Transcript for:
Lecture on Vocal for Local in Branding

[Music] [Music] thank you so much for your time today uh we have a interesting uh session on hcal for local I'll take the example from the last session uh Starbucks right who doesn't like our names calling when we enter the Starbucks right so just taking extension on this and we'll go on different categories we have our panels from different Industries so we'll just understand how they resonate their brand resonate with vocal for local so maybe I'll open the stage uh for the first uh question that how do your brand inter uh interpret vocal for local and how do you engage with the vernacular audiences right so Kunal maybe you want so uh yeah so let me introduce uh myself I represent uh India's first uh digital payment app beam uh the core idea of launching uh this uh whole T app is to reach out to the masses and the complete uh ecosystem along with the bat which is uh the reason for we are here and um as you rightly said U before reaching to the conclusion why we are in this space let's let me throw couple of data points U Indian EC times and also income pyramid says that the current income pyramid is more on a triangular shape after 20 years we'll see completely different rudimentary Diamond which will buge into middle class with more than 1 billion population um sitting with 5 to 30 lakh of annual income so these are space where there is uh more than 60% of our population is resding uh one more interesting thing is that uh the ecosystem is giving a leverage to tier 2 three cities to come up with more startups uh thanks to our whole India state which is mix of definitely UPI and Adar and the banking system um both Merchant from both Supply and the demand side is getting fulfilled and the consumer at the center stage when we reach out to Indians or bhes um definitely we should have to understand India is the most diverse Place diverse country with more culture nuances and the ethnicities language more than 23 language are considered as official language 100 languages are considered as a uh acceptance from different ecosystem and more than 19,000 languages are considered as a dialects so it is more important for marketers to understand how our consumers are reacting on different campaigns different Communications um if I simply do the standardize seven8 or 12 different adaps to my communication it will not fit into the right uh uh model so it's high time for you we to understand whether we are moving from we moved from transliteration to translation but trans creation are happening or not that is something uh where uh definitely she can come into the picture where she know that the right formats how it works and uh yeah the Forum is more into mtic so I would like to hear more from your side also that how technology can help us to jump into the effective uh Nuance Centric Communications sure thank you so much uh Sonia yeah perfect seg um most definitely I don't think for us um let me just give you a little bit of a context about my business uh I come from a digital and mtech specialist agency and we work for our clients um and we are responsible for the marketing Spence that they are making the outcome based on that um and over sort of the course of many years one has realized as he rightly said I don't think it's about translation it's really about understanding the audience right um that that's the communication aspect of it what do you say when do you say how do you say but I think most importantly um it is also kind of the distribution of the content um we've had experiences uh we work for a client uh a brand that works with Farmers um so that's sort of a perfect example of going vernacular but also based on very hyperlocal insights we tested many ways of reaching out uh and engaging with that audience right the communication was a hit but we weren't getting as many responses so when you start interpreting this whole concept it also means you know what uh channels the audience is comfortable with we realized when we did Facebook marketing we were doing it during the day the farmer was on attractor he couldn't be bothered to pull out the phone and respond and then we try different sort of timings and evening when he's back home perfect timing but then again how do you uh to solicit a a response back it it was uh for us WhatsApp worked really well for some for for north of India down south it was a m call right um so this whole sort of overlapping of vernacular and hyperlocal has many many interesting nuances and in my world uh I don't think there's a set formula because we work with so many different categories with so many different audience Cuts uh that it keeps us awake at night sure uh I'm sure you have a lot of interesting case studies also uh we might discuss it later further on uh vibu uh from your industry OT yeah hi thanks so uh steel industry [Music] next struggling time but 3 rocket science of course but start tier three [Music] Market but Facebook but of course it's a bus it's a 100% subscription based to local startup but local Metro awesome thank you so much uh uh abishek uh maybe you can take example from your uh business the cricket one because I'm sure people resonate with the cricket in India so you can tell us how it works with your brand and uh how do you segment the audiences and how it it's working for you yeah sure so first of all thank you for keeping the cricket theme in today's event it makes uh our marketing thing easier uh and uh Bin in one of the previous sessions actually gave us uh idea for our next marketing campaign he basically mentioned that there is no fun in cricket if there is no score so Crick Heroes is actually that fun in cricket uh because you know we enable scores for millions and millions of Grassroots cricketers uh having said that uh vocal for local uh has been uh part of the strategy of the company consciously or unconsciously from day one because uh when we started cck Heroes back in 2016 we knew that we are building for India uh and we are building for Cricket right which is I mean an Indian sport accidentally invented by British uh so it was not a a choice for us whether uh you know like we should have localization as a part of strategy or not right you know it was from product development to acquisition