Episode 37

How To Forge A Career In AdTech (Track + Store/(Steal?) User Data) - w/ Vishwas

Dec 11, 202400:58:59Video episode
How To Forge A Career In AdTech (Track + Store/(Steal?) User Data) - w/ Vishwas thumbnail

Vishwas is the head of Revenue Strategy at subject of today’s discussion is AdTech. Vishwas takes us thru the fascinating and cash-strapped world of ad tech and the engineering marvels it takes to serve personalized ads to users within 200 ms.

Who this is for

  • You are changing lanes and need the version that still makes sense when the story is not neat yet.
  • You would rather hear Vishwas's version while the mess is still fresh than get another polished hindsight sermon.

Key takeaways

  • Forge A Career In AdTech (Track + Store/(Steal?) User Data) - w/ Vishwas

Transcript

The full conversation, right here. Auto-captions, lightly cleaned, still very much a real human conversation.

Open source video
11,171 transcript words81 transcript blocks
00:00:01

am I right in assuming that we should never allow cookies for any website unless well inob hard-hitting point was the kind of Revenue that we were getting the money involved in attech is something that blew me away what are some skills that would make them well suited for the attech industry while talking about it I took a couple of minutes to explain it but all this opens in 100 millisecond and that's the old play do you have access to my Google searches and such Nike can come and say that I just want to Target a user who plays basketball once or twice a week a lot of apps when you install them they want access to photos when you select all photos because this app now has access to my entire gallery IEX being a Amazon product I not do anything with that data even if they're actively

00:00:41

that data even if they're actively listening maybe I watched a series about a rock band so would I start seeing ads of guars then welcome to The Ready Set do podcast where we learn from journeys of not experts who are just two steps ahead of us I'm Naman Pand and in this episode featured not expert is V shanas vishwas is the head of Revenue strategy at ver which is one of India's fastest growing short video apps and is a product offering by one of the most prominent advertisement technology companies based out of India our discussion today is centered around attech vishwas takes us through the fascinating and apparently cash stra world of advertisement technology and the engineering Marvels that it takes to serve personalized ads to users in a time frame of about 200

00:01:24

to users in a time frame of about 200 milliseconds Yes you heard that right 200 milliseconds is apparently the threshold time around which if an an ad is not served to a user the entire service is in Jeopardy we then also cover some of the innocent and otherwise things that companies do when it comes to tracking and storing your data exactly how cookies and tracking pixels work and a ton of best practices when it comes to what you and I can do to secure our data on the internet while also not jeopardizing our browsing or searching experience yes as it turns out much to my surprise those two things can actually be really deeply intertwin I should note that after our chat vishwas was kind enough to share all of the ways

00:02:00

was kind enough to share all of the ways in which riverside. FM which is what I used to record these podcasts was trying to store his data and the results are concerning to say the least in keeping with our theme of learning from somebody that's just two steps ahead of us instead of an expert my goal with this episode is to Spotlight the fascinating world of adtech and the opportunistic ways in which vishwas has curated such an amazing career in this field and how you can too this is The Ready Set to podcast and to support it subscribe to me on YouTube and leave me up to a Fest star rating on Spotify or your favorite podcast app there are also links in the description for you to check out more direct ways of supporting me and now

00:02:32

direct ways of supporting me and now without any further Ado my friends here's vishwas vishwas welcome who having me really excited to get into you know exploring the fascinating world of adtech and all that it brings um just to yeah you know set some context for our listeners um obviously this is a field that we interact with almost on a daily basis all of us are served ads and such everywhere you go you pretty much kind of are stuck with them almost but really what goes into the background of how they're served who works on them how they select who you who they serve it to etc etc so yeah a lot of these things are just black boxes at least to me and

00:03:10

are just black boxes at least to me and I'm hoping to you know demystify some of these um things that that occur in this particular landscape so before we jump into the deep end here um vishwas do you mind just walking us through a very brief snapshot of your you know career and and experience so far just to help the listeners contextualize where your information will be coming from about like you know everything that you will be sharing about this space sure and thanks again for having me you you're doing doing a great work with the podcast and I'm happy to be here so if about me is yeah a brief brief about me is I I did my engineering from PS University in Bangalore uh engineering like always everyone does

00:03:53

engineering like always everyone does engineering I had no clue what to do so I Jo engineering based on my up EC couldn't get into CSO started with EC uh was pretty decent at it my grades were great and everything but maybe in the third year is when I understood that even though I like it I might do it for like 3 to four years this is surely not something I would like to do uh for the rest of my life and in fourth uh during our campus placements I got placed at Bosch uh as a software engineer but I was always eagerly looking out for business roles as well and that is when inmobi came came to our campus for just

00:04:31

inmobi came came to our campus for just um internship opportunity during our e8th semester it was a six Monon long internship and uh I was like I really need this went for it got the internship and during the internship this how I started was introduced to arttech as a whole right I had no clue about how advertising used to work was there really a technology behind it or was it just random I just knew that Google was doing it really well Facebook was doing it I didn't really know about the other other companies that do it so immobi I would say was the introduction for me of attech and then I slowly started working at inmobi across Fields uh inmobi is a Indian Indian it's actually the first Unicon of India and it's just that be one of the top I yeah it's it's a pretty old company I would say it

