Introduction to AI and Audience

All right good afternoon my name is Heather Daset and I am the marketing and event coordinator here at net elixir I’m super excited to be here today as we introduce AI and building a successful search marketing campaign to target your high value online shoppers egg is revolutionising paid search and trust me each day it is growing in importance these dynamic changes are impacted in the entire industry and leading marketers and retailers are adapting to these changes in order to stay competitive net elixir is offering throughout the month for webinars that will help you stay current with the developments and AI let me tell you about the webinars we’ll be hosting this month which are all centered around modern search today we’ll be introducing a ioan audience tomorrow we’ll be covering the fundamentals of smart bidding measurement and creative excellence net elixir we’ll be attending the Google marketing live event in San Francisco on May 14th there we’re going to catch a new customer insights and the latest digital marketing products Google is offering on many 17th will be hosting a webinar to share the key points presented at the Google event to end the modern search once a month we’ll be joined by our premier partners at thing to share with you the most important AI insights for those of you not familiar with net elixir our company has always been able to help retailers online find and acquire new customers we go beyond looking at the data we see the value and every click and the customer behind not click the idea of humanizing every click is what we believe really separates us from the crowd and it allows our clients to position and target more strategically since it is truly important to us we have made it our mission to help companies succeed online by humanizing every click here at net elixir we are comprised of over 130 fanatically analytical global search marketers dedicated to helping retailers find and acquire new customers online since 2004 our data intensive approach and deep retail expertise have delivered success for hundreds of brands and a highly competitive marketplace we have built powerful relationships with partners like ups Google Bing and bright edge and we have connected over 400 retail clients to their customers around the world it is our job to make sure that your brand owns the search bar and we allow you to take ownership through the services we provide which are paid search organic search paid social web analytics consulting CRO Amazon ads and e-commerce and tech development services without further ado I would like to introduce you to today’s speaker and the man who started it all you dainties Eden as a founder and CEO of net elixir since 2004 he has been committed to helping digital marketers attract acquire and engage with their customers more effectively his determination and exclusive focus on the online retail industry has helped net elixir become the de-facto search marketing expert in the retail space prior to now elixir you dean was the director of business development at party gaming there he built a business unit party bingo comm from scratch which went on to become a major revenue generator for the company and today he is a guest lecturer at several universities and trade show events so before we begin as a reminder if you have any questions there on our webinar please do not hesitate to drop your questions in our chat box we’ll answer all questions at the end of our presentation all right everyone please welcoming and then join a new day in Bo’s thank you very much Heather thanks a lot for the introduction I mean this this presentation is quite unique primarily because over the last 18 months or so we have seen that the entire sem world of the page search advertising world has really changed quite dramatically but before I sort of just go ahead and share the stuff I mean I just wanted to ensure that everybody is able to see the screen and hear us okay because we have had some some complaints and some challenges if you can just a faggy press on the question box I mean if the I’ll be honorable enough and I see Dean David you guys are perfect right thank you thank you very much so just kind of starting off and going into as to what heather was mentioning it has been the it has been a time of change it has been a fascinating time of change has any such times there are there are new opportunities which are bound to really sort of rise so I just start with a couple of facts which hopefully helps you to understand the magnitude of change which the paid search world that the search advertising world has been going through the first fact is sixty percent plus of our clients paid search revenue now comes from modern search campaigns when I say modern search campaigns I am referring specifically to the the the new AI modules which Google has driven sixty percent plus eighteen months back this percentage was less than three percent so when something grows from three percent to 60 percent you know that that stuff’s just massive change happening right there the second fact which I was share is I’ve a nice or a first coat maybe a Google slide as well so overall companies who have embraced or adapted modern search modules they are seeing or realizing a 27 percent increase in conversion volume by combining the power of audience automation and attribution they are getting this lift and this is across the board for Camille so we decided to test as to what has been the lift for our clients who that have really embraced the modern search modules when it comes to paid search and that percentage leads me to number the fact number two for our clients that increase has been Believe It or Not 40 percent plus the conversion volume has really gone through the roof so based on both these two two insights or facts that I shared one is sixty percent plus of