About
I spend much of my time on the problem underneath most data and AI failures: systems that…
Articles by Malcolm
Activity
24K followers
-
Malcolm Hawker shared thisWhat’s on my reading list these days? I’m in the middle of three books, all of which I definitely recommend: 1. Software Wasteland by Dave McComb. This book was recommended by my friend Juan Sequeda so I took his advice and got a copy - and I’m glad I did. The book talks about all the wasteful spending and excessive complexities in the world of IT - something I witnessed first-hand while an analyst. This book renewed my fading optimism around the idea that perhaps one day we’ll embrace a ‘common language’ within our organizations. 2. The Ontology Pipeline by Jessica Talisman. I was given the honor of being a reviewer of this book, and I can’t recommend it strongly enough. AI is increasingly demanding that data leaders integrate knowledge management capabilities into their organizations, and this book is a much needed ‘how to’ manual for enabling that. 3. Bridging Knowledge, Data, and AI by Zach Wahl, Joseph Hilger and Lulit T.. This book is also very focused on knowledge management, and has become almost like a reference manual for me around all things related to the evolving ‘semantic layer’. It’s been lurking in my laptop bag for weeks - well after I first finished it. The next books in my reading queue: 4. Data and AI Skills by Jordan Morrow. Jordan is a skilled writer who truly understands the ‘people’ aspect of leading a data function. 5. Mixed Model Arts by Joe Reis. Is data modeling dead? Absolutely not, but it is evolving, and I’m looking forward to having Joe tell me exactly how. 6. Designing the AI Driven Foundations by Sanjeev Mohan. Sanjeev is one of the sharpest minds in our field, and has a wonderful skill for making the highly complex seem extremely approachable. I’m also looking forward to yet-to-be-released led books by Winfried Adalbert Etzel, Stacy Olson (MBA, MSIS), Ryan Dolley, and many others. What’s on your data reading list? #booknerd #data #ai
-
Malcolm Hawker shared thisIs your data team a tax on the business? If you're running it as a service desk, the answer is yes. And your business stakeholders already see it that way, even if they're too polite to say it out loud. That framing comes straight from my conversation with Nick Zervoudis on the latest episode of the CDO Matters podcast. Nick has been a data product manager since long before the title existed, including scaling data products across all of Europe for PepsiCo, and he now trains data teams to think less like data people and more like business people. We spent the better part of an hour on why product management is the way out of the cost center trap. A few of the highlights: ➡️ The service desk model has a low, hard ceiling on value. When requests arrive as tickets without context, you only deliver the right thing by coincidence. That's not customer focus. That's roulette. ➡️ Product owner and product manager are not the same job. Owners are too often hired to scale a templated thing. Managers validate product market fit first, then scale. Skipping the first step is how six months of engineering gets spent on a capability nobody actually needs. Nick shared a real story on exactly this. ➡️ The first question on any request should never be CSV or Excel? It's what problem are you trying to solve? Asking for a specific business outcome kills bad requests, surfaces better solutions, and often gets the stakeholder to say no on your behalf. ➡️ Putting a dollar figure on data work is easier than you think. You dont need precision, you need order of magnitude. Is this a six figure opportunity or a seven figure one? That answer changes what your business leaders hear when you talk about quality, debt, and priorities. ➡️ In an agentic world, defining business value gets more urgent, not less. Agents are machine customers. If you can't serve human customers well today, you won't magically serve automated ones tomorrow. My take: behind every wildly successful product, and every colossal product failure, is product management. If you're a data leader with headcount to spend next year, hire a product manager. If youre a practitioner, start acting like one now. Be curious. Push back. Understand the why. So here's my question for you: does your business see your data team as an investment, or as a tax? And if you dont know the answer, what does that tell you? Full episode with Nick Zervoudis is live now, wherever you get your podcasts. Available here: 🎙️ https://lnkd.in/enyG9f6h
-
Malcolm Hawker shared thisLast week I took my first real vacation since before COVID. I went to Spain for a family wedding. More specifically, my wife and I went to Madrid and Menorca - including two day trips from Madrid to Toledo and Segovia (that’s the photo). It was amazing, but it took me a full week before I could fully decompress. And then it was over. I definitely need to do that more often. #pto #worklifebalance #holiday
-
Malcolm Hawker shared thisWhat are you doing at 11am EST today? Join me for CDO Matters Live! Later this morning I'll be holding my monthly live data extravaganza - and this month's installment is all about 'AI Ready Data'. Is your data ready for AI? It might be more ready than you think. Join me today for a lively discussion on your state of AI readiness! Profisee #cdomatters #aiready #datagovernance
-
