I've heard "gurus" say that the Hadoop fallout and MapR fire sale wasn't an indictment of the big data era at all - just a signal that the action has moved to the cloud. Others critique the notorious difficulties of Hadoop implementations.
For some of us, that isn't enough. We're looking for a different data narrative altogether.
Like Constellation's Steve Wilson (last Friday's subject), Wang is one of the rarest types you'll find in the enterprise - an enterprise provocateur with the substance and method to support her assertions. Provocative, you say? How about this from the Sudden Compass home page:
In fact, the average return on “big data” technology is 55 cents per dollar.
And from our recent podcast, Wang argues:
Companies think that if they just get more data, they're going to make better decisions. And that has been my life's work to show them: just because you're gathering more quantitative data, does not lead you to better decisions.
She doubles down:
In fact, it can lead you to make more risky decisions and dangerous decisions. The worst part is it can lead you farther away from your customer.
Hold up - data takes you farther away from your customer? You don't hear that every day. Wang and I have a banter going about enterprise data/design problems, tracing back to our first sit down at an Enterprise UX show. She's a bit of a YouTube sensation, so I kept up with her by watching her TED gigs.
Exposing big data promises - "it's just not happening that way in reality"
At Constellation Research's Connected Enterprise 2019, we finally cranked out a podcast that delved into her thinking - and our shared skepticism of CX hype (Forget data-driven - we must be insight driven. Deconstructing big data and CX - podcast also embedded below).
So, Tricia, about that data taking-you-away-from-your-customer problem? On the podcast, she elaborated:
It's a total conundrum, but I have seen it time and time again in companies. Very few leaders want to talk about this big white elephant in the room because what it is.
I'm tackling this entire vendor industry that is swearing they can help clients find that needle in the haystack about their customer. If they just buy that dashboard, they're going to get all of their customer data, their CDP, their DMP, all of that into a big lake, and they're going to be able to make all the decisions they want to bring their company to the next level of growth. It's just not happening that way in reality.
To be fair, Wang is not anti-big-data. Her problem is idealizing (big) quantitative data, and losing the course-correcting-power of qualitative data - what she calls "thick data." At a "technology ethnographer," she comes at this differently. I took the cheese from Wang on that one: what's a technology ethnographer?
Ethnography means the science of studying people. I just so happen to study how people use technology. I made up that job title because there was no job title that fit me, My work crosses marketing, it crosses digital data and design, so I was like, "Well, how do I actually fit myself into any one of these functions?" Instead of making the job title fit me, I was like, "Why don't I make up a whole entire new job?"
As I see it, it's a pivotal time for the big data debate. While there are the aforementioned big data apologists, there is also a good deal of disillusionment about big data promises. I think most companies are ready to concede that big data isn't a magical solution. But the problem remains: how can we make better decisions with data?
Data-driven is the issue - "we need to be insight-driven"
To bring that into podcast focus, I quoted one of Wang's prior talks:
Most things fall apart in decision making. The insight loses fidelity as it goes up what I call the alignment journey. A truly new insight or idea gets distorted and through each step of the hierarchy, as people try to tack it on to what the business is already doing, through a PowerPoint on top of another PowerPoint on top of another one.
By the time these insights actually even make it up to the decision makers, they somehow feel less powerful and transformative.
That rings true - but how can so-called "thick data" help?
73% of organizations are not succeeding with their big data transformation, and Gartner's report just last year said that 91% of organizations have not yet reached a transformational level in their digital transformation. Why is this happening? Why are the numbers not getting better?
Why are we struggling so much when we already have all the technology to collect, to warehouse, to store, and to analyze all the quantitative data that we have now? It's a complex answer, but there's one big answer that is glaring. In all of these efforts, we're not able to learn from our quantitative data because we're missing an entirely other set of data called thick data. And that's qualitative data.
And why "thick data"? Wang continues:
If you've never heard of thick data before, I rebranded qualitative data, like ethnographic data stories, emotions, firsthand data, as unmediated data as possible - I rebranded it "thick data" to sound sexy in a room full of data scientists, who used to call my data sets small.
They would be like, "Well, we have collected 5,000,000 data points from our first-party data, and I would say "Yes, I have talked to ten people - and I've spent the time. But they're like, "Oh... What a small data set.