00:05:25

pretty old company I would say it started in the 2007 got the Unicorn status in 2012 itself and then really focused on uh the future of advertising and back in 2013 itself they started working on mobile they knew that in the long term people are just going to be using mobile and that's where the buying is going to happen so they were one of the leaders in this so I got to learn quite a bit about actic at inmobi and yeah I was there for around six years and then just this year I moved to a company called ver Innovation which is another Indian company uh it's the a parent company of multiple Publishers in the attic space but yeah now um I was hired specifically to build an ad

00:06:08

hired specifically to build an ad exchange or like a something we can get into later but um ad exchange of its own and yeah that is where I am right now makes sense that's you know really appreciate that walk through and all that you've seen kind of and definitely can relate to that aspect where you know obviously you get into engineering only to kind of realize that this is really not something you see yourself doing especially when it comes to electronics I won't even pretend to understand all that goes into that but all I know that yeah it's it's pretty as far as engineering Fields go it can be really you know Niche when it comes to that so I can totally relate with your um you

00:06:47

I can totally relate with your um you know ability or your want really to um go away from that and go get into business so I guess my question to you from at this point is um you referenced that when you had first started in movie were as you said you didn't know anything about ads and such other than that you know Google does it and meta does it so that's really funny because has exactly where I'm at like I'll I'll give you just a little bit of more context um one of my friends who is also going to be a guest on this podcast actually eventually um he does um digital marketing for companies based out of India so he has a big presence on as really just meta is what I think is the biggest uh or at least the most

00:07:30

the biggest uh or at least the most common thing that businesses like to put ads on so I sat with him for about 3 hours and he showed me the entire back end of how meta ads work so you I learned that day that you get to pick out um stuff like even what words have appeared on the user screen you can then use that to Target certain ads to the user obviously the demographic stuff you can pinpoint which geography which location all of that so the reason why I'm sharing all this vishwas ISS I have some background and for our listeners we'll just assume that you know we haven't been living under a rock or for all of our lives but really trying to understand in that first year at inovi what were some of the biggest learnings especially as a beginner in this space

00:08:16

especially as a beginner in this space that you know struck out to you the most or that felt the most notable because I'm sure there must have been a lot of them but trying to get that you know just the most heavy hitting items there so when I jammed in the I basically didn't even know the difference between marketing and advertising and I was always like it's it's almost the same thing right and that's I don't think why I know at either can you can you share what the difference is that's such a good point yeah you know very tap level right marketing is doing promotions across all channels organic inorganic building uh building your own uh content posting it on platforms and everything

00:08:55

posting it on platforms and everything advertising specifically is doing paid promotion and that's the main difference advertising is where you can focus on inorganic growth you can focus on user acquisition from a paid Channel perspective and that's the very top level that's the main difference but the lines are blurring right and as we evolve it's mtech and even attech are is it's becoming in like a same thing but yeah that's the top level difference and attech to explain to you what attech is it's it's not creating ads as such what is the technology be find delivering the right ad to the right person uh at the right place and at the right time so that's what I pi as basically on as as the advertising evolv uh from traditional channels like

00:09:44

evolv uh from traditional channels like newspaper brochers and everything to the internet boom in the late 90s and the uh and everything coming up Edward housing really wanted to come in that space itself and specifically in the online channels how you do paid promotions and the technology behind this is what attech attech really is and um inmobi specifically the hard-hitting point was the kind of Revenue and the kind of clients that we were getting um it was not limited to just internet companies or one category of advertisers spending on it but literally all the top uh advertisers right really wanted to focus on online channels we really wanted to focus on mobile advertising which which was a surprise the money involved in actic is something that blew me away uh and they never thought this would be such a big industry and uh I

00:10:36

would be such a big industry and uh I always knew Google as a yes they have like a search engine they do a bunch of things but I really didn't know how they make money and when you look at Google's actually they add most it's all hard since it's all uh their SDK that they have on apps and everything so uh that's when I slowly started realizing that this can actually be an industry where I can see growth like Advertiser is never going to go away with internet zero free and on that so that all makes sense yeah real quick question on that what is the way in which uh in Mobi you know serves ads so is that just directly through your phone are you just straight up getting notifications from the inmobi uh like app itself or do they work with companies to send

00:11:24

work with companies to send notifications like what is the modus operandi I guess got it so so like you said right Google ads and meta ads even when Amazon ads control 70 to 80% of the entire ad business the remaining 20% is where independent organizations have devop the same infrastructure same tools but there is not uh there is not like a black box that's more transparent in what's going on and it's more usable with uh multiple partners there so inoi as a whole has its own platform has its own ad server that they have built uh they are in the ecosystem they have all the main three components the supply side platform the ad exchange as well as the demand side platform it's

00:12:09

demand side platform it's so yeah the at inob they use their own proprietary platforms and they connect with Publishers um Publishers developers basically uh they add their own inmobi SDK into the application um and in real time whenever a user opens an sees sees a ad slot in real time I would say less than 50 millisecond is where our ad server gets a p that this is the person uh this is where the ad slot is available this is the supply side on the demand side similarly we are connected with thousands of advertisers with other attech companies as well uh it's called the demand side wherein um they're connected to it and we send those things ad Ser add notification pings to all these ad

00:13:02

notification pings to all these ad server understand if there's any Advertiser in the ecosystem that wants to deliver an ad on this receive an ad send it back to the user and render or deliver the ad on the mobile platform so even though it's a very while talking about it I think I took a couple of minutes to explain it but all this happens in 100 millisecond and that's the old play that's happening right now that's very cool because I realized that if that the delay is more than that then the user is not even there anymore and your whole thing goes and bust so for for it to work it needs to be near seamless so just to make sure I'm

