Data Analytics for Finance

Editor's Notes

• #3: TITLE SLIDE
• #5: Deloitte sums up the situation quite well. Shareholders, senior management and operations need more from finance than reporting the news. The need more thoughtful insights and perspectives on the opportunities for growth and unforeseen risks to operations and most importantly – they need it now. Quote: https://www2.deloitte.com/content/dam/Deloitte/global/Documents/Deloitte-Analytics/dttl-analytics-us-da-3minFinanceAnalytics.pdf AICPA https://blog.aicpa.org/2019/03/the-future-of-finance-how-to-thrive-in-the-digital-age.html#sthash.DvEFj8v0.dpbs EY – DNA of the CFO -- http://www.ey.com/gl/en/issues/managing-finance/ey-cfo-program-dna-of-the-cfo-part-3 Ventana Research. Next-Generation Business Planning Benchmark Research, 2015 Aberdeen “With predictive analytics, it becomes easier to understand the relationships between multiple drivers. New formulas can be created, since organizations with predictive analytics are 71% more likely to enable users to create reports, charts and visualizations using self-service capabilities” – Aberdeen Group Accenture 48% of CxOs are looking to automate admin and low-skill roles (Source: Unified Finance and HR: The Cloud’s New Power Partnership MIT Custom/Oracle 2017) and 40% of transactional accounting will be automated by 2020. Or focusing on entirely different, previously out-of-reach activities.
• #6: Different types of questions, transactional reporting only answers some Multiple cross-functional data sources Cross-functional processes: Order-to-cash Procure-to-pay Questions on revenue, profitability, spend, cash flow, etc. Your analytics needs will include not just transactional reporting – which tends to answer questions of the “what is” variety. But also historical analyses – what were the trends over the past months and years. Root cause analysis – why did something happen? Scenario modeling – what if we changed the price, what would happen to revenue if we includes the discounts in certain regions. Predictive and prescriptive analytics – how will the cost of raw materials change over time and with what level of certainty. In most cases, you need to bring in multiple data sources to perform your analyses. ERP data of course, but also HCM data, marketing data, 3rd party data, maybe sentiment data, competitive information, weather data and so on. As well, your processes, like order to cash, or procure to pay, span multiple functions. In order to properly analyze supplier performance, or the efficiency of your procure to pay cycle, you have to blend multiple data sources from various systems. Transactional reports provide the bulk of your day-to-day measure of the business. But, questions constantly arise that cannot be answered by transactional reports, either because the data needed to answer the question lives in multiple places (even external), or because the analytics are too computational intensive to be handled by a transactional database. Or both! Types of questions that go beyond the capabilities of transactional systems include historical analyses – looking for trends. Root cause analysis – finding the reason WHY something happened, What if or scenario modeling – predicting what might happen if one or more variables is changed. And generally anything future-looking like time-series forecasting. A few examples: Revenue Transactional: What is this quarter’s revenue for this product line Historical: What are the trends for the past 5 years Root cause: why has revenue dipped in this region but grown in this one What if: how price-sensitive are our different markets Data blending: are any weather events or logistics issues impacting revenue Example KPIs include: Sales by region Recurring Revenue Rate Average Revenue Per User (ARPU) Cost of Goods and Services Sold (COGS) Spend Transactional: What is the direct spend by commodity Historical: How has the spend changed over the past 5 years, are there difference between suppliers Root cause: Why have costs gone up in EMEA but not in APAC for this commodity What if: What would happen to total expenditure by supplier if we changed contract terms Data blending: can we identify savings opportunities if we combine data from suppliers, purchase orders, sales, inventory, and transportation? Example KPIs include: spend by commodity or category number of suppliers by commodity/ category average purchase order value total expenditure by supplier Profitability Transactional: How profitable is this product line? Historical: Has profitability changed over the past few years for this region or this group of customers? Root cause: Why is this product line less profitable today than last year? What if: If I gave a discount what effect could that have on revenue and profitability? Data blending: what effect would an increase or decrease in number of sales reps and marketing spend have on productivity? Example KPIs include: Gross Profitability EBITDA Customer Lifetime Value Cash Flow Transactional: What is our operating cash flow Historical: What are trends in our Operating Cash Flow/Net Sales ratio over the past 5 years? Root cause: why is Free Cash Flow trending down What if: What would changes to terms and conditions for paying our suppliers mean to cash flow Data blending: Example KPIs include: Net operating cash flow Depreciation Free Cash Flow (FCF)