to marketing to support to even hiring uh vocal for local has been uh part of our strategy all right everybody we've got the other speaker also join us in the panel discussion put your hands together for the co-founder of find we have Mr Faruk Adam in the house thank you so much just discussing the theme about vocal for local so uh we have panelist from different uh uh Industries we have from OD from uh cricket uh gaming uh and then we have from fintech so from your perspective would like to understand how the vocal for theme is implemented or how your audience resonate with with the theme uh sure uh so find uh serves close to 220 million customers uh of the country uh we do this uh via working very closely with uh Reliance retail uh our platform Powers uh the online offline and newcommerce uh businesses of Reliance retail uh specifically the new Commerce business that's where we reach uh tier 2 plus cities we reach uh rural centers and where the large part of the new Commerce consumption happens we look at this in in two ways our products yet are not uh localized and there there's a lot of work we need to do we started some some work around localizing it to local languages uh using Google API and a bunch of other open source uh work but we've never been able to kind of make it natively uh enabled for for the users users continue to uh use the English uh version of their applications and they don't want to switch back uh I also believe that we have not done a great job at that and we still uh as a country as an organization we still need to do a lot more and really think in those languages when we are making our application so that's one uh when we work with uh that part of India which talks in multiple languages all our local campaigns all our local events are in those languages uh but our End to End customer I would feel you know is there's still a lot of work we need to do to make that experience very localized uh for our uh for our customers uh that's on our road map we need to actively work on it we believe from 200 million when we move to 600 million uh customers uh we will have to think very differently uh and what we do now might not be the Playbook when we have to get the next generation of uh internet native users on the platform got it thank you uh next followup maybe question about uh what what are the challenges uh when you were implementing this theme and uh what are the opportunities where do you see the opportunities and the challenges to have this vernacular audience or how do you segregate the audiences what content to show so maybe a example of opportunity and Jaren okay so I think uh communicating with our users is the easier part I mean we we've do a lot of localized communication with our uh end users uh both at online and offline uh modes but I fundamentally believe the people who are building the platforms the product managers the engineers uh we still continue to think in English and we are never we don't have that sense of empathy that we need for our users because we're not thinking in those uh ways all the information all the literature that exists of building products of building platforms uh is primarily in English and hence that's what we kind of have a strong bias on we really don't know how we should build it uh the way you know if you look at this event itself right I mean there's hardly anything written in any other language do you write in any other language no why not that's that's a bias right that's we we are English thinking English speaking English hanging around with everyone watching Netflix and we don't really understand how and and companies like us when we build for 600 million users that people that need to work to do that uh I don't think we have that Talent uh right uh and we need to kind of really ground up think about it and and build it Technologies exist right uh Google Fant has a fantastic uh bunch of translate apis that we can use uh but the product thinking that needs to go in that that's still missing I think by and large Communications has happened a lot of people uh send localized smss we we send maybe in 17 different languages you know activation smss uh to our customers but we rarely do anything on the uh platform and application layer sure uh K moving to you uh since you you come from fintech background so I'm sure there are a lot of regulations and guidelines so how do you see the challenges and opportunities so before add to that uh question what he got is the right format where uh empathy is the main reason where why brand marketers are not thinking in a right format uh irrespective of any any business whether it's a fintech or any other app or even not in the app empathy is something which is more friction level and that to uh emotional friction so if we understand why someone is not using your app or your business uh they into three different quadrant um emotional cognitive and functional empathy comes always on the top delight and empathy both are opposite but um if we don't touch that part and if you further Deep dive the reason behind emphasizing on empathy uh we Indians though we understand English but still we love to see our mother tongue uh usage by some other person when someone just uh acknowledge me and my other I would be more uh you know comfort and more relate with that person or the brand so as a brand Persona uh I should be inclusive in nature uh with that regards beam from day one is into 22 different languages it's a big achievement for us because we understand that the languages like um miso or marari which are not the small format but in terms of volume might be less uh we are there and uh if we don't fulfill the uh promise which we are doing a brand promise there's no point to putting a single penny on it uh with regards to compliance uh definitely is not something which stops us to do that uh it's up to our will and the commitment uh what he saying is perfectly right that it starts with a product development phase and that's what something which we are working on it right sure can can I can I yeah so I think empathy is a great point and I agree with Faruk uh great point he brought up for Crick Heroes because as I mentioned you know we are building for India and building for Cricket you know like we had to be mindful of all