:15:08.000 --> 00:15:08.960
The buyers wouldn't say that.
00:15:09.200 --> 00:15:26.559
But the buyer wouldn't say that, and when it comes to buyer, if they're willing to spend, it's like I have a great campaign for McDonald's or for a brand new iPhone and I want to publish in New York, but actually my budget goes in India and I'm not selling there.
00:15:26.720 --> 00:15:28.639
Is it a good decision to make?
00:15:28.799 --> 00:15:29.759
I don't think so.
00:15:30.159 --> 00:15:30.399
Right?
00:15:30.559 --> 00:15:37.840
So I think from the from the buy spot buy side perspective, you need to measure cross-device attribution.
00:15:38.639 --> 00:15:51.759
And by the way, some of the budget are being avoided from ID list and uh publishing on premium iPhone user because of that problem.
00:15:52.480 --> 00:15:58.480
So you either spend the money unwisely or you avoid it completely because you're not measured.
00:15:58.559 --> 00:16:04.960
And we say if you have a good identity spine, then you can get two words.
00:16:05.120 --> 00:16:06.799
There is a way, you need you need to work.
00:16:07.360 --> 00:16:08.240
You need to do some work.
00:16:08.639 --> 00:16:08.879
Okay.
00:16:09.279 --> 00:16:10.960
Well, look at the publisher now.
00:16:11.120 --> 00:16:15.679
I spoke about how publishers can deal with a little bit of inaccuracy.
00:16:16.000 --> 00:16:24.639
But you know, do they understand what they need to be thinking about before they start to measure value?
00:16:24.799 --> 00:16:31.440
It's not just the scale, you know, it's engineering um a solution, engineering their resources.
00:16:31.600 --> 00:16:39.200
So they're looking at an auction and saying, yeah, more bidders, better bidding, this is great, better use of inventory?
00:16:39.360 --> 00:16:40.080
I don't know.
00:16:40.559 --> 00:16:43.039
So it's not always more is better.
00:16:43.200 --> 00:16:48.879
We see that many publishers reduce the amount of bidders, like in pre-bidded called adopter.
00:16:49.200 --> 00:16:56.480
So they reduce and they still keep the same value or reducing the ad TXT of the allowed publisher.
00:16:56.720 --> 00:17:11.039
The publisher needs to think of the future needs that he's not seeing right now and the flexibilities because what we're seeing is that many publishers they start on the web and they see great success, but then they go into in-app.
00:17:11.440 --> 00:17:18.480
Does your vendor can keep running the same service seamlessly on in-app or when it moves to CTV?
00:17:18.640 --> 00:17:25.200
Privity is evolving, yeah, Amazon APS is evolving, and all of those markets need identity.
00:17:25.440 --> 00:17:45.839
So you you think of your identity vendor as a solution that thinks for you, the doing the work of easy of integration, whether it's web, whether even sometimes uh an aggregator expand and saying, I'm gonna get my my traffic on the server side.
00:17:45.920 --> 00:17:54.880
So does your provider, does your identity partner can help you server to server without being on the page, code on page or on the SDK?
00:17:55.039 --> 00:17:57.599
Those are the questions that you need to think.
00:17:57.839 --> 00:18:05.200
Looking for the long run, I'm now on I have my content, it's on the web, but I'm gonna expand, I'm gonna be in ENAP, I'm gonna be on CTV.
00:18:05.359 --> 00:18:08.799
I want my identity vendor to be as flexible as myself.
00:18:09.039 --> 00:18:14.880
What are you evolving the company in the direction of to be able to do exactly that, to be that flexible?
00:18:15.200 --> 00:18:23.920
We see a lot of trends on adopting AI solution as part of your ability to move faster.
00:18:24.319 --> 00:18:30.240
Even a simple task as translating the site to a taxonomy with LLM becomes more accurate.
00:18:31.039 --> 00:18:36.000
We see ourselves more going into the in-app world.
00:18:36.480 --> 00:18:43.039
So far, there were solution on server-to-server API, which is easy, but we see the needs for SDK.
00:18:43.359 --> 00:18:46.799
And there are multiple
mediation platforms that need the solution.
00:18:47.440 --> 00:18:56.960
And also when looking about mobile, because the AI is hurting the web publisher and it's not affecting the in-app.
00:18:57.200 --> 00:19:08.240
So, what we're seeing is that there is a lot of demand for uh identity and suite of solution like onboarding and activation and identity, and also CTV.