• #7: The analytics, data sources, even people might all be different, but at the core, the problem is the same. The typical “band aid” solution to get to the needed answer involves getting a bunch of data extracts out of different siloed applications or systems, bringing those data sets into some storage tool – usually Excel. And then manually blending and analyzing the data to create the desired report or analysis. This process is slow, difficult, iterative and complex. It’s prone to human error. From a data perspective, it leads to questions about the data accuracy and raises security and governance questions. It seems like an ok workaround but it’s profoundly unsafe. Not to mention way to manual, slow, and labor-intensive problems around accessing, storing, securing, using data Problems around enriching, analyzing, predicting, trusting results Problems around time to results and time to action
• #8: <HOW> To solve those 3 aspects of the common problem, we propose 3 essential elements. First is simplified data access, to get value from all the data. Second, augmented analytics to power deeper insights and finally the ability to act faster on insights with analytics that seamlessly fit into the way you work Simplified data access - Data Worth Using Augmented Analytics - Insights Worth Developing Self-service and governed analytics to act fast - Action Worth Taking
• #9: <WHAT SLIDE>WHAT A financial data mart with Oracle’s Autonomous Data Warehouse and Oracle Analytics Cloud is the single solution. Let’s take a look at how it works. A slightly more technical look at this business managed architecture and process flow from data to decision. Starting on the left, you have any data source, whether that includes flat files or application sources. The idea is to pop-up a data mart to support your functional area, You load the data into the ADW using the OAC dataflow capability that is a click and drag approach with zero coding. Or IT can optionally manage this process and leverage the incumbent ETL tool. There is no limitation on the number of data marts or functional areas that can be supported. With elastic cloud you use as much as is required to support the business needs. Once loaded the ADW autonomously takes care of all data management tasks. Connecting OAC to the new data mart is quick and easy and in minutes users can being performing data visualization or ML supported analytics on that data. All user roles are supported regardless of their requirements. Any role, all data, on any device. For all your questions worth answering
• #10: If a picture is worth 1000 words, a video (or 7) should be worth a lot!. So in the next 4 slides, I’m going to explain how this all works, and show you at the same time. You can then try it out yourself. Throughout the rest of this presentation, we’re going to use a story about searching for the root cause of a drop in Net Income in the UK as a means to explore all the capabilities in our Financial Data Mart. The story, as with most of your analytics investigations, will have a lot of twists and turns. For now, we start with the end result. Your Finance data mart is up and running, and you’re developing and using the deep insights. You get an experience that allows you to get the information you need, when you need it, regardless of channel—desktop, mobile, or another application. Analytics should seamlessly fit into the way you work, not force you to work differently based on how the analytics product operates. So what might that look like? Audio voice over is turned off by default. ALT-U MUTES AND UNMUTES WHILE VIDEO PLAYS Play Videos (ALT-U TO MUTE OR UNMUTE WHILE PLAYING) To make informed decisions, every organization needs analytics. But to be truly effective, these analytics tools must work within—and across—interfaces to create a seamless experience that fits the way you work—personally and within your workgroup. Now let’s go back to the beginning and see how we build up to this result. Let’s start with the data. Video script – mobile phone video On the way into the office in the morning, you review Oracle Analytics data on your mobile phone. The information is automatically delivered to you based on your preferences. Or you can use voice commands to retrieve information. IN this case, you want to look at net income by month and by region. You can do some light analysis, filtering down to the UK, where there was a problem last August., share with colleagues, update the charts, review the summary information, and instruct the app to bring back the information at a time or place of your choosing, all from your phone. Video script – tablet video Intelligent search requires the ability to understand the question posed through speech, or text (using natural language query) as well as the ability to search all available datasets, and then surface the most appropriate results. You do not need to know the source of data before you search for it; Machine Language algorithms do it for you. Here you decide to drill into financial data. While it appears revenue is fairly flat, there’s a disconcerting downward trend to net income. As always, there are questions worth answering. You can do this analysis on your tablet, filtering to a region, lassoing the quarter that’s showing negative net income and keeping only that. You can save your work to continue later, once int the office. Video script – Laptop video Interactive visualization and dashboarding improves the way you can access and interact with data. For example, you can maximize a potentially interesting visualization, and with one-click analytics, add statistics like a trend line with confidence interval – which is of course adjustable – as well as in this case obvious outliers. Enhancing sophisticated, interactive visualization capabilities in an easy-to-use interface delivers more analytics power without compromising the exploration experience.