the overall paid search revenue really coming from the modern search modules as well as the the overall lift in terms of the revenue and the overall performance of words of 40 percent really makes a almost imperative that every we have share some of the best practices some of the new stuff which has been going on in the world and this this is really the world of modern search when we say modern search let me classify as to what exactly I’ll be referring to we have broken down modern search into four core modules the first one is creative excellence so running a set of smarter ad copies and tomorrow my colleague and our director of paid media money should be mentioning and discussing more about the creative excellence part whereby you just upload a certain number of ad copies headlines description and who will figures out as to what is the best combination for every particular audience segment the second one has to do with automation which is we’re talking about things like smart bidding so rather than manually changing the bids are using a rule-based tool to change the bid google says that we have access to a lot of signals we change these bits for you you just sort of set set overall goals that you are going to reach so that is what really has made smart bidding popular and so also has it made the dynamic search ads very popular as well we are directly using the data feed with the retailer is sharing with Google they are able to populate or generate the ads directly the third component of the third module is on measurement where it is primarily under the data driven attribution very clearly the last click attribution may not really reflect the overall customer journey and really give a tidge appropriately to the right channel and that’s where the data driven attribution has become an important part of the modern research and last but not the least which is going to be a focus of my presentation today’s about the audience segmentation and targeting so but before I really jump into all of these let me try to do a little bit of a contextual and presenting a bit of a structure or a framework which will hopefully help you understand is to wear AI modules such as modern search Modi was really sort of fit in but to do that let me go back and let me go back to something as fundamental as the definition of work I have taken this taken this diagram for this fantastic book which I would highly recommend anyone who was interested in the world of a I do need specifically from the business perspective it’s not only a technical book or prediction machines which has been written by three professors from the University of Toronto they have broken down every task or let us say any tasks that we do can be broken down into the following components for every task you are doing that task because you really want to achieve a certain outcome or a goal there is an input for the task you collect all the information to go ahead and do a task and then the core component is broken down into three so you have you are predicting as to the changes that you are making as to how would those changes interplay to produce the optimal outcome or the desired outcome you’re also making judgments what path should we choose at what point in time should we choose one path and that’s where it is entirely based on your experience plus there is a lot of human intuition and universe where and then as you really carry out those actions which you are doing we drag the outcome and then there is a classic feedback loop almost like a learning loop which keeps on which really helps you as a human being to do that task better and better and better right remember that 10,000 hour rule that if you kind of keep on doing something on a on a regular basis for 10,000 hours you can really achieve mastery on this then there is a fantastic book again called flow where they bring in the concept of deliberate practice by by putting the purpose in the practice you can really achieve mastery in anything that you want after 10,000 hours being invested in that and that’s where you’re sort of learning constantly the human the human being is sort of learning to sort of becoming more proficient in that particular idea now for a moment let us just consider this process of tasks and just consider just one small but fundamental component called decision making now any decision that you are taking let us say a search marketing agency which has been around for 15 plus years now we have as we have over 40 40 search analysts who are working day in and day out focusing on just a handful of things what are the right moves what are the right choices to make what are the key words to do what are the bits to they separate cetera so in any cases they are making choices they are making judgments they are making a strategic judgment that this is the direction that we should have sort of move forward to so that judgment component is something which makes it unique suppose you take a strategy one or let us say path one how does it really compare with path 2 and path 3 so it’s basically a question of trade-offs what path should I take and that’s where the human judgment component can sail but with every path there is a set of actions which are filmed in a punitive in nature and these are the actions which which you are sort of just jested checking different variables for all the different sets of and copies are getting tested and you’re finding out the results for that you are changing different keyboard mixes you are sort of going ahead and changing the builds probably multiple times and so on all of these are fairly definitive activities but needing to be done so that you are able to rather than only sort of setting up a path and not doing about it anything about it you can actually do something find out whether it is working or not once you are able to take out an action and map the