AI-driven Analytics | Retail- Logistics and BPM | Intellicus Dashboard Demo

in this video we would demonstrate how you can leverage intellica's ai-powered analytics to interact with your data monitor kpis and make data-driven decisions a very good morning to everyone my name is vikas and i'm going to demonstrate the product capability to you while i'm going to do the product demo just wanted to reiterate we'd spoken about the business user layer and we've spoken about the data sources so you know imagine data coming in from varied sources which are you know your erp platforms traditional data sources flag file systems how is it that you can combine all the data and bring a meaningful interactive report for your business user to consume and for him to understand how those kpis are behaving and correlate and able to make decisions with the data in today's uncertain times that we're going through this plays a very important role and data management is the near arch that we have at this point now starting off with the product demonstration here now one of the important aspect that we're going to talk about is going to be the high speed analytics report how a business user can actually interact with those reports correlate and understand different kpis now here we're talking about inventory management now inventory management is one of the biggest ask that we have in today's scenario where you would want to know what's there in stock that you have how is it that has been managed you know bringing in all kpis that are relevant and important for the business to understand here we're talking about a data for a period of three years uh where we have you know inventory which has been ordered which is about 161 000 inventory and then the bifurcation of that 161 000 into completed decline canceled and suspected now these are three years of data now for example you know you would want to look at one year of data we can look at those drill downs to happen but first let me just quickly run you through all the widgets here and then i'll be doing the slice and dice and correlations as well for you now in the second widget here you see the order versus complete tab here which is a good chart where you see three units of data now this each of these pi basically reflects the number of orders that have been received and how many orders have got completed so for 2020 for an example you see total orders that have been received is 38 730 completed at 30 643 so you see the bigger pie here that is the ordered and the smaller pie here reflects the completed then the next category is about the product categories what are the different products that have been ordered out of this 161 000 that has been placed then you have the shipping status of the product where you know what is the shipping status how many late shippings have happened how many advanced shippings have happened uh shipping on time how many shippings were cancelled then the delivery status for those shippings then we have this pictorial graph right and in this pictorial graph we've used you know relevant to the type of use case but in case if there is any specific pictorial graphs that you would want to use specific to your business we can bring in that as well here we're talking about top five products and what is the contribution we have with those top five products and then we have the order count and late delivery which talks about this is a bar graph it's a stacked bar graph which has two measures in it one is what is total order and out of the total order what is the late delivery we had and then we have a rag analysis now this is a map for the us market where you see regions divided into west south midwest and northeast where you're able to understand the number of orders that each of those regions have have you know given so west has 60 000 orders versus northeast which has 17 000 orders and midwest which has 28 000 orders and then further below you're able to look at the number of orders that have been given on a month-on-month basis and how they've been trending in terms of their in in terms of the overall uh comparison versus the previous month so in most of the months you would see there are ups and downs and specifically from january 2020 to march 2020 last year where we had experienced the pandemic scenario you know we saw those in that impact being there in terms of business across uh and then thereafter it got settled and started moving up now coming to specific uh you know giving you capability in terms of how we can interact with the dashboards for example you would want to look at a 2020 here i can just quickly put a filter and i would know how 2020 numbers look like and what are the different types of product categories so all my widgets actually would get refreshed and give me into the understanding for 2020. now imagine you know this data is coming in from different sources so all of those sources on the runtime would actually give you the information specific to the selection that a business user would do and that's something which is readily available for a business user to consume now if i would want to now go ahead and do a drill down on this i can do a drill down on 2020 and i can look at a quarter wise performance so for example i want to filter on a quarter i can then look at a quarter level performance from a quarter level performance i can go down to a month level performance within that quarter and i can actually filter at a month level and understand how did that month behave now let's look at it you know from a year to a quarter to a month i can get to a day level as well right um or within the day so last level of granularity in a daytime hierarchy can be easily achieved depending on the data granularity that is available and a business user can consume that level of granularity to interact on the dashboard now on september in the year 2020 we see that uh total orders which were placed with 3257 and then the bifurcation of those 303 257 orders which is into completed decline