00:13:41

seamless so just to make sure I'm tracking here completely exactly what's going on cuz this is all very new to me I'm I'll try and say back what you just said and you can like you know correct me if I go wrong somewhere so I have an app right I literally let's just say it's like a step calculator app doesn't even matter um in that app I have the ability to install we'll just stick with Android for now or maybe we don't have to but yeah I'll just install the inmobi SDK to that to my app and what that does is there's just a slot there on my app where when a user goes and then

00:14:12

where when a user goes and then basically there's like a space over there and then you all your platform decides what you should serve to this user based on whatever Partners you have based on whoever wants to advertise and then you pick whatever is most suited and you serve it immediately within light. 2 seconds is that roughly what you were saying exactly yeah and in between the personalization is what um makes especially in the 2020s right now uh it's how well you can personalize the ad how efficiently you can deliver it is won't matter So within that same 100 200 millisecond that we can play with you also need to make sure that the right kind of price uh is sent across to the publisher because uh in

00:15:00

across to the publisher because uh in this ecosystem there are hundreds of partners that have come up right now and uh it's very rare that you'll see a publisher working with just one attech company a publisher will be like I have everything set up on my end why not I connect with all the 100 whoever pays me the highest for that inventory I'll showcase my ad so there's a quite a bit of automation there's quite a bit of realtime auctions that happens and uh inmobi specifically is in that space of programmatic advertising uh which is basically a real time when you say Publishers sorry when you say Publishers is that the same as um just companies or is that something different or is that just what they're called in this Cas

00:15:40

just what they're called in this Cas yeah so when I mean Publishers they are the companies that have the capability to publish content on an online platform can be a mobile device can be a desktop can be a connected TV can also be a digital billboard outside our home as well right so if a company has a online space and they want to deliver that then that's a publisher for us I see and then so naturally that brings me to the you know next really obvious question is um how do you understand or how do you get the data on the user that you then later use to serve that user ad where what is the source of that data how frequently is it collected and to what degree does the

00:16:27

collected and to what degree does the user have to consent to that uh for lack of a better word yeah consent is actually a big word that's coming up and people are starting to make take notice of the Privacy part of all this right because 10 years back everything used to flow uh you install an app you never used to care if you gave them a location permission permission they doing everything it's not even relevant to you why do need camera access we never ask questions but it's good that people are asking these questions and the government is asking these questions to companies as well and are reigning the compounds so so yeah on how we collect data as I said right uh in in this so it's going to be a ad

00:17:11

in in this so it's going to be a ad server to ad server connection and inmobi specifically connects with their SDK once it's Google has the best SDK in the market is there a few other companies like app and unity that have really good uh sdks too but using this SDK they decide what kind of data that you really need need um and this is sent through a transaction when the ad server gets a ping from the user itself so when you are going scrolling through your mobile uh you would have already granted access to a bunch of parameters plus the mobile application would have the SDK in place which would require the right parameters for you mainly like what kind of content you're currently consuming

00:17:55

of content you're currently consuming where are you interacting with the content uh location is one common thing that's almost always present uh plus other that you have on your mobile so it can diversify and quite a bit of content can be sent across to the inmobi server and then it's based on inmobi on how well you use this data right because literally everyone has all the data that's there and data right now is available everywhere it's all on how well you can use it and how well you can monetize it and that's the play that attech players need to solve for would you say so you know say if somebody is currently a student right that maybe knows a little bit about this or is trying to learn um I guess what are some

00:18:44

trying to learn um I guess what are some skills that this person could learn that would make them well suited for specifically the adtech industry are there any I know you probably require a lot of functional skills from what I'm getting but in terms of maybe you could touch on that also but also in terms of technical skills um what are some things that students could do today given what he said about you know it's the upcoming thing it still hasn't even realized its full potential it's still growing so what is the best way to put oneself in the spot to leverage that growth when that comes yeah so at at a fundamental level it's the same kind of technology that you're working on that you're going to study it's going to be python it's

00:19:24

to study it's going to be python it's going to be Java it's going to be PHP go hang and all the backend uh development skills that you're going to work uh but the only tweak that I would recommend our student to make while they grow is to understand what are the latest trends going on uh in the acch industry um understand what kind of infrastructure is being used uh there's something called realtime bidding that I mentioned right so the protocol for Real Time Bing keeps on changing the only tweak that I would recommend the students to make is understand what are the latest trends happening in advertising uh in the advertising Technology field right and there such lens right now and I'll be honest with you compared to all the other Industries and how easily the

00:20:09

other Industries and how easily the information is available if you really don't know what you're looking for in the actic space you'll never you'll never find it and you might just go in the wrong direction so I I would recommend following the talk few newsletters subscribing to it from um that point of view understanding what the governing bodies are doing so there's something called IAB which is the advertising Bureau and galon the entire advertising uh industry so they have their own lab they it's kind of like a open uh internet company where they share everything so those small tweaks I would recommend where you can go through these what is the trending thing so right now sustainability is trending something's trending and you tweet your skills to it but yeah a level

00:20:55

tweet your skills to it but yeah a level I would not recommend that you just focus on ad Tech as a whole just focus on the fundamental skills understand what really you want to do and maybe once you're looking out if you really like an artic company we all the Artic companies must give a very solid training for all the freshers that join atob and even at real I see that a strong uh training is needed because even though you learn the open source code that's available all which companies have their own way own terminology and uh they do their things in their own way so uh I would say for a CO just focus on the fundamentals be aware of which are those top companies which is stranding and then the rest will just fall in