outcome and then sort of that process is learning what AI does AI with all the developments whether you talk about AI as a self-driving car whether you talk about AI in terms of probably a warehouse robot who sort of going ahead and segmenting picking stuff and putting it on the shelf or even finding out as to what stuff is missing from the shell or whether you talk about AI making those changes which are fairly repetitive in nature in terms the number of big changes which earlier used to be done by rule-based algorithm or portfolio algorithm or we probably are finding out the different audience sets demographic audience States which would be fit for your campaign stuff which would have mechanically done that is the part which AI is able to do substantially better than human beings and why is it so why does AI or why does an evolved AI module is able to provide prediction value much higher than human beings primarily because for anything to act yes this first part we has the input part so input is a bunch of data you have we are getting a tons of data and at any point in time the human being is probably tweaking around and experimenting or testing maybe about three four five ten pieces of data signals ourselves on the other hand an evolved AI technology or I would say more appropriately an evolved prediction machine is able to test out literally hundreds of in certain cases up to 150 to 200 different signals to do that prediction a lot more efficiently now these are repetitive jobs which are getting sort of done so effectively what does it what does it really mean in terms of search marketing earlier an action resulted because we were able to take a judgment even able to do a set of tasks which was more of the prediction and then you sort of learned as a process in this the judgment and learning is something with the human beings can spend a lot more time while giving off those repetitive tasks to the prediction machines which is at the center and that’s a key component for anyone to understand when you talk about anything on a I so that is a framework I really wanted to set and effectively all this is possible because as we all know we are into a massive massive massive data revolution the quality as well as the quantity of data that we are able to get as it is sort of classified let us say as big data right you have the variety of data different disparate data you have the velocity of data what is the amount of time the volume for a certain time and then you obviously have the overall volume of data as well that is massive that Google has and Google’s algorithms that are of taking all this thing so do those repetitive tasks better than probably a human being can do but there is one thing which I really have to sort of caution all of you are which is there is no no substitute and no foreseeable substitute in terms of human judgement when I talk about human judgement you can there got is about the robot which is literally the search marketing agency or our best friend at this point in time we have to strain the robot driving the robot in a certain path because if you don’t really strategically leave the robot to a certain path and the robot comes million to a set of activities in a path which can really lead to nowhere right so it becomes quite a fruitless basically useless sort of an overall deployment of AI because you didn’t really have the expertise to delete down a particular path so that’s where the human judgment which has been really essentially through a simple thing experience and then secondly there is later the tremendous value in what is called the intuition as well which has been built as a combination of experience and intuition becomes invaluable because if there are 50 different paths which have to be selected human beings can really just based on the experience and intuition right why don’t you take Path number five why don’t you take Path number 35 and so on and so forth and then apply the machines predictive capability on that so as to be able to do those tasks a lot more efficiently there is one other opportunity with this this unique division of almost like the task or the labor railroader provides the unique opportunity here is expansion in many cases just because as human beings we sort of simultaneously used to Sardinia are used to this process of making predictions making judgments minton predictions making judgments in many cases did not even have the time to spend to really make those acute number of judgments which expands the overall spectrum right so we were always focusing on let us say X percentage where X percentage definitely was not 100% and that really limited the scope now with machines taking care of the repetitive jobs or the or the predictive jobs suddenly you have time to give machine or train machine different intuitive challenges or experience led challenges which are almost like the judgmental challenges which really enables you to expand overall campaign structure significantly so and as a result of which the retailer has suddenly the same advertiser has many many many more opportunities to overall grow there in part because suddenly the addressable universe that you are looking at expands multiple fault in certain cases exponentially so what I will try to do is I try to emphasize or share as to the importance of judgment as well as the importance of predictive machines as well as through a certain example and then I sort of try to share a case study of one of our one of our dear clients Sarkar lighting who we have been working for for more than three and a half years now and in fact I sort of presented this case a study along with their Co Elba’s in just about two weeks back at the co here one conference as well which it really help you realize two things one the importance of judgment in taking a particular path so as to minimize the risks risk and maximize