• #11: Our starting point is to build the foundation of the data mart to make data available for analytics. ADW Having timely and trustworthy data is vital for your success. That means controlling your own sharable and secure data workspace – so your team can collaborate around a shared workspace, rather than emailing and reconciling duplicate spreadsheets. It means capturing up to date data, using live data, no waiting for periodic extracts, blending it with Oracle and non-Oracle data sources. It means data that is consistent across your workgroup, so you can trust the resulting analyses, while ensuring sensitive data is available only to authorized users It means adding computing power as needed, so no more worries about underpowered CPU, if you have to crunch large amounts of data. With Oracle you can increase processing power as needed, and drop it back after to save costs. Enhanced data flows Adding data to the workspace, and preparing data for analysis is a critical element of any data and analytics supply chain. You need sophisticated transformation capabilities without having to involve professional data transformation specialists, or IT. Data flows enable these capabilities. Demo video Analytics and data that are always up to date, trustworthy and available, all independently from IT. Let’s have a quick look at how to use data flows to add a data set to your data workspace. Play Video (ALT-U TO MUTE OR UNMUTE WHILE PLAYING) Video script – Simplified Data Access video You begin by creating a connection to your secure workspace. In this case, it’s an Autonomous Data Warehouse connection. You enter your credentials, username and password. This ensure ensures sensitive data is available only to authorized users. That’s it! Once the connection is created, you’re ready to add data. Choose your data set – in this case financial data from 2018 which you want to explore for an unexplained drop in net income in the UK in 2018. You preview your data, and add it to the data flow. The data flow is how you add data to the workspace, and start preparing it for analysis. There’s lots you can do in data flows, including creating and running your own ML models, and we’re happy to dive into that with you, but for now, we’re simply going to save our data flow, and run it. You’ll give it a name, and then select your newly created database connection. Name your data flow, save it, and run it. This adds the dataset to our workspace, no need for complicated ETL magic from IT. Now go inspect the data flow. This is where you can see when it was created and modified, and by who, as well as the sources, targets, schedule and history. This is important and creating data that you can trust. Before you start your actual analysis – after all the real payoff in data management is when you use the data! – let’s also inspect our newly created dataset. Again, you can verify a wealth of information about this dataset, including that it’s certified for use. Data that is consistent and that you can trust. We can also check the data elements, whether this dataset is searchable with Intelligent Search, and, very importantly, who is allowed access: full control, read and write or read only. Ok, let’s rock and roll. Create your project with a single click from this dataset. And that’s it! It’s about 2 and half minutes – less if I didn’t talk so much – you’ve started your analysis. We’ll continue in the next section, and get to the bottom of that drop in net income.