the audiences which were there right so uh even though initially we started with a English only app uh very soon we realized because of people who were there on field or even our customer support team right who are talking to our users every day we very soon realized that unless and until you know like we don't talk to them in their language uh specifically from product perspective you know like sooner or later we will hit a roadblock right so that's when you know immediately we started working on localizing our app at least the core flows of the app if not entirely right so we the the app is now available in 10 languages our customer support team speaks to uh in Hindi uh to our users right that gives them uh Comfort we also have couple of guys who can speak South Indian languages and all right so it's not just I mean just Hindi is not enough right so that Comfort also is very very important so it's not just about product but also about you know like everything in the company as a strategy how can you localize uh that you know we hire cricketers we hire people who speak different languages just so that you know they understand different different mindset and different different uh ways people play cricket awesome uh vibu uh I'm sure you have different content different languages Hindi English and other local languages so how's the response been and uh what's a split between non-metros metros I'm sure you have we we were discussing back uh uh just now that you started non metros and Metro so little bit on the Journey part Tamil South languages Mal Bali atically YouTube through social local by not [Music] Automatic Auto dubing local language DJ thank you uh Sonia you want add something from your uh branch what they doing on this yeah sure um most of the work with that we do Lands up being in English I think that's kind of the mindset or Talent Fe in the industry here right um so strategically what we've started doing is based on our clients demand to try and hunt for talent from local local markets right one of course you know there's the translation of things but uh idea be or insights or observations or local communities comes from there I don't think we can sit in Bombay uh in the office and figure out what someone wants in jarand or rachi right uh so that is one uh and sort of the second uh kind of challenge SL opportunity is that um we are all in a digital world right Nothing is Stopping Us from testing even language testing but it's not such a prominent feature in our lives right um we've seen that where we've done um Communication in vernac we've seen up to a 60% incremental uh outcome so it definitely works but I think the start point is uh a the foundation uh around which we think which everyone sort of spoken about but you your product is like that um that's a big Challenge and also an opportunity oh great and uh I mean since everyone mentioned about there's a lot of data and everything so I'm sure there's a lot of AI ml text tag that you guys use to dissect the audience or Target them in the local language or maybe different content everything so do you think uh text tack or MCH is important and how it can help you to unlock the potential I would like to jump on that question I mean uh I want my hand way early when it requires to using texting to jump on a one solution um it's all about uh expecting too much from say productivi or gen whatever to we name it whatever the version we name it because till the time we don't feed them uh the emotional aspect and that to when it comes to the finance of payment where I into uh is not there uh we Indians whether we spend or whether we transfer one lak rupees or one rupee we make sure that uh we need insurity comfort timely response we need safe secure and seamless transactions or whatever so at the time what I found that it's too early to jump onto uh say communication aspect and that to inv NE with the use of ji I would love to see something from you guys or some other player in case you want to but um with regards to uh going on a positive direction uh taking help of creating different cohorts different brand personas can be very easy when we effectively use uh AIML uh using them to uh Reach Out Right audience at right time modu with Geo fencing in a large format like India is something with definitely vouch for it and we do have a good uh result out of it sure abish you want to add anything on the Tex stack or yeah so no I mean we don't use AI machine learning uh those kind of tech for U vocal localization of the entire thing not yet but it does give me an idea so we do use AI uh for generating automatic commentary uh ball to ball uh right so you know like whenever there is a delivery you know like now we are uh putting similar kind of commentary which you will see probably on Crick info cckb so that commentary which is like written by humans similar kind of commentary is now generated by AI on Crick Heroes because virtually it is not going to be possible for us to write commentary for every match that is going on so you know we have to use such technology but now I think this particular conversation gives me an idea that you know we can actually use uh apis to translate the same commentary into multiple languages uh right and I think I think that can be become a great product another use case which we are working on right now we haven't launched it yet we are working on pre-match show and postmatch show uh generally you know like the kind of introduction uh somebody some host gives before a cricket match now that entire thing can also be generated by AI right so based on the past performances of both teams the players who were playing on those teams and the ground conditions we can create a 2-minute uh summary of the match uh you know like a pre-match show and then that entire thing can be translated into an audio with using another audio uh tool audio AI tool right so now that entire thing can also be um done using I mean localization apis in multiple languages imagine you know like a Hindi commentator and then you know on top of it you add a a known voice like Hara b or uh you know Akash Chopra saying the same thing in Hindi right now this is a true kind of localization uh for Grassroots Cricket uh and but giving them the same experience as international