now so we have our speaker Ramia as well with us so the today's topic is how data science is revolutionizing attech and the advertising industry so about ramama a quick intro about ramama so with over 20 years of experience in the industry she is based in Banglore and is a highly respected leader with a proven track of success sco Ramia is known for her strategic thinking and ability to lead change and has a reputation for putting people first by leveraging cage technology so thank you everyone for joining in Ram over to you you can take the session right now thank you thanks ad good afternoon everyone um it's a pleasure to be here today to talk about the topic that is transforming the world of advertising uh the role of data scientists in attech companies and how data scientists bring a fresh perspective to this uh fast evolving industry which is very Dynamic uh if you look at advertising industry it has gone through a lot of dramatic changes over the recent years uh and some of these advancements are driven in technology and in data and analytic space uh especially this era of datadriven decision making data scientists emerge as some of as one of the key players and key skill set that offers new insights strategies and Innovation that is shaping the future of programmatic now uh over the next few minutes let's discuss how data scientists bring a unique perspective to attech companies and why their contributions are so crucial for the success of advertising industry uh if we can go to the next slide [Music] [Music] what [Music] he [Music] awesome so that was a little sneak peek into what miq is all about um yeah if you can click on uh click on all right so so uh within miq uh if you look at Center of Excellence that is based on of Bangalore here we call ourselves a center of excellence um and Center of Excellence was set up as the technology and analytics Hub uh servicing and Catering to all of our agencies and Advertiser client needs from Bangalore um and uh we play in the field of manage services media partner in the programmatic domain where we are powering some of the global Solutions and demand in the programmatic industry and for Center of Excellence to be more future proofing its capabilities uh we look at continuous evolution of by bringing in Innovation advanced technology and state-of-the-art and best practices that drive not just Innovation and the art core but also look at how can we efficiently and move for move forward with Pace but also bring excellence in the outcomes that we work on so uh with that moving on to the next slide let's talk about programmatic industry right uh next slide um so we have uh we will show you the distinction between uh marketing digital marketing and advertising so advertising you will have to look at it as a subset of marketing that specifically involves creating and delivering promotional messages to a targeted audience the primary goal of advertising is for us to mainly communicate specific message or probably an offer to a wider audience and the intention of uh with the intention of driving immediate action such as sales or lead generation right that's pretty much what your advertising is now on the other end digital marketing is a broader term that encompasses all the marketing efforts that uses digital channels to connect with our current and prospective customers now this not just includes advertising but it also brings in strategies like content marketing uh email marketing social media management search engine optimization and many more now if you look at the goal of digital marketing uh it is primarily focused to build and maintain an online presence where we engage with customers where we nurture leads you drive long-term business growth but it's also about creating a cohesive strategy that integrates various digital channels to bring in the reach but at the same time influence customers throughout their buying Journey uh next slide now that we know about programmatic and digital uh marketing let's talk about where programmatic comes into picture right now programmatic advertising is a subset of digital marketing which primarily focuses on automated buying and selling of online uh ad space what we do in programmatic is we use a lot of algorithms and realtime bidding to Target specific audiences and we make it more highly personalizable uh personalized ads based on our data and analytics and data science capabilities now programmatic advertising generally involves various uh you know types of display ads it could be video ads it could also will be social media ads or even connected tv ads which we call as CTV ads now it operates across a wide range of platforms and websites but within the framework of automated ad exchanges and networks and the primary goal of uh uh goal of programmatic advertising is to efficiently Target the Right audience at the right time with the right message where you are able to maximize the relevance and impact of ads but at the same time optimize the ad spend now this is a much more cost efficient realtime optimized and dynamically personalized way of targeting customers uh moving on to the next slide and we all all of us know that we are pretty much in uncertain times especially after post pandemic the world is settling into new habits uh and there are you know Global political issues that are also changing Supply chains uh there is inflation labor shortages and all of this right so all of this is leading to big micro shifts now what happens when there is Big micros shifts is you will see a change in the spending pattern where your spending pattern becomes more careful and we all know it as consumers ourselves now what is happening to attention attention is fragmenting our lives today is spent on media all day but the number of devices and platforms that we use to consume media is also growing at the same time data usage is under more scrutiny and customers