the overall desired output at the same time I’ll also show us two if done right as to what is the impact it can have as well as you’ve done wrong what is the what is the negative impact it can have ass we’re talking of the negative impact I mean but before we get into this let me again try to clarify as to now we have this we have a framework hopefully of dividing a task we have a framework hopefully based on which we can better understand AI so again let me repeat some of these pieces borne any AI to at this point in time is more of a prediction machine primarily because they are able to do the number of repetitive tasks based on the number of signals that they do better than human beings right it started with there are there are a lot of chess games there was a go game which was there etcetera etcetera and there is a lot of work is going on as the as the algorithm learns constantly over time but to do that the second part which I would think is the number one value it leading down the pad comes through judgment and there is no replacement now or in the foreseeable future in terms of human judgment which has been built as a result of experience expertise and intuition so that’s the path if you really put the paths of think of AI as someone who is a collaborator for you who can really take care of those repetitive tasks and help you think of others paths or other strategies to maximize your overall growth potential so these are the four I do not really get deep into any one of them focus more on the audience iris at which the dynamic search ads and then we have the detailed demographic segmentation and dynamic remarketing a prospecting and then detail audience are I not really get too detailed into each one of them and explain because there is a lot of material there in fact we have Netflix that has produced an audience orange paper or a audience best practices checklist you can download that using our modern search link from our website as well so that’s something that you can do yourself and then tomorrow my colleague as I mentioned one each moment would be talking about automation creative excellence in the measurement we explain but our theme for both these days essentially would be to share with you some of the best practices when we say best practices we have found out a list of things through rapid testing and constant testing over the last about 18 to 24 months that really would help you improve your judgement as a human being and really lead the AI in the right direction very important improve your judgement and leave the AI to the right direction because if you don’t lead it to the right direction this is what happens so this life that I am sharing with you this is actually for a large consumer electronic retailer client that we have and this is when it q2 2018 when we started off effectively AIB sort of did a bit of a test and we like to compare it with our pre AI module in this case we are talking about smart feeding the pre smart bidding days and what we saw in a span of about three weeks or so the impressions and dropped by 36% but even more concerning the the conversions came down by 32% then the revenue came to all the way 44% now this is a massive massive overall revenue search advertiser as such and understandably – 44% drop is huge in terms of the absolute dollar number why did this happen primarily because we found out that when they were sort of testing the AI modules the path that had been selected was not the most optimal path they did not really use that expertise and the experience and intuition a combination of all of these to really lead the AI or the machine in the right path leading to a very dramatic decline in the overall performance numbers we were able to sort of bring it back to level and that campaign is now doing well but for those three peaks I can assure you it was it was a challenge for the massive challenge so that’s the reason it’s very important for me to keep on emphasizing this that expert judgment ensures that mad machine is acting on the right path and you don’t experience a sudden drop in performance because in many cases if you don’t know what to do you you can’t it can sort of get leading to a downward spiral and you can sort of blow off a lot of your budget and see a massive dip in performance is very quickly as well so a positive can be a massive positive well a negative can be a massive negative as well so that’s where use your judgement with a lot of caution now we’ve got about as to what one of the challenges are but having said that I also started this presentation segment by mentioning that we are seeing a massive increase in the overall revenues so I didn’t wanted to share with you through the example of one of our clients are colliding they are a premier lighting retailer based headquartered in Savannah Georgia and this is basically their case study so again the combination of both the right path or the expert judgment plus the power of a I drove 75% increase in our wife or circle lighting before with our little bit about circle lighting they are they’re a premier retailer of lighting products headquartered in Savannah Georgia again have been in the online space for fair quite some amount of time I mean they’re their segment customer segment is normally affluent Americans who are primarily sort of living in the coast who sort of tend to buy significantly from circa lighting so when we started building the AI campaign before that we really have to take a context as to where things are because you are looking at AI to improve what pre I campaign for pre air campaign performance was and circle I think again this is these are some of the numbers that were there as you can see I mean we had already scaled their campaign substantially that will be generating more than 300% in terms of revenue growth in three years it was an incredible growth story