 Hello, and welcome to this session in which we will discuss audit data analytics, which is known as ADA or ADAs. This topic is relatively new on the CPA exam. New means it must be relevant, and that's why the AI CPA decided to add this topic. In other words, as a CPA candidate, it's very important that you understand the mechanics behind audit data analytics. This is what I will do in this session. In this session, I will discuss the five steps in audit data analytics. So if you are an accounting student or a CPA candidate, because we are starting to teach this topic at a college level, you take a look at my website farhatlectures.com. I don't replace your CPA review course. Your CPA review course may cover the same topic, but I'm going to provide alternative or useful explanation to that topic. It may supplement what your course is providing you. It may help you understand your CPA review course better, which in turn will help you pass the exam. Your risk is one month of subscription. Your potential gain is passing the exam. If not for anything, take a look at my website to find out how well or not well your university doing on the CPA exam. This is a list of all my courses, including auditing, advanced accounting, governmental accounting, my most popular taxation. Actually, the most popular is intermediate accounting, not governmental. Also, my CPA supplemental resources is aligned with your Becker, Roger, Wiley or Gleam. So it's very easy to go back and forth between my resources and your CPA review course. And I do have all previously released AI CPA questions where you can see actual questions that appeared on the exam before. If you have not connected with me on LinkedIn, please do so. Take a look at my LinkedIn recommendation like this recording. Share it with other connect with me on Instagram, Facebook, Twitter and Reddit. So let's start by looking at the definition of the audit data analytics as defined by guess who the AI CPA. Well, how did they define it? It's the science and art. Notice it's a science and art of discovering and analyzing pattern, identifying anomalies and extracting other useful information and data. Underlying or related to the subject matter of an audit. So we're dealing with an audit through analyzing analysis, modeling, visualization for what purpose, the purpose of planning the audit and performing the audit. So ADA could be used in the planning, could be used in the performing and could be used in the conclusion. Simply put, we're going to be using data, analyze the data to serve us. And what's the idea? What's, what's, what, for what purpose to plan and perform the audit? Hopefully we'll find some anomalies, some red flags, some useful information in that process. There are five steps in the audit data analytics. And those five steps are planning the ADA, accessing and preparing the data, relevance and reliability of data, performing the data, performing the audit data analytics and evaluating the results. And basically those steps are from the AICPA guidelines. And obviously, once, if you know me, once we have specific steps, like five steps, guess what? I'm going to go through, I'm going to go through each step and explain what's in each step. And on the exam, you need to know what goes under each step. Now some of these steps, for example, step one, we're going to have another session that's going to, we're going to be breaking step one a little bit further. You're going to see why. But the point is, you need to know what goes where and what each steps entails. Starting with step one, step one, basically, in this step, we have to define what is the objective or the purpose of the audit. Not of the audit, of the audit data analytics. And there are basically four. It could be done for as a risk assessment procedure to learn about, lean, learn about the entity and its environment. It could be for the test of control. It could be done as substantive or analytical procedures to detect the statements or it could be for, it could be for evaluating conclusion. And guess what? These four steps will be discussed later, a little bit more in detail. But you need to know that those four steps are within step one. Basically, what is the objective? Why are we doing audit data analytics? For what purpose? Okay. So we're looking at what financial statements, item, accounts or disclosure are related or related assertions are being audited. We need to identify what's the purpose. In here, don't assume that one type of audit, the audit data analytics fits all. Okay. Here you have to use your judgment. You have to use your business acumen. For example, inventory for retailer will be different than inventory for grocery because inventory for grocery don't stay too long on the shelves. Whether the company is using perpetual or periodic inventory system, you have to take that into account. We have to select the proper methods here and the proper tools. Are we using spreadsheet, power BI, tableau, idea, what graphics are we using, what table, so on and so forth. We have to determine what data is available. And in this process and step one, basically the audit team, they will brainstorm amongst each other of what to do, what's the best way to plan this. And this is basically called brainstorming. The potential use of ADA in this step. One step one is done. Now we need to access and prepare the data. Now we know for what purpose we need this. We access the data and make a copy of the data. Now we have to make sure that the data is complete. Are we provided everything? Okay. For