• #12: Our goal is to power all actions with deep insights from all of your data. Oracle is committed to serving all your analytics needs, no matter how advanced—or simple. Unlike other products that require you to compromise between governed, centralized analytics, and self-service, Oracle Analytics resolves this dilemma with a single solution that incorporates machine learning (ML) and artificial intelligence (AI) into every step of the process. We are combining three powerful forces—augmented analytics, self-service analytics, and governed analytics—into a single solution that you can quickly scale across your organization and realize the greatest potential from your data. This slide and the next illustrate that combination of Augmented, self-service and governed analytics. Smart data Discovery: In any today’s dynamic business environment, getting to the right—and unbiased—answer quickly is critical. Knowing that data and processes continue to change over time, businesses need to be able to meet the demands of tomorrow. With smart data discovery, the system automatically analyzes and generates explanations to any attribute, generating facts about your data, including the drivers of the results, key segments that influence behavior, and anomalies where the data is not aligned with expected patterns. These insights can be used as a starting point for further analysis and discovery. With data-driven guidance, you can quickly get to the right answer. The goal is to rapidly deliver insights to kickstart a richer, contextual analytics experience. In this way, you can use more data and get to the right answer faster—and without bias. Interactive viz and dashboards Any data discovery capability must be easy to use, visually appealing, and enable sophisticated, dynamic analytics that can be shared with large consumer communities. Interactive visualization and dashboarding improves the way users can access and interact with data..Enhancing sophisticated, interactive visualization capabilities in an easy-to-use interface delivers more analytics power without compromising the exploration experience. Unifying the visualization and dashboarding capabilities creates a single, integrated experience. Smart data prep and blending Data preparation always takes more time than you think it will, and you can’t get to the analysis and synthesis phase until you prepare the data. Smart data preparation augments, enhances, heals, and creates richer data that can lead to improved business insights and sharper understanding. With expanded and augmented data preparation capabilities, customers will benefit from a richer, faster data analysis process. Smart recommendations can be used to improve and enhance data based on automatic data profiling and inclusion of custom-reference data to enrich data sources. These enriched data sources can be easily shared with others, giving everyone in your organization access to better data for better analysis. Oracle Machine Learning and integrated data science To predict results, better understand your data, and train models with rich datasets, you need to be able to use ML models within an analysis framework. Integration of data science and analysis into one platform enables richer insights and better predictions. Oracle Machine Learning is a SQL notebook interface for data scientists to perform machine learning in the Oracle Autonomous Data Warehouse (ADW). Oracle Analytics Cloud can use its own data flows to create models, as well as visualize the output of models created by others. You get both with this solution Demo video: Power deeper insights with embedded ML and augmented analytics. Let’s pick up our story where we last left it: a newly created, blank project, with a finance data set Play Video (ALT-U TO MUTE OR UNMUTE WHILE PLAYING) Video script – Augmented Analytics video In our last video, you had just added a Finance dataset to a secure workspace, kicked off a project, and were about to start analyzing the data to figure out what happened in the UK in August 2018. So there you are, staring at a blank canvas. Where do you start? That’s where ML-comes in handy in the form of a capability called Explain. Select Net Income and right click to Explain net income. Machine learning analyzes the data to recognize the patterns and trends in your data set to provide visual insights and enhanced statistical analysis. You can subsequently use these visual insights and statistical analysis in your project visualization canvas to interpret the data in your data set. The first tab shows basic facts about net income. We like the look of Net income by month, so select that chart and click add selected to add it to your canvas. Now you’ve got something to start with. You can begin to manipulate your visualizations to perform your analysis. You enrich this one with a trend line with confidence interval. And… here’s where a handy video editing transition occurs so you’re not watching me building out your first dashboard start to finish. I’ll just show you a few key snippets. Here you’re adding more visualization, and filtering to the UK Now you add a couple of final line charts as you suspect that Opex mught be the culprit. As you can see, the interactive ways to visualize and analyze your data are almost infinite. easy to use, visually appealing, and enable sophisticated, dynamic analytics that can be shared with large consumer communities So you can show that while revenue is ok, the culprit is operating expenses. In particular a spike in T&E from the sales cost center. Which is great, but of course the next question (there’s always a next question) is why was there a spike in T&E? And that answer isn’t in this data set. As always happens, the answer lies in blending of data from different systems.   Not a problem. Navigate over to prepare. Click add data and find a payroll dataset in your secure workspace. And add to project. You can verify that the two datasets are automatically linked across relevant common attributes. You add a second dataset of T&E data, of course intelligent search would allow you to search through all the data to which you have secure access to see what might be useful to your analysis. This second dataset is also automatically linked to your two other datasets. Let’s use some of the smart data preparation capabilities to enrich this data. Select the payroll dataset. You see a preview of the data and to the right the recommended enrichments. I have to pause for a second here to highlight this. What just happened here is the data set was profiled to produce a set of recommendations to repair or enrich your data. Machine learning is the basis of these automatically generated recommendations. For example, it might see a credit card number and recommend obfuscating it. Or a city, or country and provide the population. In this demo you decide to extract the name of the month from the date field. Apply the script and now let’s go back to our visualize tab You’ve spent a few minutes creating a new dashboard, which you’ve name UK Salary and Expense analysis. You’ve created custom calculations, such as that used in the Variance month chart, to develop deeper and richer insights You finish building out this particular analysis by adding a couple more visualizations. The variance by month chart uses a calculation that you custom built, since variance was not a mesure that existed in the dataset. The Out-of-Policy line chart completes the story.completes the story. You notice something unusual on the base salary and overtime chart, but decide to come back to that mystery a little later. Right now, you maximize the Out of policy expense chart, and add the cost center. You can confirm that people in the sales cost center did pay for a large number of out of policy hotel during that quarter, especially in August. In the next video, you’ll put together a report and recommendation to follow up on that. And you’ll also dig into a new, and unexpected mystery – did you catch it on the base salary vs OT costs? Stay tuned.