so we were really applying a ie with this sort of a foundation or a context when the when the actual campaign management or performance optimization or ours our own standard technology based bidding etcetera had led them already to substantial success we Stalin ji said that how much more can we do to do that we did the following things the first thing we defined the high-value customer persona for circulating very important right because to decide a path that you will follow decide a particular strategic course of action you would follow you really have to decide as to who’s your core customer who is a high-value customer so using our proprietary tool Alexa insights we are able to track the customer journey and period these high-value customer persona also using their internal numbers we were able to build out the high-value customer persona and this is what she wrote for satellite she is she lives mainly in California New York Massachusetts Connecticut and Tennessee this is where it becomes more specific because where we are all using our own technology Alexa insights to find out she uses long tail specific and intent driven search queries which are of the length of between four to seven keywords while searching so she she knows what she wants and she sort of likely to type in a longer search query sort of find out she completes purchases mainly on Mondays but biased the more expensive items on Wednesdays she falls in the top twenty percent of household income if you really look at the overall US census date and she would be in the top twenty percent she takes time before buying it’s understandably an expensive product a premium problem as a result of which the overall purchase time can be anywhere between three to seven days we call this in the very technical term something called latency latency is the time between the first click on a Google Ad or maybe sort of landing on the website and that she’ll party has been made she visit this time between this time three to seven days she visits the site seven to eleven times prior to communion so she is bigger finding out considering different options and so on this this this group is mostly women in the age range of age group of 35 to 45 44 she completes the purchase mostly between 11 a.m. and 2 p.m. so again if you really connect that she mostly complains approaches on Mondays and when she completes the purchase it is most likely to be between 11 a.m. and 2 p.m. so around lunchtime roughly she mostly a luxury shopper she also is a business traveler she travels quite extensively and she’s also a home decor enthusiast who loves to shop in stores very important insights now I must mention that all of these insights that we are saying is it’s very important for it would be fine your customers will whether you are sort of using using any channel whether it be sort of Facebook we do a fair amount of Facebook for circle lighting whether it is SEO channels anything or seconds I think of marketing 101 is always start by defining who the high value customer is because then unless until you do that what strategic direction we’ll use R of T right you can go into a strategic direction chasing probably a non high value customer and end up spending a lot of money not very efficiently and the last part and she uses about 2.2 devices she probably heard of does a lot of lot of lot of this browsing part or not that is a search shop and by pattern on her desktop most likely to be Mac in this case and also on our tablet as well again most likely to be an iPad as well as fall also but interestingly she we found is more still more likely to purchase using a computer primarily because I think again we have something to but I think it’s more of the psychology of buying a very expensive product when I say expensing about $1,500 plus per item is the average order value so when you look at sort of buying that level of an expensive product I mean people still prefer to use a computer to really sort of make the purchase or complete the purchase I should say so first of all step one when you are going to do anything of the hey I start by defining the high value customer persona all of these points are important we use using our own we are able to find out the data points for many other factors as well but once you have really defined that then you sort of move on to very specific things again this is a very dense light so I not only spend too much time on this part but overall I think the impact that we were able to get the adoption percentage was less than 10% about in the month of January or February 2018 and currently 62 percent of the page search revenue for circulating is coming only through the the modern search model so what have they done the first of all we have really defined the high value customer persona and I would again very clearly mentioned it is not really only one person there are about five different personas which are applicable for sucker Lighting’s friend in certain cases we work with some large companies which are very complex I mean for example for Lenovo we we have about 47 different customer personas each one is a different path and a direction so again keep il keep on emphasizing the importance of taking that right strategic part based on us – who’s your high-value customer persona that you really want to attract and mean so from the smart binning again with human manipulations we we were able to drive I mean significantly higher 154 percent higher CTR the CPC’s went down and always we were able to do that because we lent them in the right path the same for dynamic search apps again we sort of beat the industry by 200 percent for this campaign – when I say 200 percent this is further but for the industry record which I would have brought up by Google right so what happened how did we really beat the industry by 200 percent what where others doing and what