00:21:41

and then the rest will just fall in place makes sense and then I kind of want to uh explore a little bit about exactly how that process works where you serve a user ads right so you know we'll go back to our example with the SDK so say I open my whatever walking app you know and then you have information about me you know I you know my age you know where I live and maybe you know like you probably have access to actually do you have access to my Google searches and such um I guess what is can you tell us here's my question actually and sorry it took me a while to get here but what is

00:22:20

took me a while to get here but what is the General Baseline level of information that most users that you have for most users question like obviously there's age there's geography and what else so so when we look at data right this like two sets of data one is the deterministic data and one's the probalistic data and uh there are different ways of getting this data from from the user uh so when we come to the deterministic we exactly know that this user is in that category right now so like you said age the gender uh what kind of content they like consuming on a daily basis is uh what do they really how much tring they consume what do they like interacting with since

00:23:05

what do they like interacting with since this does not really depend from a user to user um fistic lens that's where we compare multiple and data to understand what the who the user might be so when I when I speak about the user there's nothing to be afraid of saying we know exactly what namon is doing or what namon is consuming this has been anonymous we don't know anything about who the user is but we just know a ID or a device ID of what's happening so when I say a user there's nothing like we know exactly what a person does it's purely from a lens of we just know that the device has this kind of user maybe sitting behind them we don't know the name of the user we don't know the personal details of the users we just

00:23:53

personal details of the users we just know that they come from a certain age group uh uh what kind of content they want and everything plus one of the main data fields that we currently look after is um from the transaction level right what is the buying power of the user uh what kind of um and this is the data that is sold to us by other aggregators in the industry wherein uh from a let's take a retail media from a retail point of view uh if you go to Walmart frequently have a bunch of transactions or what you buy is recorded from the system this is usually available by uh with all the attech companies uh in addition to that what kind of transactions online transactions you do uh so in India if the UPI player is

00:24:44

uh so in India if the UPI player is there phone Bay or uh no specific company from example but yeah all all the company and is usually store the data and anonymize the data and send it across to uh the atte companies and that's how we understand that this user is there and we categorize them into multiple personas uh because a very generic age level uh so you could be a 30-year-old person sitting out of the US I could be in the next room again 30-year-old sitting in the same geographical location but have interest might be different right so it's important to categorize them into personas and based on what do that really consume what kind of a uh buying transaction ability does the person have what do they really want to see and everything so at a very based

00:25:36

to see and everything so at a very based level even though these targeting criterias are captured by us then we have a DMP a data management platform which really segregates captures data over a period of time and segregates them into a kind of a Persona which is then shared with the advertising platforms and everything and they just select who they want to do so uh a Nike or a sports company can come and say that even though it's of 25 to 35 user I just want to uh Target a user who maybe plays basketball once a week or twice a week right so then a person in the same age group that does not even go near a

00:26:17

group that does not even go near a basketball code so all the personas are available with the uh advertising platforms which they just select with just the click of a button and add the revenue add the relevant creative and uh get that Persona users so this same exact concept also works when I go on just say any blog site right it's just literally just any random blog page on the internet and that that site probably has Google's AdSense SDK in their uh in the blog itself and what you're seeing in Mobi does Google also does the same exact thing like it's just a space user goes in it decides what it knows what user wants and then it just serves the user and AD based on its list of Publishers and then you're saying Google does this U meta does this Amazon does this in mov does this and there's I'm

00:27:16

this in mov does this and there's I'm sure many many other players that do but this is like the operational model for lack of a better word is that exactly yeah um from a data perspective yes they anonymize the data they understand the content that a user wants to follow or wants to interact with they segregate it understand multiple other parameters make probabilistic personas out of it and then it's shared with the advertising platforms which then just click based on their uh campaign that they're trying to run and uh deliver the ad that is only yeah yeah yeah that's actually cool only frightening thing that the industry and the Privacy comes in places U especially on Android Google Chrome and everything it's just not the website or app that you are on also the background websites the

00:28:10

on also the background websites the third party cookies and all these other companies that you have not even opened or don't know it's there as well are also tracking these data right and that they segregating it and they're selling the data to multiple uh companies and and that's the that's specifically what uh right now the concern is about privacy and apart from the W Gardens all the independent companies have taken a strong stance towards it is we respect a users's privacy and we're going to make sure that we don't take that kind of data from them so correct me if I'm wrong but did you just say when I accept cookies from a website that gives them the ability to track what I'm doing on other websites then just like all the time and then they like package that

00:28:55

time and then they like package that data and sell that is is that right so there's again first party cookies and third party cookies as well right um if you're using Safari if you're an user you you actually uh taken care of by Apple I don't know how well Apple respects privacy but it's not as bad as the Android or the Google uh version as can you can you dumb it down for me why exactly is that bad the way Android does it like sorry I'm not I mean I obviously learned about what cookies do in college but I feel like I actually have no idea right now so this is very interesting can you yeah talk me through that so so

00:29:33

can you yeah talk me through that so so let's take a website website for example right um Google um the way it works as anybody the website owner has the capability to add any kind of pixels out there and third party cookies as well are added in there you can actually just go to the website that you're currently on for the users that listening and go on inspect go across what what all track in pixels are there on the on the website that you're using there'll be T of pixels that you're not uh aware of that will be tracking all this information and let's say you use the website you go to another website in the same in in the same setting right you have not closed the window you have not anything in the same setting you go to another website then they have access for uh that transaction the entire uh