uh now expect that their data has to be handled more carefully than ever but let's unpack some of these uh in a uh in upcoming slides moving on um let's take TV as an example right with all the display ads social media ads and tv ads that I spoke about let's talk only about tv ads for as an example Now TV if you look at is changing that brings about you know a lot of opportunities for programmatic industry and let's just talk about fragmentation uh one is it takes a lot of computing power to analyze trillions of connected TV signals the strength of our industry to make some of the Bold choices and wisdom we have gained over the last few years points at a very changing landscape now fundamentally fragmentation exists because our viewing habits as customers have changed on an average based on some of the research data each household has about three streaming platforms and within that there are about 10% of viewers are subscription Hoppers which means they're moving from one subscription to another subscription and what really stands out to us is 93% of the consumers are also re-evaluating their subscript itions which shows that customers today are comfortable changing where they get the content from as long as they're getting the content that is needed now consumers are you know longer you know Shuckle to the TV schedule there's uh you know anecdote about watch my habits the conversation has changed but now it's now about how I behave it's about how you audiences behave as well so in the Modern Advertising landscape one-size fits-all approaches have never worked and if you look at consumers expect a personalized experience and advertisers need to continuously deliver the right message to the Right audience at the right moment now what happens when such expectations comes in is where data scientists truly come in and deliver those segment audiences based on the users Behavior demographics purchasing history and even psychographics now by analyzing this data data scientists can help create highly personalized advertising campaigns that not just resonate with individual consumers but also improve our conversion rates and enhance the user experience as a whole now let's let's think of a personalized recommendation on platforms like Amazon right or a tailored video suggestions on YouTube now these recommendations are not just random guesses but they are a result of powerful algorithms and data science models that work in the back end continuously to learn and adapt to user Behavior now data scientists make this level of personalization possible by helping connect Brands connect with consumers in a meaningful ways moving to the next slide all right so talking about advertising advertising today is not just about Intuition or gut feelings it's about leveraging data now that is Paramount attech companies today collect enormous amounts of data whether it's Impressions clicks uh user user engagement metrics or behavioral patterns the volume of data is growing and you know compounding every day now the data scientists play a very pivotal role here by using advanced analytics machine learning models and algorithms to transform the raw data that we collect into actionable insites now they also bring in structured data uh sorry they also bring in structure to the unstructured data and extract patterns which are more meaningful from all the noise and chaos but also helping businesses make informed decisions for example in programmatic Industry data scientists can predict which B placements will result in the highest engagement of the user thus optimizing campaigns for performance and cost efficiency this also gives us the ability to predict user behaviors and Target the right audiences at the right time which is a GameChanger for programmatic industry moving on yeah while the next slide com comes in um you know we also talked about outcomes which is most important for us to understand consumer Behavior right the way miq is looked at data scientist opportunity is it's our endtoend Performance Engine that makes all of this possible to achieve better outcomes be it commercial outcomes like CPA CTR or brand outcomes the as the metrics or there are also custom business metrics that brand actually cares about now the first part of our engine which which is connect allows to allows us to connect about 170 different data Partnerships first to augment the user identity then we enrich the audience digital profile and then we help build a holistic view of the customer now what also happens is our Tech platform which is Studio allows us to do this at a larger scale when you talk about the Discover over here the part focuses on how you are building these audience insights and surfacing them for Discovery via a rich data Vis visualization in our proprietary platform Hub now this helps our Traders and account Executives and our clients to build their plan and drive day Zero insights to set up campaign for audio relevancy from day one now finally what happens is our engine allows us to measure and report on all the activated campaigns to all our kpis and not just DSP kpis and we can build a custom multi- kpi Focus solution based on the user specific goal so we are able to balance the viewability with CPAs or the sustainability with CPM or something even more custom based on the customer requirement and data science here is deeply embedded in every component of our engine uh moving on all right so where are we focused on is we uh we pretty much have been having a very strong here strategically and now for us as an organization our period of focus is about building unique capabilities and continuing to stay focused on that moving on let's talk about the power of data scientist right I've tried to summarize in four different points now where where can data scientists truly be adding value first one if you look at Advanced productive analytics all of us have heard about