example, how do we know whether we are providing everything? We test this data. We want to make sure that, for example, the general ledger for inventories agree with the inventory on the balance sheet. It means we have all the data. We can look at sequence. Do we have any missing numbers and sales orders? If we have checks, any missing checks? So it may not be missing for a reason, for example, checks, but we need to know if there's anything that's missing. Do we need to clean the data at this point? We need to kind of prepare the data. We need to clean it. How do we clean it? Maybe the data, some data is presented in different format. For example, the way I write my date of birth, I'm just going to make a date, 10, 1, 19, 77. If my wife is writing her date, she would write it as 1, 10, 19, 77. Okay. She put the month first. I put, I'm sorry, she put the day first. I put the month first. So if you are dealing with an international company, my wife uses the European type of date. I use the American. I used to use the European, but over time I use the American dates. So a company might have dates from two different geographical areas. So we have to fix that. That's what cleaning the data is. Maybe we have a parent company that long time ago, not long time ago. It doesn't have to be long time ago. It was in the past, but in the past purchased a subsidiary. Well, they might have an account number for this client with the parent company in a different account number for the subsidiary. We want to make sure that it's the same customer. Therefore, we might have to make an adjustment for these type of accounts. We filter and format the data here. We consider data security and integrity, whether the data was secured. Is it missing anything? Again, it goes back to the, is the data complete or not? Here we also have to know what type of data are we working with. And we need to know the three types of data. And you need to know this for the BEC exam. I do have a whole session about the three types of data in a separate session. Which one of them is structured data. And this is the easiest data to work with. Here we are dealing with data with columns and rows that can be easily navigated. It's conformed data, relational database. A good example will be a spreadsheet. Or we could have semi-structured data, which is tagged HTML or XML data. Here there is no relational database. It's a little bit harder if you want to extract something from a website to analyze it. It's not as easy as having the data in an Excel sheet. And we have unstructured data, which is the hardest to deal with, such as stacks, audio, video. Now we need to prepare the data. And in this step, there's something called ETL. And this is something you need to know. It's called extract, transform, and load. That's what ETL stands for. And sometimes they do test this terminology because it's coming from the AI CPA. Extract is pulling the data from the sources, from sources or from a source. Harvesting the data, taking the data out. Step two is to transform the data. And this is when you clean the data if need be. This is the hardest one in the real world. It's just how do we clean the data so we can analyze it? Because if the data is not cleaned, it doesn't matter what type of analysis are we performing, garbage in, garbage out. It's not good. Then we need to load the data, which is upload the data, make sure it's in the proper format, whatever the proper format is. Excel, Tableau, whatever the format is. So make sure you know what ETL is. Make sure you know those terms. Step three, we need to take a look at the relevance and the reliability of the data. Evaluate the relevance and the reliability of the data. And we have six criteria to determine whether the data is relevant and reliable. First, the data has to be accurate. If it's not good, if we're working with bad data, it doesn't matter how good our analytics is. Three from significant error. This is, it goes with the integrity of the data. Integrity of the data. Encrypted data. If we have encrypted data, we would assume it's better than non-encrypted data. It means no one changed it. If the data is coming from strong internal control, we are more comfortable. It's supposed to be more accurate than coming from a weak internal control. If the auditor extracted the data themselves rather than someone providing the data to you, it should be more accurate. Writting obviously better than oral data. So you don't want them to give you like data and you just write it down. You want the data and writing. Original better than copy data. If you can get the original data, that's always better. That's one of the attribute of good data. Two is completeness. Complete. Complete data. Is the data complete? Are we missing anything? And this has to do again with the integrity of the data. So the accurate of the data, the accuracy of the data and the completeness of the data deals with its integrity. Make sure we review the sequence of data. If there's any sequence, chronological order or by some serial number, we want to make sure everything is there. Consistency of the data. Data as well defined and managed. The same attribute across all the data. For example, if we have an ID number for employees, we should have that ID number throughout all the data. For example, here we would look at see if we have any numbers and text fields or any text field in fields where we only should have numbers. Freshness. The data is up to date. In other words, we're dealing with the most updated data that nothing happens since we get this