Earley AI Podcast - Episode 18: Telling Stories About Data Management with Scott Taylor, The Data Whisperer

Why Master Data Is the Most Important Data and How to Make Executives Actually Care About It

Guest: Scott Taylor, The Data Whisperer, Founder at MetaMetaConsulting

Hosts: Seth Earley, CEO at Earley Information Science

             Chris Featherstone, Sr. Director of AI/Data Product/Program Management at Salesforce 

Published on: November 16, 2022

 

 

 

 

In this episode, Seth Earley and Chris Featherstone speak with Scott Taylor - known throughout the data management community as "The Data Whisperer" - about the art of translating dry, abstract data concepts into stories that executives actually act on. Scott draws on fifteen years at Nielsen, stints at Dun and Bradstreet, WPP, and Kantar, and his own independent practice at MetaMetaConsulting to explain why data management professionals are their own worst enemy when it comes to communication - and how to fix it. He shares the cautionary tale of the bakery executive who told him "I don't know what you're talking about, all I know is I've got to sell more bread," the technique of presenting data reality in increasing order of horrificness until a COO puts his head in his hands, why you should never say "data quality," and why bad data plus AI equals AS - artificial stupidity.

 

Key Takeaways:

  • Master data is the most important data in any organization because it represents your relationships and your brands - if you do not have good data about those, nothing else matters.
  • The four C's of structured master data - code, company, category, and country - provide a simple, memorable framework for explaining to executives what well-curated data actually looks like in practice.
  • Bad data plus AI equals artificial stupidity - the machine learning models that executives are excited about are only as good as the foundational data that feeds them, and no amount of sophisticated technology fixes garbage inputs.
  • The reason data leaders fail to get executive buy-in is not that their work is unimportant - it is that they talk about the how instead of the why, and use vocabulary like "data quality" and "hygiene" that clears the room of stakeholders.
  • Digital transformation has a simple definition: providing value to your relationships through your brands at scale - and the word "scale" is what makes it transformation, because scale requires technology, which requires data management.
  • Every system demos perfectly because the data in the demo is perfect - when executives see a demo drop-down and ask if it will work with their data, "absolutely" is the wrong answer and the most dangerous word in software sales.
  • Data management is macro trend agnostic - whether the buzz is IoT, the metaverse, AI, or whatever comes next, every new technology still requires unique identifiers, standardized hierarchies, entity validation, and categorization.

 

Insightful Quotes:

"Master data is the data about the most important things in your organization - your relationships and your brands. If you don't have good data about that, the rest of the stuff doesn't matter." - Scott Taylor

"Bad data plus AI equals AS - artificial stupidity. Garbage in, garbage out doesn't drive any action. It's just a snarky comment the analysts make. Let's do something about it." - Scott Taylor

"You've got these great aspirations, but you don't have the goods. Sometimes you have to rub somebody's nose in it - show them the data reality in increasing order of horrificness until they put their head in their hands and say 'enough, talk to Janet.'" - Scott Taylor

Tune in to hear Scott Taylor explain why data management storytelling is fundamentally a sales process, how a sandwich made of bread and proper ingredients became a powerful metaphor for a global bakery company's data strategy, and why the data management professionals who are most passionate about their work are often the ones least equipped to communicate its value to the people who control the budget.



Links:

Thanks to our sponsors:

 

Podcast Transcript: Master Data, Storytelling, and Why Every System Demos Perfectly

Transcript introduction

This transcript captures a wide-ranging conversation between Seth Earley, Chris Featherstone, and Scott Taylor - The Data Whisperer - about what it takes to communicate the strategic value of data management to executives who would rather talk about AI and digital transformation. Scott draws on decades of experience at Nielsen, Dun and Bradstreet, and his own practice to share the techniques, analogies, and cautionary tales that have actually moved organizations to act on their data problems - from the bread-and-ingredients metaphor that resonated with a global bakery, to the shame report presented in increasing order of horrificness that made a COO put his head in his hands and say "enough."

Transcript

Seth Earley: Welcome to today's podcast. I'm Seth Earley.

Chris Featherstone: And I'm Chris Featherstone.

Seth Earley: Our guest today is a passionate self-proclaimed data advocate - an author, speaker, master data management storytelling advisor. If you are in the master data management space, you have been amused and enlightened by his creative, humorous approach to educating companies about the strategic value of master data management. Please welcome The Data Whisperer, Scott Taylor!