00:30:24

for uh that transaction the entire uh website details what kind of interaction you have done in and everything so T party cookies is something that tracks all this so what exactly is a tracking pixel like is that just a pixel on the website whose job it is to track what what else is going on on the website or correct correct so you can just think of it as um a JavaScript or or a bunch of fields that a company adds and tries to get the values based on the interaction of the user with the website uh so if you are on a News website Channel and you click on a specific category of news when you click on it let's say you do not give any additional data on it but uh you just serve the

00:31:15

data on it but uh you just serve the internet for a longer time on a category those kind of information is gone to the user and um yeah at a whole the kind of user the behavior that you show on the website is recorded by multiple companies so then is that am I right in assuming then that we should never allow cookies for any website unless you're using Safari or what's the uh you know what's your advice on that front so I can what I personally do is um I don't use Google Chrome at all uh unless I really have to get get like a very personalized see Google Chrome has its own advantages right it's it's the best search engine in the world let's say you type something you get the exact thing

00:32:00

type something you get the exact thing that you're looking for but there are many more alternatives to it like tapt go uh if you are not a safari user like you have D go which is like a they do not track any of the cookies that comes uh of the user so these kind of websites might not give you the best results but on the other side you are not giving them anything in return as well and you can be more safe while surfing through it so yeah uh if possible go to Alternative search engines which do not record cookie data do not really sell it and Google is very open about it this right it's it's just not there in your face but if you if somebody really wants to dig it up understand what Google does

00:32:45

to dig it up understand what Google does with the data it's all there in that privacy policy and everything it's just that the def switched on all the time and nobody really goes it switches it off and everything so uh yeah uh yeah sorry real quick question on that um about what you said about uh Chrome you when you normally go to Chrome and when you search something you probably find it I don't think I followed that part entirely what did you so do you just mean like Google search like what is Chrome doing in terms of finding things that's better than a safari not in comparison with Safari uh but from a search engine perspective Google has the best search engine to give you the right of the B on the keywords you add so from that lens is what I was speaking about I

00:33:32

that lens is what I was speaking about I Mak sense so but then you saying that if I were to just use Doug du go but then if I use Google search on dougo then I'm immune from all of this stuff or is that right or am I still in trouble uh it is right to an extent where in G comes out and says we don't take any cookie uh from you but when you are there on Google and you click on a website through it I'm not entirely sure of what all is getting tracked again through that website right so uh personally I just stay away from even if I do use Google Chrome my third party cookies are always switched off even

00:34:10

cookies are always switched off even first party cookies are switched off but websites do understand this right uh since internet is free the only way to make money for most of the internet companies is through advertising so yeah they will give you a better they'll give you a better experience in turn we're just going to take a bunch of data monetize it correctly we respect the the person behind it we're not going to add the person we're just going to add a ID towards it and monetize and show personalized ads for you but yeah all almost all especially in the US and Europe specifically is very stringent about privacy and um us only you get whenever you open a website you usually get on the bottom or on the side saying these are the cookies do you want to

00:34:56

these are the cookies do you want to accept it reject rejected right do you want to accept only the necessary ones so this has changed just in the last four to five years this was not there 10 years back right but that's because the government is looking into it and they're like uh especially after the 2016 uh election us election and how there were studies that shared that uh a user was shown the kind of content and asked to vote for a certain party and now that the government knows that a user can be influenced uh with this kind of content they are adding bunch of regulations in place and uh that's something that's going on in the industry yeah um yeah in that's yeah I feel like that is the single most insightful thing I've learned through one of the many actually uh insightful

00:35:45

one of the many actually uh insightful things I've learned through this whole exchange which is that yeah it is infringing on my privacy sometimes but on the flip side of it it is making my experience better because stuff that I'm using using or searching or seeing maybe forgetting later will not be forgotten by you know that platform that platform will remember that I looked at this even on a subconscious level I think your example with the news uh categories was a really good one actually because sometimes you'll just click on stuff without even thinking about it right but there's a reason you licked on that and I think that is really interesting aspects um aspect of this I actually yeah I'm now I'm thinking of you know more uh like what are what are some other Secrets I guess that you can share

00:36:30

other Secrets I guess that you can share about just this stuff generally that you've learned that you would think most most uh people would not be aware of when it comes to really just privacy on the internet it doesn't have to be focused at ads but privacy and ads kind of go hand inand right like there wouldn't be a debate of about privacy if ads weren't the thing so yeah what are some other you know maybe notable stuff that maybe people should know that they don't because they're not from this industry the f i' say is anx is not negative as a whole right we we are bombarded with more than tens of ads per day and it's kind of built a negative image of oh if it's an ad and then I'm not going to buy it's so since I'm from

00:37:11

not going to buy it's so since I'm from the industry I know that if the correct average is shown to the right person in the correct time right it it's it's going to save the users time if it's something they're going to buy instead of searching for it online doing the right kind of research if on your screen you are shown the right kind of content and the right ad that you want to buy then it's actually saving the users time so ads as a whole is not not really a very negative thing it's just we Us in the industry need to do it better we just need should not bombb the user with multiple ads but just show them the right ad make it in make the ad into kind of a Content why should it be different from a Content make it into an