00:19:08.319 --> 00:19:26.880
So we see ourselves as expanding our buy-side solution, yeah, but still maintaining a very um hard KPIs on accuracy when it comes to publisher on flexibility, judging you by performance and revenue.
00:19:27.440 --> 00:19:30.559
They have a lot of pressure with AI to bring the best.
00:19:30.720 --> 00:19:36.720
So if you need to do the best, you need to be equally investing on all channels.
00:19:36.960 --> 00:19:43.519
And we see ourselves strongly on CTV, strongly on mobile, and expanding to GEOS.
00:19:44.079 --> 00:19:46.960
The idea is scaling with intelligence.
00:19:47.200 --> 00:19:53.200
One part of that is AI, which we addressed, but it's also scaling overall to meet the needs of the market.
00:19:53.759 --> 00:19:58.960
We understood that Germany is uh a very important market.
00:20:00.079 --> 00:20:08.960
Even though the publishers are more conservative, there is a demand for identity and there is a demand for understanding how to overcome idealists.
00:20:09.119 --> 00:20:17.119
We even hired uh um a general manager and open a branch, local branch here in Germany, the same as in France.
00:20:17.200 --> 00:20:36.799
Yeah, because we believe that you have to be in the market in order to influence, and that's why we're coming here to spread the news, to share our success stories within Germany, because we believe the German market is a few years behind with adopting technology, but it's getting there.
00:20:37.200 --> 00:20:40.559
The publishers who've been on my show today telling me about that.
00:20:40.640 --> 00:20:44.400
But I'm interested what questions they're asking in Chen IQ.
00:20:45.200 --> 00:20:53.759
So they're they're asking how to maintain privacy without hurting their performance.
00:20:53.920 --> 00:21:01.359
Uh, there is a process of reconsent where anytime you add another vendor to your CMP, you have to do a reconsent.
00:21:01.519 --> 00:21:01.680
Yeah.
00:21:01.920 --> 00:21:03.839
And that's one challenge that they ask.
00:21:04.079 --> 00:21:09.039
The other challenge is what are the demand sides that are working for you?
00:21:09.279 --> 00:21:12.000
Okay, the Europe, the the US market is huge.
00:21:12.160 --> 00:21:14.160
Are they the same in in Germany?
00:21:14.319 --> 00:21:29.119
But we come to realize that we need to penetrate the local SSP, and that's what we're doing with local SSP, because many publishers here in Germany have a split of maybe 50%, which are there, those buyers do not exist in North America.
00:21:29.440 --> 00:21:31.839
CTV is always intriguing for me.
00:21:32.000 --> 00:21:33.039
What does that look like?
00:21:33.759 --> 00:21:39.440
So CTV is challenging because this market is yet late to the game of identity.
00:21:40.240 --> 00:21:46.960
But there is a need to understand exactly what does it mean a household share device, a Wall Mountain device.
00:21:47.119 --> 00:21:50.480
They're holding their phone while they're watching and they're installing stuff.
00:21:50.640 --> 00:22:16.160
So for us, is working with uh um attribution companies, providing them identity solutions in the form Of a device draft that has also the ability to connect Ruco device, Samsung device, all those uh channels that have their own platform, their own technology, but the user is expect exp exposure more to CTV.
00:22:16.480 --> 00:22:18.799
Now it's kind of a house level match.
00:22:19.119 --> 00:22:28.640
But we see a trend where CTV is moving also to um devices like iPad and phones where people take consume their content there.
00:22:28.880 --> 00:22:29.279
Yes.
00:22:29.519 --> 00:22:38.160
But we're trying also to accommodate the need for those platforms, and it's more about I think attribution on a house level.
00:22:38.880 --> 00:22:48.400
And that's where we found ourselves uh trying to understand how can we educate the buyers on CTV to ingest the signals.
00:22:48.720 --> 00:22:51.680
Yeah, let's just let's just talk about it plainly.
00:22:51.839 --> 00:22:55.359
The change and the challenge that AI presents.
00:22:55.680 --> 00:23:06.720
Because when you're thinking about it from a public perspective, your customer could be your biggest customer, maybe AI bots.
00:23:08.000 --> 00:23:22.000
Yes, so luckily when we analyze the signals and deciding whether the criteria to form a household and then a person has multi-dimension criteria.
00:23:22.240 --> 00:23:35.839
So obviously, a bot is being detected, not always, but it's challenged whether it's someone scamming the site, and now you are trying to identify a bot that is doing something for the person.
00:23:36.240 --> 00:23:43.440
I think still we see that publisher getting hit and the the amount of page views are shrinking.
