Customer relationship management needs to be a system of intelligence, not a record of who did what weeks ago.
Microsoft’s Dynamics 365 is a CRM service with plenty of AI-powered tools that you can extend and customise for specific industries. So why is Microsoft working with C3.ai — an AI services company run by Tom Siebel whose eponymous salesforce automation company established much of what we think of as customer relationship management — and Adobe to build an ‘AI-first’ CRM?
Although Microsoft builds its own tools, like Customer Insights, on top of Dynamics, at the end of 2019 it started up ISV Connect, a partner program for ISVs like C3.ai to use Dynamics and Power Apps as a platform instead of building their own SaaS applications. C3 AI CRM promises to combine Dynamics with what C3.ai knows about some key industry verticals, without every customer having to tailor their own solution.
“For Dynamics 365, we offer a collection of modern interoperable and individually adoptable SaaS applications, and we build these applications for departments and lines of business,” global director for Dynamics 365 Dina Apostolou told TechRepublic.
“Our products are finished SaaS applications: we take them to a certain point in terms of helping organisations be able to use them out of the box, but also to customise and tailor them for their specific business workflow and their needs.”
“What C3.ai has are the industry models and templates so they can do the customisation for an industry, instead of it being a custom solution every customer needs to invest time and resources in. They have a long-standing history of large deployments at an industry level with specific extensions and models.” C3.ai customers are large organisations like 3M, Shell and the US Department of Defence.
The combination, Apostolou said, is “industry solutions that are powered by Dynamics, that are off-the-shelf, and that start to work out of the box.” That’s especially important these days: “Typical CRM implementations that were known for months of implementation have happened in literally days.”
C3 AI CRM has industry-specific CRM solutions for telecoms, financial services like banking and insurance, oil and gas, utilities, and manufacturing, C3.ai CTO Ed Abbo told TechRepublic. These solutions are preconfigured for handling not just general sales and marketing, customer service and churn prediction, but also specific tasks like wealth management or corporate banking. Apostolou also mentioned aerospace, automotive, public healthcare and defence.
Both execs view CRM as something much broader than salesforce automation. “Organisations need to connect their front office to their back office; they need to think about the entire journey across sales and marketing and service,” Apostolou said, calling them “modern systems of intelligence and action.”
“The CRM products on the market today are very similar to what we created in the 90s — they’re tracking systems,” Abbo suggested. “The effectiveness of those systems to do revenue forecasting, product demand forecasting or retaining customers by figuring out what to offer them is very limited because they’re operating off limited data in the CRM system. We’re activating all the data that’s available to inform a targeted engagement with each customer — that’s data across the business, not just the CRM system, and also data outside the enterprise.”
Dynamic disparate data
This all builds on the Open Data Initiative (ODI) work that Microsoft has been doing with Adobe and SAP. “We always said other companies would be part of ODI and part of that vision,” Apostolou noted.
“The more data running into these models, the more efficient it becomes. That’s the beauty of bringing the industry expertise and models of C3.ai into the common data model that underpins ODI. It’s bringing the behavioural, the transactional, the operational, and the financial. When you think about the ODI layer, the Azure layer and Azure Synapse bringing unstructured data into this as well, the possibilities are really limitless for the analytics and AI models you can start to generate.”
C3.ai uses Dynamics’ many connectors to pull in sources of data inside the organization, plus public data feeds for interest rates and equities; weather and terrain data are also relevant for utility companies forecasting demand and maintenance.
“It’s crazy that in 2020 we’re having to grapple with this, but we spend a lot of time taking data from different systems,” Abbo said. “This is the challenge that most corporations have: They have too many systems and their data is fragmented and siloed across the systems. In banking, the customer view of their portfolio of currencies and commodities and fixed income and equities is strewn over dozens of systems. And each system has a different identifier for the client or the account.”
Data lakes can get data into the same place, but they don’t help when two systems use a different identifier for the same invoice from the same customers, which is one reason that 99% of organisations’ data goes unused.
“We’ve solved all of that to allow you to aggregate this information about the client into a unified view, and we keep that current as you trade or as you withdraw money or deposit money, in real time or near real time,” Abbo said. “And it’s not just that you have current information, informed by the broadest set of data within the company as well as external market data, econometric data, demographic data. With AI and machine learning, you can characterise micro-changes in demand or in supply, micro-changes in what customers are buying in what locations or geographies or price points, and then react to that much more quickly.”
“In banking, if you’re looking at the likelihood that a customer might leave the bank and take their deposits with them, you have to look at what are the market rates, what are the competitive rates other banks are offering? Look at equity prices. When the customer applied for a loan last. Was the loan approved, or was it denied? These are all the factors that go into determining whether somebody is likely to leave the bank; it’s not just whether the financial advisor called them every quarter.”
Manufacturers now have access to far more information about their products after they’re sold, Abbo said. “It used to be that the manufacturer manufactured the product, and then sent it to a distributor who would send it to the customer and the manufacturer actually had no idea what the customer owns or what they’re doing with the product. Today, whether it’s a car or a medical device, the product has sensors on it, providing information directly back to the manufacturer. That allows the manufacturer to actually know which features are being used on the product, and which are not.”
That might change future product designs, but it also lets manufacturers get into aftermarket sales and service, Abbo added. “Which products out in the market in customers’ hands are likely to fail that require proactive service? Based on that, you have highly accurate revenue forecasts because you know what products you should be selling to your existing customer base. With better product demand forecasts, you could do a much more effective job in your manufacturing and in your supply and your inventory. What parts and components and new products you’re going to need, what product mix to manufacture, which customers are likely to churn so you can save them in advance, what is the next best product offer.”
C3 AI CRM builds on more than just Dynamics: it uses Azure Cognitive Services and other PaaS services — Abbo is looking at Azure Synapse for connecting unstructured data — as well as Power BI, Teams and Power Apps.
As well as rattling off a list of the common CRM use cases — “much more accurate revenue forecasts, product production forecasts customer churn, next best offer” — Abbo noted that unifying data “enables citizen developers and citizen data scientists to do literally tens or hundreds of other analyses or build tens or hundreds of other applications. They can use drag-and-drop interfaces to do further analyses and come up with price sensitivity models for customers, or look at the adoption of new product introductions into various customer segments. There are hundreds and maybe thousands of analyses that they can create with these low-code, no-code tools.”
C3.ai is essentially a bet that cloud is the best way to build enterprise software and services, rather than creating software that organisations install and customise, and it has picked Microsoft’s cloud services for that.