00:37:54

different from a Content make it into an immersive ad make sure that the user interacts with the ad have some fun with the ad and then or adds value right adding Valu is also really underrated yeah correct so if you're in a gaming app uh if you play Sudoku or any of these um gaming apps which have ads right uh the ads are more from a interactive lens it's it'll also be an ad something you need to click something you need to do and then you'll get a reward saying this is a coupon or this is an ad so from the cleans this a lot of things going on in the ad space which which needs to change from a user perspective as well is how they look at ad and uh that's actually our

00:38:37

ad and uh that's actually our responsibility as an industry to go towards it uh other than that from a secret secret lens uh you yes we need to be more precautious about what kind of apps we download which websites we open really understand if the source of the website we're coming from right and something that people don't really talk about is AI is everyone speaks about AI from a very positive lens right in the a industry once AI has come into picture the fraud the Bots the malicious activities has gone up significantly and um uh that's something that's Fred is something that a user needs to be precautious about so do not install or do not view something that you're not sure about plus do not connect with even a Wi-Fi network that you're not that's not protected basically or you don't know it's coming

00:39:34

basically or you don't know it's coming from a trusted Source right because once you do that do that everything is ACC accessible for all the people who might misuse the data we will so so that's actually interesting and I know you always see this where you know you should not connect to public Wi-Fi on your work laptop and all of that and this is very interesting because I come from Tech you know I like studied basically computer science but some part of me for whatever reason is just not able to appreciate the actual threat of somebody hacking into my system just by because I'm on a public Wi-Fi is that something that can actually or does have you seen that happening um because you just referenced that so I'm kind of curious about that so from a hacking lens it's a different topic that I guess you would know better but from a data

00:40:22

you would know better but from a data lens right if you really want to protect yeah data I meant the data yeah exactly so from a du talents yes I think it's it can be easily accessed uh it might be the very basic kind of data but still you're giving it to somebody that you you never know how it's come going to come back and hit you right uh maybe your content on social media platforms is going to change because of one interaction uh that somebody else on the network also did so if it's a public network let's say 10 to 20 people are connected to it and ntech Company just that through this network these users were looking at this kind of data so they record the device IDs of those users so all the 20 users are shown a specific kind of AD so there's quite a

00:41:10

specific kind of AD so there's quite a bit of okay uh because they were changed that Wi-Fi oh sorry they were all in that Cafe together that's why they over served that wow this is the probabilistic aians targeting that we do right um will atte there's a lot of targeting that happens this deterministic scale targeting is very minimal you have to rely on probabilistic ways to increase the scale of the Persona segments that you're trying to reach out to that's that's how attech companies make money uh if you know a single person who is your 100% confident is right for the campaign then you maybe find 10 other people surrounding them across multiple parameters and might have even a 70% probability of buying the product product and that's how advertising works so uh so from that lens I would say

00:42:02

so uh so from that lens I would say that's something that um I would recommend the users to be careful about obviously you I don't want people to be paranoid that there are quite a few people who are paranoid who cover the camera uh send don't connect to anything and always there but uh something just you need to take care on a daily basis just uh so from that lens those are the things that I would recommend the people do so that opens the door then to when you are as as this company that whose whole job it is to serve ads when you doing that what percentage of that is the algorithm decides that for device ID you know 3 2x1 we will serve this ad

00:42:44

you know 3 2x1 we will serve this ad versus um you or us as humans writing certain you know parameter matching so I actually won't even know exactly how you would do that but is that does that come does that algorithm come from the user or is that kind of a set and forget type of thing where you just write the code once and then for every new device ID that comes in with a certain as you said probabilistic model about certain behaviors it automatically becomes the part of like a cluster and then that entire cluster is treated the same exact way in which they all are shown the same thing so does do you do that or does the

00:43:22

thing so does do you do that or does the algorithm do um there's a of both currently and it's SL moving towards the algorithm the AI taking care of it but I would still say we're not there yet um uh but yeah the percentage of how much the algorithm does it is significantly increasing when I compare the industry five years back to how it going now I would sayit is being done by the AI algorithms itself they really understand what kind of so also the number of apps and the data signals that we get is in millions it's in billions that's blue uh yeah so so if you look at the number of Publishers there it's I would say in 50,000 plus Publishers are connected to one ad tech company and um each

00:44:11

one ad tech company and um each publisher would be sending millions of uh iquest per day so it's billions of iquest per day and at the scale that we are in right now bit is taken care by the the AI uh algorithms that we have what only where where I've seen still a lot of manual work being done is when from an Advertiser lens they're trying to optimize they still don't know if a certain Persona is going to get them more performance or a secondary person is going to get them more performance and that is where I still see a lot of um campaign Traders available in the market who do a lot of manual optimization they go through based on their learnings understand from a different Advertiser what kind of

00:44:55

different Advertiser what kind of results they have seen and optimized for a certain Advertiser so yeah it's kind of a mix of both but in the future it was going to be completely skewed towards AI algorithms so going back to our earlier discussion around you know any students that want to get into this I don't know how many students are able to write such complex algorithms but it sounds like you don't have to be a student but any professional really that is looking at this or is you know trying to build for the future it sounds like it would be a worthwhile Endeavor to look at how to solve this problem or how to use AI to um you know do this in a way that's computationally less