Scott Taylor: Hi everybody, great to be here. Thanks for having me.

Chris Featherstone: Who gave you the title "The Data Whisperer"?

Scott Taylor: I gave it to myself. Inspired by the horse whisperer - they calm horses down. I felt that coming from the data management space, I help tame and calm data down. It was an effective moniker. I put it on a badge one day at a conference and got so many positive reactions that it stuck. But we're all data whisperers in the data management space - we all help calm data down. We're psychologists and therapists too. There's probably a whole practice area around data psychology.

Seth Earley: Tell us about your background. How did you get to where you are?

Scott Taylor: I'm from the data management side of the space. I bifurcate the broader space into two camps: data management - governance, stewardship, master data, reference data, metadata, MDM, PIM, DAM, all those foundational activities needed at the enterprise level to curate and create a proper foundation - versus business intelligence, analytics, data science, AI, ML, all the cool sexy stuff. I come from the data management side.

I think I got started because my parents told me that when I was a kid, instead of building with my Legos, I sorted them. Immediate early indicator. I'm not actually a data practitioner, though - my first real data job was working for a company that is now part of Nielsen. We had a standardized list of every supermarket in the country. It seems dull, which it was, but we made it exciting. That is where my first storytelling about data came from, even though we didn't call it that back then - it was called marketing. We went to consumer packaged goods companies, they were trying to keep their own list of stores, we had a better list. It was essentially outsourced customer master maintenance, and we repositioned it with a branded integration key that ran across all these different data systems. This was pre-Y2K, before data sharing was a big thing, but people have been using data for a long time. I love when people say data is the new oil - data has been around before computers. It was around before electricity.

That opened my eyes to the power of highly structured, well-curated data and what it can do for an organization. I was at Nielsen for fifteen years and built a whole network effect business. Then I went and consulted with WPP, Kantar, and other world-class data companies, and spent time at Dun and Bradstreet. I've always been on the supplier side, always representing actual data companies rather than software or services. I've dealt with every kind of enterprise at every level of data maturity in every category all over the world.

I went to Berkeley and studied the very important data preparation subjects of history and dramatic arts. I pull on my dramatic arts training constantly. I have a fear of not public speaking.

After leaving the corporate world I went out on my own under the moniker The Data Whisperer. I saw other thought leaders and influencers making a go of it and actually building a business. My question was: do I have a point of view that anybody cares about if I'm not representing Nielsen? Could I build a following? Could I get brands to sponsor me? The answer was yes, yes, yes in pretty rapid succession, and the last couple of years have been a blast.

What I had observed over all those years meeting data leaders at enterprises was that they exhibited two very similar emotions. One was this incredible, determined passion around their commitment to make data work for their organization - especially structured master data, reference data, and metadata - and how they knew it was going to transform all kinds of problems. But that was generally coupled with intense frustration, because nobody would listen to them. Part of the reason nobody listened to them was because nobody understood them. So I've lived happily in the niche between the IT data side and the business side, helping the data side explain its value to the business side.

Seth Earley: When you think about storytelling and translating dry, complex, abstract material into something that resonates - how do you approach that?

Scott Taylor: The challenge a lot of data leaders have is that they talk about the how rather than the why. My background is sales, marketing, strategy - natural communicator and storyteller. I love that storytelling is a thing now in the data space.

When you take it apart - data storytelling - there's the story part, which is the narrative, and the telling part, which is the technique to communicate it. The questions I focus on: how simple can you make it? What's the fewest number of words I can use to articulate a concept? Can you make things relatable? Do you use analogies, metaphors, narratives - or do you just walk in with a hundred-page deck and expect to get through it while the CEO goes "what do you really want from us?"

For instance, I describe master data - which is my favorite kind of data, and I go out there and declare it's the most important data - and I haven't had anybody really challenge me on it. My rationale is that it's the data about the most important things in your organization: your relationships and your brands. If you don't have good data about those, the rest of it doesn't matter.

When I'm talking about what's critical - unique identifiers, standardized hierarchies, taxonomies, geographies - I came up with what I call the four C's: a code, a company, a category, and a country. That simple framework is a way to walk through what good structured data looks like. But in most cases, it's focusing on why things like this structural data enable the strategic intent of whatever the enterprise is trying to do.