Malcolm Hawker posted thisIf you're working to capture data definitions in a catalog, then you're 'doing' data governance. You may not be calling it governance, or have a formalized governance function, but the definitions of data objects and attributes are the foundational building blocks of any data governance program. And like any other data governance policy, they require collaboration, consensus, and ongoing change management. The most important of these definitions are those for objects that are shared widely across the organization, and are used extensively within business-critical KPI's that span multiple functions or departments. Revenue. Customers. Products. Locations. Assets. etc. etc. etc. The establishment of shared definitions for these objects, commonly known as master data, establishes a shared language (aka 'semantics') for both analytical and operational uses of data. Put another way, these definitions act as the 'connective tissue' within complex and and highly siloed business processes, like 'quote to cash', or 'procure to pay'. These shared definitions literally enable our businesses to run, and for our business executives to trust any analytical output - including AI. Not all data is created equally, which means not all data definitions are of equal importance. If you're working on definitions in your catalog, or your 'semantic layer, then start with the data that matters the most, master data. And keep in mind, that establishing rules for data that's widely shared across the organization requires collaboration from everyone affected by those decisions. And by any other name, this is still data governance. #datagovernance #data #semanticlayer
-
Malcolm Hawker posted thisI've long positioned the challenges around data governance as an incentives problem. If you want to understand human behavior, understand the incentives. And when it comes to data governance, the incentives of data consumers are different than the incentives of data managers and governors. It's that simple. The incentives of data managers align to control over data. The incentives of data consumers align to business outcomes. So until we solve the incentives problem, we won't fully solve for governance. What's the answer? It all starts with measuring the economic impact of data governance, optimally through both the lens of cost avoidance, and business enablement. Yes, data teams must start to measure the business value of data governance. Next, we need to measure the cost of data governance. And once we understand both the costs and benefits, we can then begin to better understand how to close the incentives gap. We can build financial models that more closely align the costs to the beneficiaries. We can understand when costs should be shared across multiple beneficiaries. We can ensure that those 'paying' to govern data can be adequately compensated for their work. We can start to treat data governance as more of a P&L, where decisions on how (and where) to fund governance efforts are completely rational, and not emotional. And all of this starts with measuring it's cost and benefits. If we do this, we can solve for the incentives problems, and data governance programs will flourish. Else, data governance efforts will continue to struggle. The bigger question to me is this: what incentives exist for data leaders to continue the status quo of *not* measuring the business value of data governance? This is arguably the harder problem to solve. #datagovernance #governance #aigovernance
-
Malcolm Hawker posted thisIf you think you're 'behind' on AI, you probably aren't - at least by comparison to others. How do I know? It's a big part of my job to understand what's really going on in the data and analytics space. Do to that, I travel the world talking to data leaders, speaking at conferences, and interacting with practitioners across every conceivable data role. And I'll be honest. When it comes to operationalizing AI at scale across a majority of companies, I think that most of what we're hearing is hype. There are maybe 10-15% of companies on the bleeding edge of AI, but when it comes to any sort of highly customized AI solutions, ~80%+ are still mostly just trying to figure it all out. This means: ➡️ Most have not figured out 'context', and are a long way from having any semblance of a robust semantic layer for AI. Most efforts here continue to be on deploying a data catalog as a 'source of truth' for data definitions. ➡️ Many who say they have deployed agents have actually deployed chatbots, and those bots tend to support more internally-facing use cases, without much in terms of customized capabilities. ➡️ Highly customized (and complex) retrieval patterns are not the norm, and most companies are just now starting to get acquainted with the growing number of vendor tools available for bespoke agentic workflows. ➡️ Most have not figured out how to integrate their internal structured data into any agentic process, which also means that most are still struggling to adapt their governance programs to best support AI. Most understand that they need to harden their foundations for AI, but most also struggle to understand how those foundations must adapt, and more importantly, to find the time, expertise, and funding needed to adapt a program that most continue to manage as a cost center. ➡️ Many companies that are claiming to have AI solutions in production have only enabled LLM use at the desktop. ➡️ Most continue to treat AI as an IT deliverable, and are not investing in the process changes / transformations needed to realize meaningful business impact from AI. None of these statements should be construed as a judgement. Ultimately, when it comes to AI, companies are moving at the speed they want to move. Rather, these are just my observations of reality vs. hype. So if you think you're a long way 'behind' on AI - I suspect you aren't. You're probably in that massive majority that's just trying to figure all of this out. If all the hype helps to motivate you, then great. Otherwise, I would take it all with a massive grain of salt. #ai #datagovernance #cdo .