• #13: To act quickly on insights, you need to use all three capabilities - augmented analytics, self-service analytics, and governed analytics – as one single solution. The systems must adapt to the way you work, not the other way around. You also see these capabilities shine throughout the mobile experience, as shown previously. Intelligent search In order to make analytics and data available to everyone, systems must adapt to the way you work, not the other way around. With intelligent search, you can easily find the right content—by searching via text or speech. By removing IT bottlenecks and delivering results faster, intelligent search makes all of your data accessible to everyone. It allows you to find answers to what you’re looking for faster—and with greater ease. Interactive vis and dashboards Any data discovery capability must be easy to use, visually appealing, and enable sophisticated, dynamic analytics that can be shared with large consumer communities. We continue to enhance the interactive visualization and dashboard capabilities of our analytics, with the goal of a unified environment supporting both discovery modes. Businesses will have stable, repeatable, analytics and new, agile, visualizations—all in one interface. Unifying the visualization and dashboarding capabilities creates a single, integrated experience. Experience continuity To make informed decisions, every organization needs analytics. But to be truly effective, these analytics tools must work within—and across—interfaces to create a seamless experience that fits the way you work—personally and within your workgroup. Smart collaboration To expand the use of data for generating insights, organizations need to be able to easily share and collaborate on analytics content. Providing both structured and unstructured ways to collaborate across all analytics activities builds community and consistency for both agile and governed types of analyses.By harnessing the collective wisdom of everyone in your organization you can drive the sharpest insights, leading to best actions and optimal outcomes. Demo video: Act Faster on Insights with Analytics that Seamlessly Fit Into the Way You Work. Let’s wrap up our story with sharing analysis results and smart collaboration. Play Video (ALT-U TO MUTE OR UNMUTE WHILE PLAYING) Video Script – Self-service video You just confirmed that at least some of the spike in opex spending was related to expensing out-of-policy hotel rooms by the sales cost center. You want to share this so that sales management can take action. You create a narrated story. Click the Narrate tab. Select the canvas that represents your analysis results and drag to the bottom panel.. you update the page title to represent your findings. You also add a note to highlight the out-of-policy spend, format it, and drag it to the relevant spot ont eh canvas Click the Share icon and save your story asan image. You can send that to the sales managers, and to your boss. Now, you remember that mysterious blip in one of the charts. Back to Visualize. Maximise the salary vs overtime chart for a better view. Very odd. Why would base salary average drop off like that, while overtime goes up? You can only guess that experienced staff has left. But why? You decide to loop in your colleague in HR to get to the bottom of this. You create an image to share and save the project in shared folders so she can log in and work with you on this. You click save as, navigate to your shared folders, and create a new folder called Finance HR collab. You inpsect the new folder’s properties and see where you would add access permissions. There will be both HR and finance data, so the sensitive info needs to be secure You slack your HR colleague the image and a request for help. She logs in to the shared project and quickly adds HR data. This new dataset is automatically joined to the others, thank you machine learning and smart data prep. She also reviews the machine learning generated recommendations before adding the data to the project. All 4 datasets are in the project. She adds a canvas and gets to work analyzing the data to answer your question. The HR dataset contains a measure of volunteary turnover. She uses Explain to get started choosing a bar chart showing volunteary turnover by month. Refining that visualization, she sees that the call center lost 22 people in a single month. She decides to dig deeper to understand why. We rejoin her having built out most of a dashboard digging into this question Let’s add one more chart, using intelligent search to find the attributes and measures. Change the chart type from bar to tag cloud so we can better see the reasons given for voluntary turnover. Filtering to that month where 22 people left, she sees that most left for higher pay rate. And so it goes. Another question, another analysis. It never really ends does it? You’ll always have more questions worth answering. And oracle analytics will always be there. We are committed to serving all your analytics needs with a single solution that incorporates machine learning (ML) and artificial intelligence (AI) into every step of the process. We are combining three powerful forces—augmented analytics, self-service analytics, and governed analytics—into a single solution that you can quickly scale across your organization and realize the greatest potential from your data.