00:45:33

way that's computationally less intensive that is just as you said speed is of the ultimate Essence here so it needs to be quick but it does sound like yeah with all of the new age I don't know if the new age but really just the new AI algorithms that are now becoming mainstream um it does really sound like this you like this whole industry is is going to see a lot of really rapid developments I feel I mean that's true for all at this point but it does sound like yeah even this is not too immune um I guess and on that same note how does the this whole thing work with other platforms so say um if I were to be watching uh like a binge watching

00:46:14

to be watching uh like a binge watching an attach Show on Netflix um would that then give advertisers the ability to find out what I watched on a separate platform which is Netflix and then use that data to show me stuff maybe I watch watched like you know a series about a rock band so will I would I just start seeing ads of guitars then on my search you saw how how does that part work this is something that I've heard my friends also say right is they looked at a certain kind of content they looked at a movie and they got the right kind of it depends on the platform that you're using to be very honest Netflix for example does not really sell the uh what

00:46:53

example does not really sell the uh what kind of movies you're wishing all this data to the Attic platform but let's say uh MX Player now was recently bought by Amazon uh and Amazon TV MX Player is the same and within their own ecosystem they do understand what kind of movies or what kind of series you're looking into and recommend the right kind of series to you but still forting them based on Ahad is something that most companies do not do right now uh but you might see The Coincidence of it happening if you have seen a series go on to the Chrome or gone to a website flick search for a let's say you like the clothing that I'm wearing you look the brand went and

00:47:35

wearing you look the brand went and searched for the brand then only from that in transaction or interaction you've done with the website on the brand then you'll get the relevant ads coming in so uh it's not really from the content that you have watched or what the clothing was inside the content but if you have done anything additional to it that's when you start seeing the relevant ads interesting but from what what it from what you're sharing it sounds like it's it's not crossplatform though it has to be kind of within the same entity in this case that was Amazon but Amazon is probably not getting or buying Netflix data that you are aware of obviously you know you can't know all these things but from what you yeah know from what you make sense the other

00:48:16

from what you make sense the other really usually scary thing that uhuh sorry go ahead yeah you yeah usually the top companies do not really send data outside it's a smaller company means it's the unknown sources that do and that's how they make Revenue right because they don't have their own infrastructure to run ads correct the reason Amazon does not do it is Amazon has its own DSP and they use all the data and Target ads and a close system the same goes for Google same goes for Apple but for a smaller company who does not have the infrastructure to run the own ads the only way they can make money is sending that kind of history or data points to other rtech company uh and make and and that is how the

00:49:00

uh and make and and that is how the circle you know how like a lot of small fishes combined to appear like a big F to to fight off credit this is what you're saying sounds exactly like that and on that same note of you know really just conspiracy theories but of late we've been seeing that a lot of conspiracy theories are you know just realities but this seems to be a growing and at this point I think it's just accepted that you know Google is listening to our mics at all times some people even say and I know this has been debunked but it's kind of interesting that some people are like then you say um you know what's a device I don't have Alexa I can say Alexa I didn't want to

00:49:38

Alexa I can say Alexa I didn't want to say si cuz that would just trigger stuff and break recording but if when somebody says Hey Alexa a lot of people say that how does it know that you said Alexa it's because it's listening the whole time so on that same kind of note uh even with this you know ad Tech stuff is it true that they have if they wanted to be malicious they could look at you know video they could look at people's cameras look at their listen to their audio is all that just a conspiracy theory or is that just not actually what happens of Alex listening to us yes active listening is is a part of it but

00:50:16

active listening is is a part of it but when you say hey alixa that's when it's activated and it start listening to you and again we need to there a kind of bit of test that you have with Alexa and it's being a Amazon products so you kind of have a trust that yes it's coming from Amazon they might not do anything with that data even if I don't drust I don't know if many people have that yeah yeah if I left it to Amazon or some a company they actually use all kind of data but right it's under the regulations of a government and uh me being from the industry I would trust those government regulations right yes makes sense yes surely within the government there's a bunch of things that might happen which we will never

00:50:58

that might happen which we will never know about but um yeah from a top company perspective surely it's not the case nobody is listening to the interactions or videos the surrounding videos to understand and show ads to you but there were a couple of studies recently released where a few smaller media agencies did come out and say that yes they were actively listening and they did Place ads uh based on it right it's it's kind of being from the I was a bit taken a back but um yeah there were a couple of two listening stories that that come in play and yes the agency did say that we didn't know the user we didn't use use it in a wrong way and

00:51:41

didn't use use it in a wrong way and everything but it just opens up many doors um so hence like when you install an app this really happens from a app perspective right once you install an app uh you might have just installed some gaming app to play for 5 minutes a year back and the app is still on your phone and you would have given microphone access video access everything to it just as a precaution keep the app on your phone clean just use the ones that you really use on a daily basis and even if it's um you've never opened up in the recent few days it's always active on the background right you never know what kind of data it's collecting so i s to delete it if

00:52:23

it's collecting so i s to delete it if you don't want to delete it at least do not give it permissions don't give any kind of uh device permissions to it and you can just give it time of use usage right at least on Pixel my phone you have like always on only this time or never allow at least you can do this only this time then always allow so two things on off of that so the first just being that I don't know if this has been your experience but it has definitely happened with me that when I'll be talking on the phone with my friend about like a product or something next thing I go when I go on Gmail I see the

