Thoughts on the Microsoft-SAP-Adobe Open Data Alliance
This week, Microsoft, SAP, and Adobe (I’ll call it the MAS alliance moving forward) announced an “Open Data Initiative” or ODI. The general consensus is that this is a competitive response to the increasing domination of Salesforce.com in the CRM and marketing technology space, whose PEPM (per employee, per month) SaaS (software-as-a-service) CRM platform has continued to eat up share. Oracle is also not included in this alliance. I think this is a huge event in the marketing technology / CRM / data science world. In this article I’m taking a marketing technologist / analytics perspective to try to deconstruct this and ask some questions.
The ostensible reason for the Open Data Initiative is to make machine learning, AI, and general data transfer much easier between and across systems. The promise seems to be, “if you buy Microsoft, SAP, and Adobe, your data scientists will be able to develop models much faster, and real-time AI algorithms will be able to use data seamlessly from across the systems.” This is a nice promise.
Any marketing operations / data executive knows the frustration that comes with knitting disparate customer-related systems together. Thousands of hours of coding and testing seem to be required with each new system, and even at that point, a lot of manual download / transform / upload analyst time seems to be required. One thing is for sure: This is a needed solution that, if it works, will be in high demand.
More specifically, it seems that three guiding principles have been stated for the ODI:
- Control: Every organization owns and maintains complete, direct control of their data
- AI Support: Customers can enable AI-driven business processes to derive insights and intelligence from unified behavioral and operational data
- Partner Ecosystem: A broad partner ecosystem should be able to easily leverage an open and extensible data model to extend the solution
It’s clear that these three differentiators were chosen deliberately as a response both to Salesforce, and to very real pain-points of organizations, particularly when it comes to marketing / customer data. Let’s go through each in a bit more detail:
Control
This is the most obvious response to Salesforce, whose PEPM walled garden frustrates CTOs who have to pay $80 for an employee who needs to log in once a month to fill in a campaign form. It’s true that apps can be built outside of Salesforce via API, but even so, they make it very clear that they want it to stay inside the system. APIs—particularly for small companies—are rate limited and can be clunky. Most smaller companies, and even some larger ones, still use bulk manual imports and exports using data loader. MAS seems to be saying “you own your data, for reals.” We’ll see how true that is.
AI Support
AI is at a point where “websites” were in 1997. Seemingly overnight, $20M firms arose whose only competency was coding HTML and standing up some active server pages. Today, most technologists still aren’t clear on whether AI modules will be provided mainly by vendors, or developed internally. Software providers clearly want marketers to use (and pay for) their AI solutions, instead of hiring data scientists and building bespoke algorithms. The MAS alliance is smartly promising that AI endpoints will be easily accessible to companies that choose a MAS solution. Of course, they’re still hoping that these companies use the built-in AI being developed by each, or the AI being developed on the Azure cloud. But, by providing an “option of freedom”, they’re putting executives’ minds at ease.
Partner Ecosystem
Finally, the Partner ecosystem approach is really about the open and extensible data model. A trend we increasingly see is the desire for a “platonic ideal data model” for customer and marketing data. This is absolutely possible, just like it is for accounting data (XBRL) or healthcare enrollments (HIPAA 834) (Yes I know these are both old and need refreshing, but they’re still standard.) By providing this open and extensible data model, companies can build applications and data exchanges without fear of deprecation, meanwhile trusting that other companies will also build their software using the same object and data definitions. This is really smart, but it could set off a VHS / Betamax war if Salesforce follows suit and announces its own standard.
Other Observations
Structured Marketing Data Platform vs. Data Lake
What’s also interesting about this announcement is what it’s not, which is a marketing data platform. It’s clear that each member of the MAS alliance wants to keep its part of the sandbox separate, thank you very much. The idea of a co-hosted real-time data store between the three—think an “AI and data exchange real-time clearing house”—was probably discussed but discarded. It’s too risky, too expensive, and frankly too heavyweight and tightly coupled to be successful.
However, a standards-based partnership does support a data lake approach, in which real-time data are fed “raw” from each system into a high-speed, cheap-storage cloud environment, which coincidentally (lol) Microsoft is happy to provide. I’m still not convinced that data lakes are the long-term future for scalable marketing analytics and AI, because of the mess that results and the human data cleaning required of it—but this could change that, and make the data lake concept more realistic for more conservative enterprises.
Speed and Uptake
One thing that will be interesting is to see if the standard data model is quickly deployed, documented, and used. MAS probably knows how many large enterprise customers they have in common, and I’m sure several are lined up to pilot and prototype. Forbes points out that this isn’t just a paper launch, and that a lot of work has already been done.
Blockchain?
One other thing to watch is blockchain. Blockchain is totally overhyped, but this is a use case where it could work. If Microsoft is sharing an opportunity back and forth with Adobe in anything other than a one-way handoff, a distributed ledger could make that opportunity in essence a “shared object” that remained secure. It would be clear who wrote what to the record, and when—even with the record existing simultaneously in two (or three) systems. It’s kind of like a quantum realm of marketing data. It’s both a particle and a wave… or maybe I’m torturing that analogy.
Competitive Response
Lastly, I’ll predict that Salesforce.com further strengthens its partnership with Amazon. Microsoft clearly wants all of the cloud storage and processing business for the MAS partnership, and Amazon is still storing most of its massive CRM data on its own proprietary (and obsolete) cloud of siloed Oracle databases. That’s changing, as Salesforce has committed to put more of its cutting edge products on AWS—but will the walls come down more, and will Salesforce think about decoupling its data storage from its UX? If it does, those PEPM fees will go with it, but it would be an interesting counter-move to the open-extensible alliance that emerged—albeit in an inchoate form—from Microsoft, Adobe, and SAP.