• #14: OPTIONAL SLIDE if the customer has Oracle SaaS Data is the lifeblood of any analytics system. Access to data, regardless of the source is paramount. Native access to more data enables richer, more diverse analytics. Oracle Applications Connector supports several Oracle SaaS Applications. You can also use Oracle Applications Connector to connect to your on-premises Oracle BI Enterprise Edition deployments (if patched to an appropriate level) and another Oracle Analytics Cloud service. With smart connectors, you will direct connect to Oracle SaaS, inherit security from Oracle SaaS, and combine real-time and transactional from your applications Oracle applications connectors: https://docs.oracle.com/en/cloud/paas/analytics-cloud/acubi/oracle-applications-connector-support.html Supported data sources: https://docs.oracle.com/en/cloud/paas/analytics-cloud/acubi/supported-data-sources.html Demo video: Connect to a wide range of data sources Play Video (ALT-U TO MUTE OR UNMUTE WHILE PLAYING Video Script – SaaS Connect – optional for use if customer has Oracle SaaS and wants to use smart connector You begin by creating a connection, selecting connection type Oracle Applications. You give the connection a name, and enter your credentials, including username and password. This connection will inherit security from the SaaS application. That’s it! Once the connection is created, you’re ready to add data. Click create data flow. Add a data set… from your recently created connection to ERP cloud. You want to do some analysisis on supplier spend. so you go find that folder and analyses. You select your data. Click to preview the data. Visually check that these are the data you’re looking for, and click add. You could add any number of steps to the data flow, but for now, you’re going to save the data. .Give it a name Choose your data storage. In this case you want to add it to your autonomous data wahrehouse. Save your data flow and run it. That’s it! You’ve just used a smart connector to connect to an Oracle application and added the data you wanted to your secure workspace, ready for analysis. You can immediately create a project. You’re brought into the Visualize tab, with canvas and your dataset. Youu know nothing about this dataset and are staring at a blank canvas. But we can fix that. To get started, since you’re interested in how much is spent on different suppliers, choose the Supplier attribute and click Explain Machine learning analyzes the data to recognize the patterns and trends in your data set to provide visual insights and enhanced statistical analysis. You can subsequently use these visual insights and statistical analysis in your project visualization canvas to interpret the data in your data set. You select a couple of the charts that explain Supplier and click Add selected. This now gives you a starting point for your supplier spend analysis. And took about 2 minutes.
• #15: NHS http://www.oracle.com/us/technologies/big-data/big-data-insights-3073862.pdf Outfront https://blogs.oracle.com/oraclemagazine/analytics-for-business KFC Netherlands https://www.mt.nl/bijlagen/cloud/hoe-kfc-de-klant-nog-slimmer-gaat-bedienen/567854
• #16: CTA
• #17: https://www.mt.nl/bijlagen/cloud/hoe-kfc-de-klant-nog-slimmer-gaat-bedienen/567854
• #18: https://www.mt.nl/bijlagen/cloud/hoe-kfc-de-klant-nog-slimmer-gaat-bedienen/567854