this but if you look at analyzing historical data and Trends data scientists can predict future consumer Behavior you can also predict the market trends and even the potential success of a particular advertising campaign for example when you look at a productive model that can forecast how a specific audience segment will respond to a new product launch or how different creatives would perform across various channels this allows advertisers to tweet campaigns in real time ensuring that they are always one step ahead of consumer preferences and Market shifts the second uh area that we talking about is attribution and Performance Management now one of the longstanding challenges in advertising industry has been accurate attribution which means understanding which touch points in a customer Journey are driving conversions our traditional methods that we have been we have used in the past like last click attribution most often than not fail to capture the complexity of what a modern consumer Journey looks like which often spans across multiple devices channels interactions and all now data scientist bring a fresh perspective by developing a multi-touch attribution models that track and assign value to every interaction a customer is having with its brand now these models help advertisers not only gain a holistic view of the customer Journey but they also allow them to optimize spending across various touch points and increase return on investment the third point is about enhancing creativity through data now many of uh many think that data scientist is all about numbers and algorithms but it but it is Al it also has a powerful role in driving creativity now data scientists can provide that creative teams with insights into what kind of content resonates with the specific audiences by allowing us to land impact storytelling and advertising now for example if we have to analyze which images colors or messaging strategies perform best with certain demographics of people Crea teams can tailor their assets for maximum engagement now this blend of creativity and datadriven insights that we see leads to ads that not only capture attention of the customer but also Drive action and finally realtime optimization and automation which we can never ever uh you know forget in programmatic world is in the fast based Dynamic Advertising environment realtime decision- making is super critical data scientists play a major role in automating and optimizing ad campaigns in real time through our machine learning models what we do is we automate bid strategies in our platforms which means that we are adjusting bids in milliseconds based on user behavior and campaign goals now this not only saves time but also ensures that ads are shown to the Right audience at the right price and increasing overall efficiency additionally realtime optimization also provides advertisers to react at a faster Pace to changes in the marketplace now if a certain ad creative is underperforming for example data science data scientists can quickly identify the issue and recommend adjustments ensuring that we are still staying on track with our campaign delivery now with the all of this background on how data scientists add value to uh at teex space we will now move on uh to talk about the tips to crack a data scientist role okay the first tip um mastering the fundamentals I'm pretty sure all of you would have heard about this but fundamentals becomes uh super priority especially when you brush up your key Concepts like probability linear Al algebra calculus hypothesis testing you know these are crucial for you all to understand how machine learning models work second is we will have to focus on program programming skills be it python or R they are most widely used languages in data science and even Master libraries like pandas which comes really handy finally data manipulation and visualization as to how you learn to clean and uh visualize data using tools like powerbi tblo or uh you know thought spot uh in or cbon in Python is super critical when you have all of these foundational elements what you then look at is understanding the machine learning algorithms which means that you're starting with a supervised and an unsupervised learning algorithm such as your linear regression decision trees or random Forest that then you look at how do you deepen your knowledge by progressing into exploring some advanced level topics like deep learning neural networks natural language processing and but at the same time you also have to keep in mind that model evaluation is super critical as to how you evaluate models using techniques like like cross validation uh confusion metrics or Precision recall all of these are super important once you have done this your data querying skills is critical for querying databases learn how to write efficient SQL queries to extract from large data sets you will have to understand the basic database Concepts also in database management like it could be normalization it could be indexing or even Performance Tuning because as data scientists often work with large data sets stored in most often the not relational databases once you have all of this in place data cleaning is is super critical when it comes to cleaning all of that noise and messy data you have your hands on so mastering the techniques of handling missing values outliers data inconsistencies learning to pre-process techniques such as normalization scaling or encoding category uh categorical V variables becomes important so first get your fundamentals right and master every aspect of it the second recommendation that we have is you have to really get handson with projects you may have heard this but until you apply your theoretical knowledge to a Hands-On real world data you will not be able to understand where can problem solving and business requirements fit in so we look at using platforms like kaggle