I learned the hard way. I describe a seminal moment in my career in my book - I went into a company, talked to the IT side, they loved it, they got it, they said "the business doesn't understand it, come meet the head of sales." I explained it to him and he leaned back and said: "I don't know what you're talking about. All I know is I've got to sell more bread" - and it was a big bakery. That was a long ride back to the airport. I realized the terminology wasn't working - too technical, focused on process rather than output.

Speaking the language of the business and making sure the vocabulary you use already means something to the people on the other side is how you shortcut your process.

Seth Earley: Did you eventually come up with a story about bread? How did this evolve?

Scott Taylor: There's a motif here. That was decades ago with one major bakery. Last year I was brought in to another company - actually the largest global bakery on Earth. They were putting in a new sales system and they understood the value of the outputs and end results, but they were struggling to reinforce the value of the data stewardship and governance they were putting in place. I used a bread example with cartoon-level graphics.

I started with a sandwich. I said: this sandwich represents your objectives. This is the tasty result that is all the benefits of the new system you're putting together. The next graphic was a bunch of people making bread - that's all your systems and processes and software. But then I had a bunch of ingredients: flour, yeast, water. If you don't have the proper ingredients going in, you're not going to get that bread, and you're certainly not going to make that tasty sandwich.

That was representing the data. Just another variation of garbage in, garbage out - but you have to make it resonate with the specific audience. They were like: yes, we get it, we're going to use this graphic when we bring new people in. The objective is a sandwich, but we need to make sure we have the right ingredients.

Chris Featherstone: And when you put bad data into AI, you get artificial stupidity.

Scott Taylor: Exactly - bad data plus AI equals AS. Artificial stupidity. Garbage in, garbage out doesn't drive any action. It's just a snarky comment the analysts make. Let's actually do something about it.

Seth Earley: People nod their heads when you talk about the value of data. But then when they see what it's actually going to take in terms of investment, process change, organizational change, and technology remediation - they pump the brakes. How do you get them to focus on the basics?

Scott Taylor: Part of the problem is we use vocabulary that doesn't resonate. "Data quality" - if you want to clear a room of stakeholders, start talking about a data quality initiative, or a hygiene program. Nobody wants that. "Data governance" - has that ever worked for anyone?

Flipping it around: I think it's incumbent on data leaders to understand their business as deeply as possible before they start adding data to it. What do we do? Why do we do it? Where do we bring value to our relationships? What's our brand experience? Then map that to the data you need to get there.

I have a fairly simple definition of digital transformation: providing value to your relationships through your brands at scale. Scale is the key - what makes it transformation. Providing value to relationships through your brands is what business has always been. But at scale means technology. Technology means hardware, software, data. If you've got data, you've got data management needs.

Seth Earley: So how do you actually get executives to confront their data reality?

Scott Taylor: You have to show somebody that their baby is ugly. At Nielsen we were very overt about what I called the shame report - here are the fifty-seven ways you spell 7-Eleven, here is your customer vertical classification pie chart where the biggest slice is "Other," here are the hierarchy structures that are a complete mess. You map that data reality - which is a horror story - to the executive who just told the board about how we're going to manage our global relationships at scale. How can you possibly do that when the data you've got is in this state?

Let me prove it to you. If you've got thirty-year-old duplicate records, you can't talk your way out of that. It's a process problem. Software isn't going to do it for you. You've got to fix it. How about "search before you create" - how about trying that once in a while?

Tying together the really high-level strategic visionary stuff a company is trying to do, and smacking them in the face with the data reality: you said in your investor day presentation you're going to scale the business. You don't have the data to get there. Our data situation doesn't mean we need more data scientists. We need more data stewards. We need data management to build that foundation.

My best business story is from when I was at Dun and Bradstreet - I was with a rather large company talking to the COO. We had excellent retailer hierarchies, and we took their retail hierarchy and put it against ours and started comparing: okay, you've got this many stores in this market, we've got this many, you're missing nine. And the story we put together was in increasing order of horrificness - on purpose. About halfway through, the COO put his head in his hands and said: "Enough. Talk to Janet." And Janet - our contact - was beaming, because we were going to get this done. The seasoned part of me didn't say "oh but I've got nine more examples to show you." We stopped. Sometimes that is what you have to do. You've got these great aspirations - here is the ugliness that stands between you and them.

Seth Earley: Tell us about the book.