cancelled and suspected now if i were to look at it what is a different product category so for example campaign and hiking had 575 water sports at 965 so this looks to be the most highly ordered product so you know i can just go ahead and put a filter here on that product and all of my uh details would actually give me the understanding of that 965 orders so you know that's that's another aspect that i can look at now in in this particular use case you're looking at 628 are the late shippings that you had shipping cancelled is 262 advance shipping and ordered which is 204 and out of the overall orders that you had 417 were actually early deliveries now you had 359 late deliveries now let's understand why these late deliveries and what is the impact how many orders got cancelled on because of these late deliveries so i can just put a filter here and i know decline were 40 315 were completed and four are suspected to be declined now this is how a business user can understand what really has happened to each of those action labels and is able to understand the the impact and which category do they fall into now they can be multiple such kpis that you can actually bring in now let me just quickly remove all of these filters and then uh take you to the pictorial graph if you go to the pictorial graph here on the top five products we're talking about you know these five products so for example if i would want to look at say the girth gloves so i would just put a filter here and i'll be able to look at the golf clubs so here we are on on the top five products so if golf globes you know in case if i would want to quickly filter on golf clubs and understand how did that behave so i can look at that data and understand how did that data do for the overall three years so i had 21 000 orders and so on and so forth in terms of my overall so this is the product category that we had actually selected now in this particular product category on a month on month how did we perform in terms of delay overall delivery so i see that on a month on month i see there are late deliveries that have happened now there are late deliveries that have happened in the month of july which is the maximum so for example i want to drill down and i want to look at a monthly day level data i can do that and i can select a specific day and understand how did that area do in this particular bar graph you see that the bigger bar graph is the number of orders and then you have a dark blue smaller bar graph inside which is called the stack chart and that's the late delivery account that we have now if i would want to just go ahead and give you the understanding of those 69 orders business user can actually look at those 69 orders and understand how does that happen i'm just going to clear all the filters again and quickly now move on to the lag analysis now on the rack analysis we're talking about the us market here now if you look at the us market here we're talking about different regions right say for example midwest west and and also you want to look at a ytd you just click on it and you would have a yearly number available you would want them to be compared to different years you can look at comparing them with different years you can just go ahead and do a selection and the moment you select one so if i it is of 2020 it will actually include 2019 and will give you two years of numbers now if i remind if i just go ahead and minimize this you see two years here which is 2019 and 20 and the entire rag analysis gives you that view and the entire data available would actually give you that view let me just quickly go ahead and give the understanding of the order trend here now in the order trend here you see that in the year 2019 and 2020 there are certain highs and lows so the graph can actually give the understanding to a business user as to how you know the orders have been trending compared to the previous month so for example from the month of january to the month of february there's a growth of about 13 versus from the month of feb to the month of march there is a decline which is there for 43 and we know that we had actually gone through this phase which was in the pandemic where we had seen the downsize trend now if you want to look at that for a specific product segment you can just go ahead and click on that product segment it will give you that information as well you want to look at a specific kind of a region you can click on that region and you can find details for that region so a business user can actually look at multiple slice and dice of dimensions and understand how those dimensions are impacting the measures and how these kpis can impact him in terms of its overall decision a business user can actually correlate kpis coming in from different sources data coming in from different sources bringing it on to the report designing layer at the dashboard level and interact with those kpis now i will quickly move on to a next important aspect of our interactions that we generally do and that's the customer interaction so in in today's scenario customer interaction and customer experience is one of the most important factor for us to sustain our business and to sustain our brand and to ensure that we've got loyalties around so you know it's important for you to know and understand what's really happening there now in a scenario where you were actually in a in an environment where you are attending customer calls or where you're in you know attending customer emails or you are actually into a you know platform where you chat with a customer so all of those transactions can be monitored all of those important kpis can be monitored and they can be brought into a visualization layer for a user to understand how those interactions have been now here in this scenario we're talking about a scenario where you're receiving customer calls now a customer call can be for any nature and what are the important factors you should know how many calls that you've been receiving how many