were we doing differently compared to that so that is where I think the judgment of the expertise part comes in and then you have the responsive search ads which is coming up with those combinations of the most optimal ad copy combinations for each individual audience type so that you are able to win maximum percentage of audiences then we had the this would be more of the the audience ID as well new detail demographic targeting customer mass smart display campaigns emails and so on but I think the purpose of these slides they only go very deep into each one of them but the idea was again if once you are able to set it up in the right way the numbers can be extra as you can see just look at that line which has audience maximization around site visitors you would see that the traffic went up by four thousand percent and I am telling and this is this is not suddenly out of nowhere I’ve already shared with you in the three three and a half years that we have worked with some time before we even started up this thing we were on a very strong platform because the entire account was going to matter people those are those are the numbers that you probably cannot achieve only through human judgment you really have to get the prediction machines to work for you you really have to open up the space using the power of modern search of AI so as to be able to get to those extraordinary numbers as such the impact of modern search campaigns specifically when talk about the audience the Hawaii as I mentioned earlier went up by about seventy five percent the click-through rates were again through the roof so how does this how does this work so in a very simplistic sense let us talk about the audience campaigns so earlier through the different targeting mechanism your holes already once they have defined the high value customer you are able to set campaigns around each high value customer persona now if you really do it manually you probably would have an X number of potential customers that you can really attract or get them to buy from your website that is where the prediction machines of the the audience the smart champion optimization the audience optimization brief sort of comes in once you have really defined those very variables very granular design by the customer what is our OS for a unique individual segment what is the overall ad copy miss I think that you would ideally sort of want to share with that segment and then you set it up with the smart bidding algorithms suddenly you are able to see as to how we are able to not only extract efficiency from those campaigns but how essentially the audience maximization algorithms which of the area algorithms again that Google has developed helps you to expand the pie suddenly the addressable universe grows from maybe 10 to 100 or even 1000 now these are the customers who probably didn’t even know that existed and the way that Google is able to find out is by relying on multi signals remember that massive the big data slide that I shared with you there are so many signals that we don’t even know that we will have access to just a simple example let us say they have access to the weather conditions they have access to not only the device type but hour of the day but they have access to a lot of information about that instantaneous buying decision as well as more when I say more they also have access to what are these these same customers or these strange shoppers beyond purchasing on circle lighting what are the other websites that they are engaging with what are they doing on those websites and there is something called an affinity audience which sort of matches if someone is sort of probably more of a luxury shopper she may be a high likelihood of being a target for this particular segment as well so it is doing those affinity audiences the in market audiences etc etc now in-market audiences is someone who has expressed an interest in terms of making a purchase like this so those are the signals where it is going just beyond your business does specifically your account numbers and so on and expanding the pi sub me as I mentioned suddenly you have an addressable universe which you thought was ten people now it has become 70 hundred people a thousand people and that is an advantage when you have really led Google in the right path you have really set up the campaigns fail and you have defined the high value customers fail you have said the ROAS thing and then you select as to which which algorithm should we even be selecting what is the most optimal algorithm which would make the maximum sense should I really apply the TR OS of target ROAS as the key goal or should I look at the TCP see as the key goal or should I believe you look at the T maximize conversion as every goal so I think all these are the judgmental part and then when you have set up the remedy to work as well as the expansion work I think AI is able to do it phenomenally well now I just try to have a wrap up with a list of list of best practices which can be there now all of these best practices will help you to improve your judgment when you are setting up an AI campaign or an audience maximization campaign they are they are practices that we have developed as I mentioned over the 18 to 24 months working with a wide range of customers and this will really help you set up and really run the AI algorithms if you need help we are more than happy to sort of LP out on that but these are some of the best practices you can also download this checklist I forget to get to a link right after this so test all the audience types because first of all you won’t really you have the high value customer persona but at the same time maybe there are certain audience types that you don’t really have at this point in time so testing constantly is very important so as to find out the the various types of high value customer