Scott Taylor: "Telling Your Data Story" - data storytelling for data management. On the front cover there's a label that says ninety-nine percent buzzword free. I didn't want to over-promise. It came from knowing I was effective at explaining the value of data management, and wanting to capture those themes. Part of what I try to do is give people a little solace - to realize they're not alone. The frustration that data leaders have sometimes comes from thinking their organization is uniquely bad. Guess what? You're an enterprise. This is the nature of enterprises. It's not your fault. But you've got to fix it.

I also try to be pointed about the things that have been tried for so long to help data management that simply haven't worked - like the word "quality." Even though it's important, it doesn't sell well. And getting down to the key point: the story you are going to tell about data management is not an epic journey or a suspense story. It's a pitch. You've got to sell this in. When you're talking to executives, you want them to take action on what you're explaining - and at its essence that is a sales process. It's not that you have a quota, but those sales techniques are what I try to put into the book.

I also talk about the two kinds of data storytelling out there. There's storytelling with data - for analytics, BI, putting data in a business context to drive action. And there's storytelling about data - explaining why managing your data is important in the first place. Both are needed. They are not Sophie's choice. But they are distinctly different, and the analytics side too often ignores, dismisses, or is openly demeaning toward the data management side. I've never heard a baker talk about how flour is worthless until they turn it into bread, because bakers respect their ingredients. That's my appeal to the data science community: stop being adversarial. You confuse the business, who just hears you bickering and thinks you can't even agree with each other.

It's on Amazon and on Technics Publishing. We'll put a coupon code in the show notes for twenty percent off.

Scott Taylor: One thing I love is that data management is macro trend agnostic. No matter what comes along, you still need it. I play a parlor game: whenever new technology emerges, I figure out where the master data, reference data, and metadata story is - because it's always there.

When IoT came out: everything needs to connect to everything else, the "when it should" part requires identification, entity validation, hierarchy structures, categorization, geographies - right back into code, company, category, country. The metaverse: brands are going to participate, you need branded entities, validation, transactions, standardized data. AI: bad data plus AI equals artificial stupidity. There's nothing in technology that doesn't need a proper data management foundation.

There is always a counter-current of vendors swimming against this tide saying "you don't need these things" - because they've already made the decisions about classification and data standards and architecture before they sold you the tool. That's why everything demos perfectly. Everything demos beautifully because the data in the demo is beautiful. That drop-down in the demo? Where do you think that list came from? They defined it themselves - perfectly - before showing it to you. "Will it do this?" "Absolutely." That is the most dangerous word in software sales.

Chris Featherstone: What can we look forward to from you?

Scott Taylor: More content, more events. I'm an event chair at a big data conference in Vancouver - I love that work, being in between sessions, making sure the energy is there, coming up with spontaneous bits and comments that relate to what just happened on stage. And there will be more puppets. If you haven't seen "Too Much Tech Talk" - the CDO, the chief data officer, and the IT guy who speaks only in buzzwords - they're going to meet up soon with a cat consultant from Meow-kenzie, and an analyst from Gard-purr. The corny puns play globally, and people somehow think it's my best work.

I've also got another book I'm reading to my grandson - there's "The Emperor's New Dashboard" which basically writes itself, and "The Three Little Data Scientists and the Big Bag Data" which just about writes itself too. Reading the Little Red Data Hen to my grandson was wonderful - he started ad-libbing, going "what's a silo?" I had to keep answering in character. He is now affectionately known on LinkedIn as Data Whisperer Jr.

Overall you're going to see me leaning more toward the entertaining side. There's room for all kinds of voices in this space. I stay true to my focus: helping people understand the strategic importance of proper data management, and finding as many fun ways as possible to say it.

Seth Earley: Scott, it's been a real pleasure. Thank you again. The show notes will contain links to your videos and your book - they are very entertaining and great fun. Keep up the wonderful work.

Scott Taylor: Thanks so much. Happy to help anyone who wants to talk about talking about data. I'm on LinkedIn as The Data Whisperer, on YouTube, and at MetaMetaConsulting.

Chris Featherstone: Scott, pleasure. Thank you so much for coming on. Have a great weekend, everyone. Thanks for the time. Talk soon!
Meet the Author
Earley Information Science Team

We're passionate about managing data, content, and organizational knowledge. For 25 years, we've supported business outcomes by making information findable, usable, and valuable.