data because we need the data that we are dealing with up to the period that we are testing. Timeless. We need this data available when we need it. Otherwise, if we get it when we don't need it, then it's too late. Clarity and relatedness of the data. The data field are clearly defined and related to the objective of interest. Whatever we are doing, the data is helping us achieve that purpose. Step four is perform the ADA. At this point, the auditor will execute the planned application of the ADA. It could be just running a regression analysis. And here we identify what we called and address any notable items. Notable items or something called the red flag, something to pay more attention to. And by the way, when we perform this step, I'm going to discuss this a little bit further in a separate recording when we are looking at the purpose of the ADA, whether it's for the test of control. What is it's for to detect material misstatement and notable items are basically indicative signs of risk of material misstatement. They're going to give us, they're going to provide information that's useful once we know those signs of material misstatement. We're going to have more information in designing and tailoring procedures to address those risk of material misstatement. And though some of the risk may not be, may not have been identified before now, since we're on the ADA, now we can see this new risk. So it represents a higher risk than anticipated by the auditor. This is what a notable item is, something to pay attention to. What happened if we find a large number of those notable items? Well, we can do the ICPS as you can group them, apply filters, try to find common characteristic between them, find the nature and the cause because it's going to be a lot. You have to cut it down. You may not have enough time to address all of them. So some of them might be false positive and they're false positive. There's no further necessary procedures you have to perform or sometime further response is necessary. For example, let's assume we're looking to compute the net realizable value for inventory. And we're looking at any inventory that sits on the shelves more than 120 days. It's basically we consider it as it's too long. It should not be, should not be with us more than 120 days. When we run this analytics, we find out that 30% of our inventory meet that criteria. That's a lot. 30% of the inventory is a lot. Well, we looked at the inventory a little bit further and we find out that our inventory is very different. We have some slow moving inventory and we have some fast moving inventory. Think of a company like Walmart. They may sell grocery, but they may sell TVs that are very expensive or computers. So the grocery sells very quickly. The other items, they sell slower. So what we need to do, we need to factor out the slow moving inventory. Usually they should have a higher margin items and we want to make sure that they are indeed selling at a higher margin item. Then that's why they're so slow moving. The percentage, this 30% might drop to 10%. Now we are dealing with 10% of notable items. 10%, it means the other 20%, we can say that they are false positive or they don't really, they look like notable items. But when we look at their nature and cause, they're not really notable items anymore. Now we can discuss the 10% a little bit further. So it all depends on the assertion being audited, the company that you are working with, understanding of the business, the industry. This is what determines what a notable item is. At this point, you may need to revise previous risk assessment as necessary, depending on the results of the ADA. Step five is evaluating the results. Here we need to determine if the purpose was achieved. Did we achieve our purpose? We have to investigate and flag any balances. For example, if we're testing inventory turnover, what happened? Is it as expected, not as expected? Was the ADA properly planned and performed? We need to evaluate this. If not, we need to redo the whole thing. Also in the last step, we need to document. What do we need to document? First, we need to document our objective. Why did we do this? What's the purpose of this audit data analytics? The risk of material misstatements we're trying to assess. What are we trying to do here? Why? Basically, those two are related together. The sources of the data, where is the data coming from? Did we extract the data from the internet? Did they give it to us? Did they download it from their main frame? Where is it coming from? How did we generate any tables and graphs? And also have screenshots of those tables and graphs. Names and dates of people who performed the procedures. And who reviewed them? For example, MF performed the procedure, and MN reviewed it. So we need to know who did what. In the next session, what we're going to be doing is, remember at the beginning, I said, what's the objective of the purpose of ADA? And we said there are four objective and purpose. We're going to start to talk about these four objectives. Could be for risk assessment to learn about the entity, test of control, substantive or analytical procedures, and evaluating or evaluating conclusion. At the end of this recording, again, as usual, I'm going to go ahead and invite you to take a look at my website, farhatlectures.com. Once again, I don't replace your CPA review course. You invest once in your lifetime in your CPA. Throw everything on it. Take it seriously. It will pay off down the road. Good luck. Study hard. And of course, stay safe.