driven data or even Google data set search to find data sets and practice building models or insights then you would also have to Showcase your projects on GitHub or a personal website this is all about building your portfolio real world applications even if it is personal project go ahead and demonstrate your ability to apply that theoretical knowledge and finally don't stop there my recommendation is also start practicing or participating in competitions if you are engaged in kagle or a similar data science competition they will help you hone your problem solving skills and at the same time gain experience working with diverse data sets third once you get Hands-On with projects the third one is all about focusing on business problem solving which means that you're not thinking about data but now you start shifting your perspective to think about business companies want data scientists who can solve business problems using data however you have to understand the business really well if you have to figure out out an opportunity now when you start practicing working on business proper and translating them into data science projects that's when you will reflect on your areas of improvement next critical thinking where you're developing an analytical mindset that is focused on asking the right questions identifying which data or methods will help solve the problem is crucial after you do all of this you also need to understand that tools and Cloud platforms are super critical for example we at version controlling where we are looking at you you learn git for Version Control uh as data science projects often involve team collaboration second is uh you will also have to get familiar with Cloud platforms like AWS Azure or Google Cloud which companies use which are increasingly used for deploying data science Solutions and finally Big Data tools Big Data tools is all about large scale data problems where you learn about big data Technologies like Hardo or spark and distributed computing and our last recommendation once you are able to nail the business problem solving is about building soft skills in a corporate world if you have to stri uh Thrive you need to have the right communication skills where you are able to explain complex technical ideas to a non-technical stakeholders as a critical skill you're focusing there on your aspects of Storytelling with data and presenting insights clearly the second is collaboration where in a data science role you will often work in Cross collaboration teams where you're developing skills in collaboration or you're working with project management and uh Team communication becomes crucial next is adaptability data science tools and techniques evolve quickly and all of us know that but you have to be prepared to continuously learn and adapt to new technologies and methodologies now how do you prepare once you have all of these how do you prepare for interviews become super crucial first one is having exposure to technical questions you have to be ready to answer questions about programming machine learning algorithms data manipulation uh data querying and all of those you have to practice coding and work through problems on platforms uh like hacker rank as an example then you pick case studies you practice data science case studies where you will need to approach a business problem you have to choose the right data and model Solutions and finally start giving mock interviews where you participate in mock interviews to familiarize yourself with the real world interview scenarios and improve your confidence with all of these done never ever forget to network and seek mentorship you have to join data science communities if you have to participate or attend meetups webinars or conferences where you can network with like-minded people who can also give you the leads and insights but it is important to also find a mentor in the data science field who can offer you all advice review your work or help you navigate your career path uh with that I think we uh come to an end thank you so much RAM for the amazing session I hope the candidates got to learn few things for sure so we have few Q&A as well attend so the first question is that what are the some of the biggest challenges data scientists face in the attech industry industry and how can they overcome them sorry I was on mute all right uh good question uh I think there are a lot of challenges and opportunities to uh you know uncover or unpack in uh programmatic space I will give you a very high level overview uh you know some of these challenges can be dynamic in nature uh based on the ecosystem it can also increase our Focus on other aspects of the business for example one of the challenges we often run into is data privacy and regulations uh where you will have to you know keep in mind the constraints around gdpr CCPA or even the latest one India's digital personal uh data protection dpdp bill that got implemented so you will have to keep privacy in mind at at the core uh when you are designing Solutions and building capabilities especially with some stringent rules uh that requires data science is to be you know up to date about how do they extract value from the data that we are onboarding the second aspect is also cookless feature uh future that we have not spoken about but you know you if you're closer uh closely observing programmatic industry you will get to know that uh the deprecation of third party cookies has been something that Google has been talking about uh but for adte companies tracking user behaviors across sites become a challenge when uh you know cookies are removed now where miq stands out is we have continued to build your contextual capabilities you find alternative identifiers and that's the role of a data scientist where you develop machine learning models that can predict user Behavior by either using first party data or contextual