calls you've been able to answer what is the average handle time that you're spending so average handle time is the total time that you spend on on a call so that you know what is the total time that you've spent to manage the overall calls that have been given to you for the day for the hour for the week for the month or for the year now here we're actually talking about two years of data so which is about 19.04 million transactions uh 18.72 million have been answered the average handle time per call that has been about two minutes and one seconds the service level which means that how many calls i've answered for example you know the definition says that i have to answer 90 of the calls within 20 seconds that means the ring should not go beyond 20 seconds and i have to answer the call within that time so that the customer has limited wait time now that's one of the matrix that we've brought in here then you've got average speed to answer which is 12 seconds then the service level variance paul answered how many calls were answered out of the total which were offered and how many were actually abandoned so combination of this would be actually a hundred percent now uh if i were to go ahead and quickly look at filtering on 2018 you know i would know how 2018 performance has been i can further drill down on 2018 i can look at a month level performance in this particular bar graph chart you see the service level as a bar graph here and then you have a service level variance which is a dotted red dots in the bottom below now why they are at zero percent right now is because they've been able to meet the target that has been defined now for example i want to drill down further i want to look at a weak level data from a week i can drill down further and i can look at a day level data i can filter it down another day and understand how did that day perform so a business user can actually look at interact with the dashboard and understand how they have been able to perform each of those kpis have been able to perform now for example i want to look at from a day level to an interval level data now here in this particular graph chart you see that it's 0 to 24 now this is half hourly interval data that is available now that's the kind of granularity where they would know what has happened on each of those transactions that have come in at a interval level at a half an hour interval level right and then other attributes which are important for business to understand which is in terms of the total talk time that i've spent on a call now why do you see these lines as red is also because they have not been they've been outside the uh thresholds that have been decided uh that have been defined for this business so here you see fourth of december none of the materials are are met you know the service level average speed ones or sla variants and so on so forth have not met they're all red and and subsequently you see below is also the calls offered number of calls that have been offered how many calls have been answered within those intervals average speed to answer it is a red line why because it has been above the threshold that has been defined average handle time has also been above the threshold that has been defined hence it is all red now because of these matrices not in order we've not been able to meet the service level right now we've understood how is it that i would want to look at in terms of a product level management at a supply chain we've looked at how is it that i would want to manage a customer interaction or a customer experience in a performance insight now i'm going to move on to another use case which is going to be a machine learning use case now this is the wfm prediction use case now in this use case we would basically be demonstrating how a machine learning model can help you plan more efficiently in terms of a manpower requirement or you can actually think about a scenario where you would want to know what are the different types of stocks that you need to order for your inventory based on the past sales that you've had you would want to know what type of customers you want to onboard you would want to know how would your revenue look like for the next couple of years think about all of those scenarios that can be related to your type of business and and you can actually look at building that kind of capability within your enterprise that based on the historical data that is available with you and based on the transactions that have happened and based on external factors which can be seasonality festivity any other external factors customer sentiments you're able to give that learning to the machine learning use case and you're able to predict and empower yourself to understand how your future is going to be look in terms of the kpis that you want to monitor so here we're actually talking about a manpower prediction use case now in this use case we're actually talking about the data for a period of about three months we'd actually used data for three months and we predicted the data for about eight eight days here now in the first section that you see is the forecast now this forecast is basically a number that was achieved through a very calculated approach when i say calculated approach it means that it had actually looked into the growth factor and actually given that numbers of this number was given by the business that they anticipate about 25 000 calls that they were receiving for the for the particular month for those number of eight days but when they actually received calls they realized it was only 10 231 calls so the variance between the forecast and the offered was 150 which is the deviation percentage forecast here now when we utilize the historical data we understood that the variance between the forecast and offered is huge now we need to go ahead and bridge this gap and that's where a machine learning model can actually help you to be more realistic to the scenarios to situations and give you more accurate numbers