persona so the entire concept is every business doesn’t have just one high value customer persona they probably have five ten as I mentioned in case of the Novo 47 different high value customer personas so have you identified all these high value customer personas is a big question the second part is allocate spin to focus on the top performing servants again that’s the more of a judgmental part how much money should you invest in each one of these segments how much money should you invest in each one of the high-value customer segments and before so you are essentially training or giving the information to that algorithm to start working so you have these are trying to make that algorithm kind that algorithm in the way that sort of optimizes overall account structure you have to determine if you really need an audience layer versus standalone targeting so it really is this high-value customer segments so big that you really want to have a separate campaign built entirely around it or should it be just an audience layer where it all filters out of anything which is not in line you can test combined audiences as well so suppose there are two audience segments is there any intersection to think of almost like a Venn diagram right two circles intersecting and the intersection point I think we have found to be exceptionally helpful but that can come through a process that really have to have to implement which is rapid testing unless I’m going to test you never able to find out the intersection point what is that customer segment one customer segment to what point in time do they intersect and so on so again you are using the power of human insight more of in this case sting and the inside that you are able to get out of this thing are in the human judgment becomes important apply smart bidding with select audiences to maximize the impact of against don’t apply smart bidding everywhere I’ll give you a simple example suppose you are running a season and campaign so we have Mother’s Day coming up right now effectively any GI algorithm to run requires a sizeable chunk of data and that’s what the input data and then the AI learns which is typically they’re sort of there is there is a feedback loop as I showed you then for the feedback data and the third component is where the human judgment you are trying to train the AI advice that comes our training data so we talk about data there are three types of input data feedback data and training data right now in this situation for seasonal campaigns you do not have adequate time probably to run AI or any any audience maximization campaigns primarily because it’s a limited time and during a seasonal promotion like Mother’s Day or let us say Valentine’s Day or during the holiday season lives around Black Friday the promotion keeps changing so dramatically and so quickly the technology does not get enough time to get an accreditor to act right so sometimes your numbers your campaign performance will come can be completely off the back if you don’t really accept that fact so AI algorithms are very applicable on steady state campaigns I would say statistic happens of course a regular online campaigns and they are very powerful once they start having an outdated or a slowly scale up and beat but for short-term campaigns very specific period short period campaigns they would have we would we would definitely recommend exercising caution or probably in certain cases not not applying them at all and last but not the least engage in continuous performance testing because as I mentioned the I tried to highlight by giving the example of the performance deep within 44% deep within a span of three weeks that’s substantial so you have to constantly keep on top as an ask of under module question is what is in what working at the same thing you have to remember that able to get together enough time if you keep changing every week they understandably not only we will say any I and the world would not really work for me so you will give it any we’re time and we feel that the adequate time is anywhere depending on the size of the data set that you have but we have I sort of mentioned I think during the cohere 1 conference keynote presentation as well the adequate time when we see substantial performance left happening comes anywhere between around 125 to 130 days after the AI has been applied on a success campaign we thought it has enough signals enough data points performance points of me it can try to start optimizing as such so these are some of the best practices you can you can go ahead and download the best practices checklist so we currently have the the audience maximization this practice is the ones I shared with you along with the write tough explaining as to what in detail it is specifically that you can download from this particular link using this particular link and I mean that stuff’s what I had in my presentation we have this month is we are celebrating the modern search month because we have had a first-hand experience of success that AI is being doubled entre Drive obviously if you’re a builder of directed in the right way and we really want to share all of these cases and the best practices with you so tomorrow my my colleague and our director of paid media Manish Mohan would be presenting on the other modules which is more of the automation the smart feeding component the creative excellence a measurement on May 17th Manish and I will be attending the Google marketing live event in San Francisco so we’ll be coming up with the T object that has been historically one of our most most well attended and highly attended events because this is like fresh from the press we are the first ones to come out with her come out with a webinar on that and last but not the least on the main 30th we we have our friends from being we are a bring bring select partner as well and