targeting or cohort based approaches and all of those and uh I'll I'll probably uh you know highlight one other thing is scalability in realtime processing we deal with large amount of data sets that require realtime decisions on Campa campaign optimization or bidding data scientist will need to design algorithms and models that can process all of these large data sets at scale with low latency responses but at the same time maintain accuracy in Predictive Analytics and recommendations of the systems and I think I've also touched upon cross device activ identification in uh in the slides that I went about and finally one last thing is about uh in interpretability of the models uh you know which means that many machine learning models used in attech today are often looked at as black boxes now making these models interpretable to stakeholders be it your agencies or advertisers or marketers or brands that you're interacting with while ensuring that they are actionable and effective is a very important challenge that data scientist face most often than not thank thanks ra for the response so next question is that how does miq Ensure the ethical use of data in its advertising strategies especially with growing concerns about user privacy okay um that's a good question so if you look at the ethical use of data it's actually through a robust set of policies and Technical Innovations and our commitment to transparency and privacy in a programmatic space now a few things that we do uh you know at miq is one is you have to be compliant with data privacy regulations which means that uh you know internally your legal counsel will have to you know uh be uh sensitizing people about major data privacy laws uh again which is gdpr CCPA uh now DB uh dpdp bill in India uh then what we do is we follow all of these uh regul requirements for transparency and user control over the data which means that you're using robust consent management platforms to track and manage user permissions second is in all the capabilities that we build we look at privacy first approach which means that you're anonymizing uh data but also using capabilities like data minimization we use this extensively within miq to protect personal information so that way all the sensitive data is not exposed or misused and only the data that is necessary for a campaign is collected and processed by reducing the risk of overreach and this is also an area where cookless Solutions the shift towards cookless future we have invested heavily in privacy safe Alternatives such as um first party data contextual targeting cohort based advertising and all of those and maintaining campaign Effectiveness uh the other aspect I can think of is is fairness and non-discrimination like you all have heard about AI right I'm also trying to bring the AI perspective here so uh with every company you know trying to adopt AI into their product and solution capabilities uh We've also looked at where AI is applicable within miq and we're committed also to ensure that our data science models are free from any kind of biases that could unfairly Target or exclude certain groups of people now what we do is we continous Monitor and evaluate our Ai and machine learning Al algorithms that aim to mitigate any unintended bias or delivery and uh of course like any other organization we also have our own internal governance and ethical training practices we also look at Advanced Data security practices uh where we are looking at encryption uh secur storage uh access control or protecting consumer data from unauthorized access and preaches uh so all of the is put together I think uh our holistic data practice to build trust with both our agencies uh and clients and consumers uh help us bring a more balanced approach thanks raama so next question is that what advice would you give to aspiring data scientist who want to enter the attech industry thinking I've covered all the tips there I'm thinking uh so first thing I will highlight is U one is of course I'll keep in mind that you're a data scientist who I'm addressing today but irrespective of data scientist or uh individuals who are aspiring to get into programmatic there should be few things that we should focus on if you have to build your career for the long term which means that you have to look at developing your skills both in technical aspects but also industry specific skills which means that you're exposing yourself to Unique challenges and dynamics of that particular industry which is more fast based uh I'll go back to saying that you have to master your fundamentals really well you have to gain real strong understanding of attech Concepts and I always say this attech is not an easy domain lot of jargons get used there and you have to you need need at least 6 months to familiarize yourself with at ecosystem be it realtime bidding demand side platform supply side platforms there is dmps which is uh data management platform all of these components inter I mean uh understanding is very very crucial now after you understand that then you comes all of the key metrics now uh key metrices is it could be like your click- through rate which is CTR or your conversion rate Impressions viewability uh CPA which is cost for acquisition uh or your roas which is uh return on ad spend how all of these impact decision making is also important once you familiarize yourself with atch ecosystem then the key metrics the third comes the attribution models now as a data scientist you will have to be very familiar with the applicability of uh attribution models like MTA which is multi-touch attribution last click attribution or any other models that measure the effectiveness of ad campaigns across different touch points now uh what else maybe you know develop expertise in handling large scale data uh you know B there are big data Technologies where that helps us handle huge volumes of data but you