00:23:43.839 --> 00:23:55.039
So it's even more challenging to yield better performance on the on the share that's still left for the real person that's visiting the site.
00:23:55.200 --> 00:24:06.160
We identify the user agent when it's coming on a bot, we identify their IP address, we identify a lot of signals when it's not a person, but it's not completely sealed.
00:24:06.640 --> 00:24:12.559
So to be honest, it's like an ongoing battle once AI take over more and more.
00:24:12.799 --> 00:24:26.880
But then you ask yourself, how about providing for those giant also identity to say, okay, you know that person, but I can know other stuff that the user actually still surfers, still going in the website.
00:24:26.960 --> 00:24:33.759
So we think that our next job would be getting identity to those big AI chats.
00:24:34.079 --> 00:24:42.880
We see what the user are
doing on multiple sites, multiple vertical, and I think we could also add supplemental information even to AI.
00:24:43.200 --> 00:24:50.240
And to your point, if you think about it, you know, the value of the gold now, the human gold, will be much more valuable.
00:24:50.400 --> 00:25:04.559
And probably I wouldn't say the pressure, but the opportunity for buyers as well as publishers to get that right, to understand the difference between matching and matching accurately.
00:25:06.000 --> 00:25:10.799
So matching and matching accurately, I think it raised the stakes, I think.
00:25:11.200 --> 00:25:18.319
You have an auction right now and you need to provide something rather than checking what is the accuracy of what you've provided.
00:25:18.559 --> 00:25:42.240
And we we said that at the early stage, publisher has a lot of pressure to provide value because if they're shrinking and they still need to show growth, I'm not sure yet that all of the publisher are aware of the accuracy issue, but once chargeback is coming, it's going all the way from the top, from the advertiser to the DSP, SSP, and back to the publisher.
00:25:42.559 --> 00:25:42.880
Okay.
00:25:43.039 --> 00:25:48.319
An identity vendor tells me they have enormous scale.
00:25:48.640 --> 00:25:57.519
What's the first question that I need to ask in order to determine if that scale, if that more is better?
00:25:58.640 --> 00:26:08.079
Evaluate and measure your own first party data against what the identity vendor is offering, because then you'll gain control versus what they're telling you.
00:26:08.319 --> 00:26:10.799
That's testing that testing that 5%.
00:26:11.359 --> 00:26:25.279
Yes, take your 5% of your first party data that you know their email, you know their identity, they already visited, blend them without telling them, and send it over and see what the results or best 5%.
00:26:26.240 --> 00:26:28.400
But what's really the number?
00:26:28.559 --> 00:26:34.160
What's really the cost of inaccurate identity?
00:26:34.640 --> 00:26:46.880
So hurting the experience for the long run, being exposed for chargeback, the ability to stand for your content and the audience that you are selling.
00:26:46.960 --> 00:26:48.960
Because at the end of the day, you are giving it free.
00:26:49.359 --> 00:26:55.599
Understand what is the accuracy, understand what are the sources of this data that is being fed into your streams.
00:26:56.079 --> 00:26:56.319
Okay.
00:26:56.480 --> 00:26:58.000
Well, it all comes down to trust.
00:26:58.160 --> 00:26:58.880
We talked about it.
00:26:59.039 --> 00:27:00.000
Measurement, trust.
00:27:00.160 --> 00:27:01.279
We talked about the metrics.
00:27:01.359 --> 00:27:14.640
So in identity, trust has to be what trust is something you earn by showing a real value for long-term relationships.
00:27:14.799 --> 00:27:21.279
You're making sure to stay high on the numbers because at the end of the day, it's an it's a it's a number of profits for the publisher.
00:27:21.599 --> 00:27:26.319
But adding a solution for advertiser and buy side is kind of creating a holistic.
00:27:27.039 --> 00:27:44.000
So keep on the guard for the changes, but don't lose accuracy, don't lose reliability, because you need to earn the trust that you're not bluffing, you stand for what you're providing, you're not rounding corner.
00:27:44.799 --> 00:27:53.680
If you want to keep for the long run, keep straight, keep technology, and keep research always on top.
00:27:54.319 --> 00:27:56.160
And that's for any technology vendor.
00:27:56.319 --> 00:27:56.960
Keep research.
00:27:57.200 --> 00:27:57.599
I love it.
00:27:57.759 --> 00:27:58.880
But it was a great interview.
00:27:59.119 --> 00:28:01.279
May I thank you for it, thank you so much.
00:28:01.519 --> 00:28:02.319
Thank you, Peggy.
00:28:02.400 --> 00:28:03.519
It was great.