00:53:02

thing I go when I go on Gmail I see the top Banner is an ad of that thing is that just coincidence then or how does like yeah is that I just coincidence how does that happen because it happens a lot to me and a lot of other people I know yeah I would say it's most of a mostly a coincidence unless you have actually researched quite a bit about the product and then you call up someone and say yes this is the product but it's also when you speak about it you notice say more right uh if it's not something relevant to you and you're not really looking at it you might not even notice it so I would say it's a coincidence it

00:53:36

it so I would say it's a coincidence it does not happen much uh in the industry at least the companies have worked for really value the government regulations and uh privacy is taken into very high standards right now okay perfect that is that puts my heart at ease last question before I let you go here is that um a lot of apps when you just install them they want access to photos and you normally have an option of all photos or you have an option of Select what ones you want to share at least on an iPhone that's that works like that I'm sure it's the same on Android also when you select all photos does that mean that this app now has access to my entire gallery and every picture in so remember four years back or so there was some Chinese app that

00:54:21

or so there was some Chinese app that went famous that used to make people look old uh that was the I I got the name of the app per say but uh yes it does it does happen where the access can be given into the wrong hands so unless why would the audio phone need access to photos you need to ask few questions while when actually like uh really starts asking you these details right and I would say just deny everything unless it's like a mandatory thing and you're not able to access a feature in the app just keep on denying everything and that's the best way to be but there have been instances there have been leaks that been leaks there have been leaks that have come out in the news so imagine the

00:55:06

have come out in the news so imagine the leaks that have not even come out yet there quite exactly yeah quite a bit happening right now yeah but yeah fun access so contacts access is another thing that we need to be very careful about it's not just data about you once you click on contact access a user has details of all your contacts they can reach out to them they can misuse misrepresent you quite a bit can be done so one thing that I would like users to take away from this podcast is really give attention to what they are installing read through those policies uh even if it's to our a top lens and understand because because all the users are basically spending time on their phone right now more more than interacting with friends we are learning

00:55:55

interacting with friends we are learning from our phones we learning from website so it's very important on what we consume on a daily basis we live on our phones at this point yeah exactly so so take that extra couple of minutes go through the source if it's a very famous company it's not like I we are safe but still they have a few regulations uh in place so more interesting than unknown source uh at least in India I know quite a few Chinese apps were were BL listed right we don't have TI talk we don't have like B 50 50 plus Chinese apps and it was it was done due to this main reason uh and at the top level there

00:56:36

reason uh and at the top level there might be hundreds of other things going on but from our dat tens it's best if we if we just go through these details that makes sense which was thank you so much for sharing uh all of the best practices really that you've shared I feel like obviously you know I've been on the internet for a really long time and yet somehow I feel like I was doing most of these Basics wrong which aren't very hard to get right you just need somebody to kind of walk you through what it means like you know for the for example the stuff about cookies like I click that button on multiple websites multiple times every single day right without entirely understanding exactly what's going on with that maybe that's

00:57:16

what's going on with that maybe that's on me kind granted I you know I get that but it it just helps sometimes to really think deeply about some things and because you know that just informs you and more importantly it keeps you safe like you don't have to worry about somebody again me personally my stance is that sure like take my data like I don't actually care but I can very easily understand how for someone else that can be really you know an invasion of their privacy they're like no please don't I don't consent so please don't do that yeah so yeah really really appreciate you taking the time today and I feel like I'm going to go away from this Having learned so so much and yeah really excited stay connected with you

00:57:56

really excited stay connected with you and to watch all the amazing things that I'm sure you're going to do with your career in this industry which is really just feels like just starting to take off so really exciting kind of the head thanks a lot T and and you've been doing great stuff with the podcast and I look forward to your videos every week just to understand which new thing you're going to talk about so great work here great production work uh and and yeah thanks for having me again absolutely thanks so much that brings us to the end end of episode 37 of the ready said do podcast thank you all for sharing these episodes with those that continue to benefit from them if you would like to

00:58:33

benefit from them if you would like to support me the easiest way to do that is by subscribing to me on YouTube leaving me up to a FIV star rating on Spotify or any podcast app of your choice and telling literally all of your friends about what an amazing podcast this is catch you all in the next one new episodes every Wednesday

Transcript-backed moments

A few lines worth stealing before you hand over the full hour.

Open on YouTube
00:00:01

am I right in assuming that we should never allow cookies for any website unless well inob hard-hitting point was the kind of Revenue that we were getting the money involved in attech is

00:00:10

the money involved in attech is something that blew me away what are some skills that would make them well suited for the attech industry while talking about it I took a couple of minutes to explain it but all this opens

00:00:19

minutes to explain it but all this opens in 100 millisecond and that's the old play do you have access to my Google searches and such Nike can come and say that I just want to Target a user who

00:00:27

that I just want to Target a user who plays basketball once or twice a week a lot of apps when you install them they want access to photos when you select all photos because this app now has

00:00:35

all photos because this app now has access to my entire gallery IEX being a Amazon product I not do anything with that data even if they're actively listening maybe I watched a series about

Show notes

Vishwas is the head of Revenue Strategy at subject of today’s discussion is AdTech. Vishwas takes us thru the fascinating and cash-strapped world of ad tech and the engineering marvels it takes to serve personalized ads to users within 200 ms. We also go over some innocent and otherwise things companies do to track and store your data, exactly how cookies and tracking pixels work and why it can be risky to allow third party cookies, and a lot of best practices on how we can best secure our data online while also not degrading our search experience.

More in AI + Tech Careers

Same mess. Different guest. Pick the next conversation that feels closest to your real life.