cricket got it uh viu uh you add something intern of course quarterly orans of audience okay far you want to add yeah sure so uh between find and uh Geo uh very specifically on two types of foundational models we are localizing it uh there is B GPT that is being built by uh Geo uh we're also building another version of that called tiger boost at find where we are fine-tuning uh fine tuning Lama 3 with a lot of internal data which is you know India specific business specific uh and some part of it is uh localized but that's that's been fine tuned uh one is being foundationally built and one is being fine tuned we're also building uh bhat diffusion uh one of the problems that we uh came across when we using you know open Ai and mid Journey uh a lot of the images that you try to make for India and Indian use cases uh they just don't turn out well uh and we're a country of 1 point you 1.4 people you know we are you know a country of country in in some sense and if we are to kind of do this we'll have to build Foundation model which serves our our need right I mean if and in in the model that we've made uh you can actually describe things in in in local languages right you can write you know panipuri with Chutney on a five star plate and it will make you uh that so all of that is you know contextualized for India so those are the foundational work we do uh a lot of the IML work that you you kind of expect to happen in a company like us as all of that is happening but these are the two interesting localized projects that uh we feel will move the needle for for the country got it uh maybe we have uh uh few minutes now so maybe one or two uh key takeaways for the audience that if someone is sitting in the audience from your particular sector or domain what should they do a single point or two points how how should they strategize to implement this theme oh for the audience foundational models or maybe some strategy that you tried it worked out and uh maybe one line or two lines what they should try doing it uh I mean I mean making Foundation model is a fairly expensive activity uh I would I would advise if you're building generative applications or if you're going after generative uh use cases uh focus on the top application layer uh there are larger companies there are country specific projects that are running uh which will give you the foundational models don't spend too much time and effort to try to you know get CPUs gpus Andy to get this large training data set on which you can train it uh use anything out of the box go go backwards from what the user needs and then figure out how you can build it generatively magically create all sort of next generation of applications products and services uh rather than spending too much time on on on on building foundations uh that's a very expensive activity very few companies uh in the world can afford it uh so just focus on that and that would be my simple advice you know and anyway all of the uh riches is going to be in the application layer right I mean no one is going to become rich by giving just another Foundation model because the moment you make one Foundation model someone else will come with a better one and there's always the Llama Llama 3 four five sort of you know Company open models that come out sure uh abish you want to add anything any inputs no I I I agree with Faruk no point creating the foundational layer of this technology the technology is there even we took a lot of long time before implementing anything uh related to let's say generative AI or any similar technology because you know it took us a long time to find out the use case uh finally you know as I said the ball to ball delivery uh automatic commentary is one use case which we have found which cannot possibly be solved with any other approach right so you know like until and unless you find that so I mean apart from what faruk's advis I would also advise that you know like just make sure that the use case is also extremely scalable you know like unless and until uh you know the activity is something which cannot possibly be done you know without any other alternative no point uh you know like uh trying your hand on even application layer you know like for example you know his example was also a great one right automatic generation of subtitles in multiple languages now this is something you know at scale it is not going to be possible it will be very very expensive so then this is the right problem to solve with the help of generate UI or similar technology actually be sorry problem actually more or less Sol I have heard the demo of thing which my our team has created you can actually have voice which is let's say very similar to an actual guy and tomorrow as I said I'm pretty sure after hearing that I'm pretty sure we will be able to license Ravi shastri's voice one day and we'll be able to see that can be done yes yes awesome uh gral from fintech side any suggestions before jumping into any business model or U what do you call development phase my suggestion can be very simple that find out the right use case a it has to be scalable B uh the India Tech is ready it's a it's ready format and everyone is using it and we are going out uh to give it to some other developed countries so it's high time where we should have to understand that we understand our segment first uh whom we are approaching segmentation has to be right and perfect uh drill down to multiple different formats of uh Persona mapping if it's not happening perfectly on day one your foundation is we uh India is data reach country uh we do have a perfect MX Tech available with us different players are there so no need to jump on that direction make sure your persona and cohort creation are right from day one otherwise your entire money or enre exp goes on not train and then the drills are very simple we have seen the good examples iners awesome thank you uh any any questions from audience uh do you want to ask any question about your business or strategies of particular districts VES avability Poss [Music] soft stand alone stand alone stand alone stand alone thank you any other question from audience okay uh thank you so much again for your time today and uh it was great I think uh great insightful thank you thank you [Music]