AI: How to get from hot mess to business value
- Posted on March 14, 2018
- Estimated reading time 4 minutes
If you’ve ever had to call into a call center for service on just about anything, you inevitably encountered an interactive voice recognition platform of some sort. What most people don’t realize is that there is a lot going on behind the scenes of these systems and that they are one of the most common forms of Artificial intelligence being applied today. Another might be your interactive chat session on a company’s website which might be driven by a chatbot or automated system answering your questions. Today, 66% of C-level executives and IT/business decision-makers already use artificial intelligence, such as chatbots or virtual assistants to improve user experience.
If we look at how these systems work, they all baseline off a database or dictionary if you will, of the keywords that people will most likely be asking questions about. Your questions align to a series of categories, sub categories and specific answers. But the key to it all is creating the database of record, creating the data pointers behind the scenes or the data map.
Now let’s apply the use of artificial intelligence in its multiple forms to a larger enterprise scenario, across departments, businesses, multiple locations, products etc. It’s easy to imagine the spaghetti string of different business systems, sensor data, customer data, product data. Then add on top of that less categorized information like knowledge repositories and document management systems. If your organization is like most, it’s a hot mess.
AI starts with managing your data estate
As you probably already know, making sense of all that data is a huge challenge. Over a third of businesses know it, and say that the inability to analyze customer data due to disparate data sets, data skills, systems and tools is an obstacle to creating a good customer experience. Before you can even start analyzing or automating for personalization, you need to ensure your data - structured or unstructured - is in good shape, consistent and ready to be queried. Technology and processes are only as good as the data they feed on. Especially with AI where you’re no longer relying on human judgement, but on a machine, that can only learn from and act on what’s put in front of it. Essentially, 80% - 90% of the AI and analytics is in data wrangling – reshaping the data sets so that algorithms can detect and learn from them
So what’s the answer? Getting your data into shape is much like trying to change your diet, you can take multiple approaches – you can dive deep and do an elimination diet or re-structure your data all at once. This is a huge project and often can take on an entire life on its own.
You can take a progressive, iterative approach, adding water to your diet, exercise and upping your activity. Similarly, you can take an iterative approach with building your data map and effectively prioritizing your priority data that is most relevant to the questions your business needs answers to.
You can even calorie count down to the individual food level of your diet. And, yes, you can train your data to create the map -however, most organizations lack the data scientists to tune the algorithms to get there.
Six steps to get ready for AI
At Avanade, we work with clients every day and recommend a roughly six stage approach to iteratively building your data map to be ready to take advantage of artificial intelligence applications today and tomorrow. We combine design-led thinking to your data and an agile approach to building your business questions and testing your hypothesis. This allows you to answer your most important questions first and provide tangible business value back to the business. It not only helps you build your business case for continuing forward but it makes you truly ready to take advantage of data for artificial intelligence in the future.