so the product has a capability to be integrated with r in python scripts and use industry level algorithms to provide you the output that you're looking at for your business you can test different machine learning algorithms models within intelligence and you can understand how those models are behaving and you can utilize those outputs to actually understand the behavior of your business and how effectively you can plan them now we've got calls prediction here now this particular kpi was brought in by a machine learning algorithm which we had used random forest here uh based on the historical data that was available of the call arrival the model predicted that you can anticipate about twelve thousand one hundred thirty four calls which was more nearer to ten thousand to one and thirty one which was the actual offered hence the deviation of the model of the machine learning model was only about 18.60 therefore it saved about 159 man power for the business now that's a huge number to save if you look at the variance here you know i would just show you the variance of the offered and the prediction stuff and so you look at the variance this is the forecast staff gun which is 25428 and that's the upper line uh offered staff count which is 10 231 which is the below line and here offered staff count which is the below line and prediction staff count which is the upper line here now you look at the gaps in between both so if you were to plan more effectively basis the model that you currently anticipate the business to behave uh you know you may be in a scenario that you would look at certain losses to be booked versus if you were to look at a machine learning model where it can understand how those data trends have been impacting you it will be more realistic for you to look at the future planning so that's how it can help you all right so you know you've gone through the stock management you know what what's gonna happen what's happening in the past in your business how they've been behaving you know how to go ahead and look at different types of kpis to go ahead and monitor your business you know how you need to go ahead and look at uh planning in terms of machine learning use case now what is the output that you want to look at right that's another aspect that you would want to evaluate now you've done your planning you looked at your historical data all of that is in order now what to expect that's another use case that you can look at and that's going to be uh you know the most important factor for you now in this case we're actually talking about a service level prediction where actually if you remember in the second you know the the customer experience interactive dashboard we'd actually spoken about the interactions of the customer whether they're actually being answered at the right time or not whether you have the adequate band part to achieve that or not and and subsequent other kpis which were important so in the similar case we're actually relating the uh what if analytics use case now into customer experience dashboard there now here we're actually talking about the data for a period of uh you know 31 days i'm just going to go ahead and talk about the selection here so this selection that you see is basically the the solid bar graphs which is the historical data and then you have the dotted bar graph and the line graphs here which is the predicted data that you have now let me quickly tell you you know the service level is a bar graph here asian count is a blue line graph and calls offered is a yellow line graph here now the model had actually used the data for about 31 days and predicted you know how many calls you can you know you can expect for the remaining 15 days what is the count of agents that you need to answer those many number of calls and what is the output that you can get so it says that i will not be able to meet the service level hence my all bar graphs are in red right now i can get on to filters and a what if scenario is a scenario which empowers the business user to create you know different type of scenarios what if i do this or what if i do that what is the kind of combination i can look at to achieve what i want to achieve now here in this use case we're talking about volume right so in order to attend the volume i need resources to be there now i'm not able to meet the target basis the volume that i have now what is it that i can do i can just go ahead and improve the agent counts for example i increase the agent currency by 40 percent now by the way when i increase the agent count my average speed to answer would also get reduced and why they will get reduced is because they are correlated with each other more the people higher the efficiency i'll just click on apply and this would actually give me an understanding as to what output that i get so i it says that i'll be able to manage 19 000 calls but i will still not be able to meet the service level now i can go back to the water scenario here and for example i just reduce the average speed to answer by 25 and let's see what happens it says that in most of the scenarios i'll be able to meet the service level so now eighteen thousand calls i need to attend i will be able to answer seventeen thousand calls average speed to answer target would be ten percent ten seconds agent count will be 133 ht will be 131 and my output is going to be green which is at 77. now that's how it can empower you or the business user to understand what type of target he wants to keep for all those variables that can impact you to achieve what you plan to achieve that's the kind of power that the business user can get thank you for watching for more information on how analytics can help your business visit intellicus.com or mail us at info intellicus.com

Channel: Intellicus Tech

Views: 440,806

Keywords: business intelligence, data analytics, data science, business dashboards, big data, machine learning

Id: MOTwkcw3c1A

Length: 23min 55sec (1435 seconds)

Published: Tue Jul 20 2021