we have someone from being an expert from being giant Manisha’s with to conduct this webinar as well but with that let me let me open it up for any questions that you may have feel free to sort of added in your text box I know we have I have substantially more what from the time but we felt that this is such an such an applicable area and something that we have realized so much success we needed to share these best practices with all of you so that you can’t really be as successful if not more thank you all right it looks like our first question comes from a knight Renee I would like to know with all these new ways artificial intelligence impacts my search marketing what qualities or skills do you recommend when looking to hire marketers for my team question winning the know the first quality I think they’re definitely the marketer the fundamental strategic skill set has to be very strong so when you talk about the strategic skill set I’m not talking about the fundamentals of marketing again let us just cut through all the the the paid search or the search marketing every minute count so person should be able to understand the fourth piece of marketing for your business very well she should be able to understand as to what are the goals and objectives for your business she should really have an experience in running campaigns in your industry or your category if not sort of running your website but since you’re talking about hiring she probably wouldn’t have experience on your website maybe a comparable website Oh sir because that experiential value would really help someone to build that level of judgment and that competence which can direct experience you as I mentioned I mean it can work both if someone doesn’t have the experiential value and just are trying to impose something then the results can be dramatically that and at the same time when you are you know the stuff have experience on this trafficking one the second part which I feel is not only for probably every market or every business person the only only formula of success I emphasize this word only formula because that’s how we will do – mental power or the value of your judgment improves is through a lie but by really being a testing champion at heart you constantly need to be creative enough to identify different testing hypotheses and testing problems and constantly keep on testing those problems as such and then recording the problems learning from it and hydrating something and then sort of moving on so I think those are the fundamental for something who has understanding of your business categories fundamental thing which they would not have so that they are able to guide layer in the right direction and secondly someone who is a tester or an experimenter at heart who is constantly the searching of these numbers the different points and finding out the inside because the goal of AI is basically AI is going to serve you and when I say you you are the master only when you are able to guide it down the right path otherwise it can be a massive disservice to you as well so from that perspective only way for you to expand your horizon is through testing and experimentation learning yeah okay we have one more question this question is from Caitlin Caitlin would like to know what are some limitations to the audience targeting options available today secondly I think that’s a great question Katie lemon has been constantly expanding the audience are getting options there are some some limitations which are based on purely the number of modules which are available so one of the key things which probably would make or even in that tribution component we are seeing that the customer path to purchase is not really linear fight so there is something on the mobile they pass as I mentioned 2.2 devices and so on so unless and until you are logged in and Google or someone has been able not to figure out that this is the same user who is using the multiple devices in many cases it probably Google algorithms are still reacting or responding or working on very selective data points on this until you are able to connect together to different device points and the journeys within that it becomes a selective match but even selective that is very powerful as you can see in this situation I said upsetting one of the gaps connecting the different devices in the customer journey around this thing the second part is something which is which is a challenge I think for every every industry and every marketer are purely the privacy norms which have been which have been sort of blocked right so there is only so much information you can find out about a part of your customer customer preferences which is good because you don’t really want any any but I think to be so creepy that is sort of and then they are constantly tracking you but I think then you the privacy norms are pretty restrictive the third component is which is more of inadvertent I would say many years probably it is important in many cases people attempt to refresh our removes a clean of the cookies and if you are cleaning the cookies are using an ad blocker there is no way for you to collect or for Google to collect all this information about the audience and so as a result of it when I talk about addressable universe it’s important to keep in mind we are talking about only a subset and you aren’t really including any three of these situations that I just mentioned which is the cross device connectivity you’re not talking about let us say a cookie removal an ad blocker and the privacy norm sort of impacted parts in the in the overall mix its pristine not smaller than the flowers alright and this concludes our first day of modern search month we hope you enjoyed it and found it useful to resolving your real-world solutions have a great day and we look forward to seeing you back here tomorrow take care.