That qualitative data set might look small, but ignore it at your peril:
For companies to think that they can only invest in quantitative data, and actually try and grow their businesses without any of that thick data, to me that is ludicrous.
Wang dresses down another darling phrase: "data driven."
Data-driven is actually the issue. We need to be insight-driven. We need to understand the insight that the data is telling us. And then that's what going to help a decision-maker act - or make a decision.
Embracing qualitative (thick) data - the Netflix example
"We've exaggerated the promises of the quantitative." For Wang to make this argument convincingly, she's going to need examples - and she shares several on the podcast. The Netflix example always jumps out:
Netflix had that big million dollar contest, where they said, "What engineers out there can actually help us write a better algorithm to improve our recommendation algo?" And I think in the end, it didn't actually make a measurable difference in the recommendation with whatever teams that they had submit.
Enter the qualitative:
It wasn't until they said "Okay, we're going to go embrace thick data." They hired an ethnographer, a really well known one named Grant McCracken, who went out there and just gathered thick data.
What does that mean? He literally just hung out and spent time watching TV with families, watching media, understanding their daily lives.
That's when it emerged. He was like, "Wait a second, people are watching shows consecutively, back to back, and the best part is they're not embarrassed about it.
Instead of giving you a recommendation for a different show, guess what their recommendation algorithm did - you don't need an algo to do that.
Why we need a "department of the unknown"
There's one more foundational concept we need to grasp Wang's approach: her call for companies to create a "department of the unknown." What's that about? As she sees it, enterprises excel at managing known variables, and reporting on existing customers and assets. But that's not enough in today's market:
For a company to not just survive but to thrive in this era, because the world is so complex now and you have all these new kinds of customers coming in, new markets, you also have to know how to manage the unknown - not just manage it, but then take advantage of it - and have the unknown drive your growth.
No, she's not expecting companies to literally put "department of the unknown" on signage:
I say it facetiously. I don't technically mean you have to create the department of the unknown, but what I mean is you need to have people say, "You know what, you're in charge of bringing the unknown into the company, and you're going to get the resources to be supported to do that, because that's where we're going to look for growth."
If you think this is another code word for "create a little incubator" or "disrupt yourself with an internal startup," that is NOT what Wang has in mind:
And not in a siloed way, so I'm not just talking about only in an innovation lab, throwing more money out there. The company needs a department or a person or a group of people and processes, and a culture that really knows how to find the unknown - and ingest it into the culture and drive that into the business.
What we don't need in the enterprise is one more armchair critic. Wang and her Sudden Compas co-founders have taken this critique, and fused it into a methodology:
This is a new practice, a new skill set we teach - that we think is needed for the 21st century, which is the ability to integrate different kinds of data sets. In particular, data sets that are on the quantitative level and also on the qualitative level - and you have to integrate that to tell a coherent story.
This is why we're teaching companies, and we're working with universities and also professional programs to make sure that this new skill set that is taught and learned.
In the podcast, we get into the Sudden Compass methodology, which includes formation of a data strategy, implementing and scaling a data practice for "insights," and their Unlock Sprints project approach. It comes down to this:
How do you integrate qualitative and quantitative, the big data and the thick data, because if you can do that, then you can start being able to play with different kinds of data sets to make better decisions for your customers.
The good news for companies frustrated by the millions they have poured in, who don't want to start over: Wang's team doesn't advise starting over. With a tools/platform agnostic approach, their goal is to help companies get to what they call "Data 3.0." 1.0 was about technology investment, 2.0 was about talent/skills/data science teams. And 3.0?
Data 3.0 is about insights, about how do we communicate with the data. Now that data is universal; it is pervasive within the organization. I did my job as a chief data officer. I got all the data in one place, but now I have to make sure that all the functions can communicate with our data in such a way that it makes sense for the customer so that you're, in your words, consistently coming through for the customer.
Of course, to do that, we have to wade through a bunch of CX hype - another term Wang and I attempt to dismantle during the podcast. Writing about that will have to wait. For now, we'll have to settle for banishing big data hype and "data-driven" infatuations in the same article. I'll leave you with one more Tricia Wang zinger for the weekend:
I wish I could start a support group for everyone who has bought technology from vendors and then not gotten what they wanted.
End note: one news item to look out for, which we cover in the podcast: soon, Sudden Compass plans to open source their Unlock Sprints framework for "integrated data thinking." I'll add in a link here when that is live.