Cloud Data Wars 2:  Which Data are Most Valuable?

This blog was created in collaboration with Dave Elkington, CEO and Founder of XANT.

Read part one of the Cloud Data Wars here.

Wall Street got it wrong when it comes to Microsoft’s acquisition of LinkedIn. They said it was “one pricey deal.” Analysts reported that $26.B was a “50% premium over LinkedIn’s closing price,” citing “79x EBITDA” and “8 times sales.” Despite the fact that both Salesforce and Microsoft were bidding at these levels, Wall Street widely lambasted Microsoft for overpaying.

More recently, Wall Street had priced the Qualtrics IPO at $4.5B, until SAP scooped it up for $8B–representing a 78% premium and the largest enterprise software acquisition in history.  What does industry understand that Wall Street does not?

In the early 2000s, private Industry also saw something Wall Street did not.  Venture valuations of subscription-based software companies skyrocketed, while Wall Street struggled to see the magic of SaaS. Silicon Valley had no interest in backing perpetual license business models and switched “cold turkey” to subscription-based SaaS, but public investors took their time understanding and embracing the difference. 

Now we are seeing the same divergence of valuation related to data-centric companies.  LinkedIn and Qualtrics are two examples of data-centric companies being acquired for far more than public multiples would justify. But these are not anomalies. There is a wave of data-centric companies being valued by venture investors in a way that defies Wall Street valuations.  

In our first article about the Cloud Data Wars, we highlighted the current battles to assemble B2B data. We outlined the multi-billion-dollar bets being made by Microsoft, Amazon, Salesforce, Adobe, and SAP on data acquisition. We also defined “Collective Intelligence” to mean insights derived from data collected on customers across multiple companies. The challenge of AI in B2B companies is a lack of data. No single company has enough data to make accurate and granular recommendations about the future, and when they do, it’s only based on their own history. B2B needs cross-company data for AI to be powerful.

So Which Data are Most Valuable?

If it’s true Wall Street doesn’t have a unified approach to valuing Cloud Data, but Cloud Data is worth $billions to tech juggernauts, how do we bridge the understanding gap? The key is to understand which data are valuable and why. From there we can build a model.  

All data are not created equal. Collecting data–lots of it–and analyzing it has become easier and easier with tools like Redshift, Hadoop, Cloudera, Azure ML, Apache Spark, etc. But more data is not necessarily better. Far more “big data” initiatives have been launched than have produced actionable intelligence. And in a world where Collective Intelligence rules, the type of data one has access to is paramount.

Consider the following analogy (adapted from submission by Rick Smith, president at Techcyte):

Suppose a grocery chain has kept track of what shelf they store their peanut butter on for the last 10 years for all 100 stores. A store manager then approaches an AI company to predict the best place to put the peanut butter to maximize sales. There is a lot of “big data,” so he assumes good predictions should be easy. Right? No. The first thing the AI company will request is all of the peanut sales data for the same period.  Oops! Somehow that was never stored anywhere. Collecting the right data for one year would have been better than collecting the wrong data for 10 years.

Cloud Data in B2B Sales

Let’s apply that to a B2B sales situation. If we want to optimize decisions to affect better outcomes, we must define the rules of the road:

  • What is success?  
    • In the peanut butter example, success = peanut butter sales
    • For our B2B example let’s say that success = closed sales, measured in dollars
  • What levers are under our control?
    • In the peanut butter example, the relevant lever is shelf placement
    • For our B2B example let’s say the lever is “who to call?”

Now we can search for data that help us choose which leads to call such that we can close more sales. Data valuable to this analysis display certain characteristics:

  • Collective Intelligence or cross-company. If we want information about who to call, we can consider people we’ve called in the past, or we can consider “all the people.” If we want the best people, we shouldn’t limit ourselves to data about only the people in our own history.
  • Behavioral, with outcomes. We need data on an action we have control of, correlated with the outcome we are optimizing for.
  • Evergreen. Data has currency, and its value attenuates with time. The best data is refreshed continuously, ideally through some sort of passive crowdsourcing.

In our B2B sales example, if we could observe every conversation with every prospect, along with the outcomes of those conversations, we could use that information to predict the outcome of calls to certain types of people and recommend more of what works and less of what doesn’t.

Where else could this apply (theoretically)?

  1. Best method of contact
  2. Best frequency of contact
  3. Best collateral to present
  4. Best price to present
  5. Best purchasing team to assemble

We predict that companies will store more data, not less, over time.  We assert that the right data to store follows a CIBE framework (Collective Intelligence, Behavioral (with outcomes), Evergreen.  

A New Approach to Enterprise Data Architecture

Collecting data like these systematically requires a fundamentally different way of thinking about enterprise data architecture. If you were able to rebuild enterprise data architecture, how would you do it?

We suggest that you would most likely build it the same way the Internet was built. Today’s prevalent enterprise data architecture pre-dates the Internet. We inherited it from old client-server data architectures that were lifted and shifted into the cloud.  As such, all silos and partitions of yesteryear still exist.

But if we had a chance to re-do data architecture for enterprises, we would build it such that the more people who use it, the more valuable it becomes for each participant. This is how the Internet works. When Amazon tells you that “people who bought X also bought Y,” it’s not because you bought Y–it’s because other people bought Y. Each of us benefits from patterns contributed by other users of the same Internet.  When you log in to Netflix and are presented with a few screens of suggested titles, what percentage of Netflix’s catalog is being presented?  The answer–6.5%. And these titles are carefully selected, not because you’ve already watched them, but because other people with viewing patterns similar to you watched them and enjoyed them.  The Internet works such that the more people who use it, the more valuable it is for all of us.  Enterprise data architecture can and will begin to mimic this structure over time. But unfortunately for us, that is not true today.

The way our enterprise architectures are built, we can only see what our own company has experienced in its own history. We can’t see any possibilities outside our four walls. So if we are looking for a way to chart an optimal course into our future, the data set we have access to fundamentally limits us.

That will change. As our ability to manage massive amounts of data improves, while still honoring data privacy, protecting personally identifiable information, and operating with the bounds of GDPR, we will continue to see data and data patterns co-mingled across enterprises and customer sets, much the way the Internet does it today.


Some forward-looking enterprise software companies have already architected their solutions in such a way as to produce this very type of collective intelligence. Eight billion dollars is the largest ever acquisition of an enterprise software company to date (Qualtrics acquired by SAP). Bill McDermott–SAP’s CEO–outlined the strategy of this acquisition as being related to getting the X and O data into the same place (X= customer experience data, O= operational data). McDermott called this the biggest idea of his career, and he made a massive bet on it.

But that $8bn record will be shattered. And it will be shattered by another company that systematically collects, organizes, and analyzes data across traditional corporate boundaries. Companies like Clearbit, Infer, Bombora and InsideSales are already doing this. New entrants will also race to the top of the food chain to the extent they build their data architectures to be collective, behavioral, and evergreen.

And when these and other companies outperform their peers, don’t be surprised. Just as early SaaS companies outperformed their premise-based peers, this new breed of enterprise software companies will also set a new standard. These are the data pioneers, positioning themselves to win in the Cloud Data Wars.

Cloud Data Wars 2:  Which Data are Most Valuable?

Related Posts

Posted in