need to familiarize yourself on spark or Hadoop uh some kind of Big Data tools or distributed data databases um of course real time data processing becomes important um may be specializing more in predictive modeling and optimization uh the reason I say that is uh you know when you focus more on predictive modelings for user Behavior your clickthrough rates or CTR or conversion probabilities becomes more accurate now the experience with recommendation systems or time series forecasting such as these are super valuable for us and optimization is one of the key techniques in attech where you particularly bring the context of ad bidding strategies or campaign uh budget allocation for campaigns so how do you optimize for real-time decisions and maximize the campaign performance is very crucial and apart from that you have to stay curious you have to stay hungry you have to always you know understand where the industry is moving towards as humans we always find you know comfort in growth but we don't find comfort in curiosity and my recommendation is learn all about Solutions uh that uh that are happening in the programmatic industry be it culs future be it connected TV be it retail media networks or Omni Channel campaigns all of these act as uh you know a trigger point for you to go deeper and understanding the industries challenges and how you can build that Competitive Edge for the company that you're working for and where data science plays a key role in optimizing and measuring cross Channel ad performance and all of those uh and of course soft skills we spoke talk about uh yeah and uh continue to network and uh keep learning thank you Rama so we're taking the last question for this session so the last question is how do you see the role of AI and machine learning evolving in the future of attech okay so one of the examples that I quoted about AI um was definitely around how we eliminate biases in the models I'm also kind of thinking aoud as to how we bring Ai and ml into the future of hattech so one is U making the consumer Journey more personalized and helpers in audience targeting for sure um I'm also looking at you know your contextual targeting or Predictive Analytics capabilities the second time the second area is Real Time bidding optimization where you bring automated bid strategies where AI powered algorithms are you know further optimizing all your real-time bidding strategies uh by continuously learning from your auction data and make the most effective bidding decisions for the company uh then you have cost optimizations that you cover like your cpcs cpms uh all of those um and one thing that I not touched upon is fraud detection in programmatic maybe that's also an opportunity where AI is ability to detect patterns of fraudulent activity in real time could be helpful which will make it more precise reduce ad fraud and ensure there are more genuine and reliable ad engagements um I would also say attribution and uh probably you know cross Channel campaign optimization where you bring data science models like mtas or cross Channel optimizations that look at different channels be it including uh display search social connected TV and more so you can optimize this spend allocation by analyzing performance across different channels in real time data privacy solutions could be an area uh we will have to explore and see how AI can you know Drive privacy and preserve machine learning models and uh where it can help us on first uh first party data utilization but the other curious aspect that I'm personally curious is about customer Journey mapping and uh uh you know analyze uh analysis where you're able to build Ai and ml to track and analyze uh customer journey across multiple touch points but at the same time those behaviors from awareness to conversion and how can that deep understanding allow advertisers to tailor their marketing messages based on where the users are in their customer Journey that's going to be super critical and when you look at the lifetime value um where AI can truly optimize that full marketing funnel is by predicting customer lifetime value identifying high value users recommending strategies to engage uh and making it more personalized me messaging uh could be uh interesting one of the other areas that internally we have been exploring is conversational AI um for some of our own inside report uh where you bring in voice-based activities uh which are more widespread and are enabling new forms of advertising tailored to voice search Behavior Uh and then you can also leverage NLP and uh you know AI driven Voice assistance to deliver those personalized contextually relevant ads um and finally I think just being ethical Ai and responsible advertising where you are mitigating bias bringing in AI governance and all of those so in summary I would say personalization automation privacy and ethics and cross Channel optimization would be four key areas that probably the future of attech can heavily rely on thank you so much raia for the detailed response and thank you so much for the amazing session hopefully the candidates are able to learn few things for sure if you have any closing remarks or words feel free to add or we can end deis now thank you thanks yeah thanks z i team I hope this discussion has been help ful uh programmatic is still a booming domain in India uh you will see a lot of this adoption happening in other markets so if you're keen on uh you know looking at adopting data science as a uh stream and programmatic as an industry uh please keep yourself up to date put yourself in some of the challenging situations try to understand how problem solving and yeah uh just reach out to people Network get some thoughts um you know get mentored and uh yeah do whatever it takes for you to you know work towards your goal and we wish you all the very best and thank you for this amazing session thank you thank you