-
Malcolm Hawker shared thisI just spent a week at the Finovate FinTech conference in NYC, and here are some of my top takeaways: ✅ There is a large and growing market for data management solutions being developed specifically for Financial Services companies that exist completely outside the traditional data and analytics industry. These solutions are being sold to heads of Innovation, Chief Risk and Operating Officers, CFO's and many others - not the CIO or CDO. This forces me to conclude that the data function, at many bigger banks, is simply not innovating at the pace they need to, which is forcing functional heads to seek their own solutions. ✅ The format for this conference was unique, and it created a special dynamic not usually experienced at most data conferences. Two of the three days were devoted specifically to showcasing the solutions of smaller startups via 7 minute pitches of their solutions, where most of the participants were seeking funding from the many VCs and private investors in the room (see the pic below). This had the effect of attracting attendees who were all very eager to learn about the 'latest and greatest'. People wandering the exhibit hall actually *wanted* to hear what Profisee does, and the value we could possibly bring to their company. ✅ Vendors in the Fintech space are working to solve for AI Governance, but without calling it AI Governance. ✅ While AI is clearly a major focus for most, it actually *wasn't* the top concern of the most senior leaders in attendance. Their top issue was solving for growth in an era where younger individuals are less and less engaged with the traditional banking system. So rather than focus on the minutiae of data and governance tactics, many conversations with the attendees were instead rooted in business transformation and strategy - which was an extremely refreshing shift from most data conferences. ✅ A focus on showcasing the 'latest and greatest' tech also helped to recruit a lot of very senior leaders from banks - who seemed to be using the conference as a means to stay abreast of the rapid advancements in their fields. I had many conversations at the Profisee booth with several bank Presidents and Board Members - people you won't ever see attending data conferences, including Gartner. In all, it was a great week in NYC and a learned a lot. The Fintech space is booming, and data leaders at financial services companies would be well served to get more connected to it. #fintech #finovate #ai
-
Malcolm Hawker posted this9/11/2001 started just like any other day. I was working as a Product Manager in the Advertising Technology team at America Online, in northern Virginia just outside Washington DC - very close to Dulles airport. Our team built the systems responsible for the scheduling and delivery of all of AOL's online advertising, with almost of my internal customers/users being based out of our NYC office in Chelsea. AOL had a private jet that would shuttle back-and-forth from Dulles to NYC every day, and often I would be on it. But thankfully, not this day. Within minutes of arriving to the office, it was clear this was anything but a normal day. In 2001 cell phones were common, but they didn't have video capabilities, and most texting relied on SMS. Also, video streaming to desktops was painfully slow and not widely adopted. That meant that most people needed to rely on TV and radio for updates, and on this morning, everyone in our office was huddled around the one TV we had connected to cable. Watching everything unfold was surreal and horrifying. Many people on our team had loved ones in both the Pentagon and the WTC. And all those people were furiously trying to make contact with their loved ones, but couldn't. That's because the local phone lines in both DC and NYC became quickly overwhelmed by people trying to contact their friends and family for updates. Instead, all they would get is a fast dial tone. I was lucky enough to have a device that didn't rely on SMS for texting, and was able to get messages out to my family, but my situation was pretty rare. Ultimately, everyone in our office would be sent home out of a concern that the last unaccounted aircraft would try to return back to Dulles (where it originated from), and our offices were literally in its glidepath. Instead, that plane went down just to the north of us, in rural PA. I had friends and coworkers who lost loved ones in both DC and NYC. My heart breaks for them, still to this day. Their lives were forever changed, as was mine. I will never forget. #nyc #911 #dc
Malcolm Hawker commented on a post
12h
My apologies to anyone I tagged on this post who may have received multiple notifications here - when I went to make the post, LinkedIn said it 'failed', but as it turns out, it got posted three times. Go figure.
Malcolm Hawker commented on a post
14h
Said another way: both truth and meaning are context-dependent. Roughly speaking, there are about a dozen 'contexts' in any big company that align to business functions (sales, marketing, finance, etc.), and each will have its own semantics, and its own set of truths. That said, within any one bounded context, there can only be one version of truth. So the concept of a 'single source of truth' is not dead, it's just multidimensional.
Malcolm Hawker replied to a comment
15h
Shachar Meir I'm bummed I missed BDL and the opportunity to meet you, but hopefully some day!
-
Malcolm Hawker liked thisWhat’s on my reading list these days? I’m in the middle of three books, all of which I definitely recommend: 1. Software Wasteland by Dave McComb. This book was recommended by my friend Juan Sequeda so I took his advice and got a copy - and I’m glad I did. The book talks about all the wasteful spending and excessive complexities in the world of IT - something I witnessed first-hand while an analyst. This book renewed my fading optimism around the idea that perhaps one day we’ll embrace a ‘common language’ within our organizations. 2. The Ontology Pipeline by Jessica Talisman. I was given the honor of being a reviewer of this book, and I can’t recommend it strongly enough. AI is increasingly demanding that data leaders integrate knowledge management capabilities into their organizations, and this book is a much needed ‘how to’ manual for enabling that. 3. Bridging Knowledge, Data, and AI by Zach Wahl, Joseph Hilger and Lulit T.. This book is also very focused on knowledge management, and has become almost like a reference manual for me around all things related to the evolving ‘semantic layer’. It’s been lurking in my laptop bag for weeks - well after I first finished it. The next books in my reading queue: 4. Data and AI Skills by Jordan Morrow. Jordan is a skilled writer who truly understands the ‘people’ aspect of leading a data function. 5. Mixed Model Arts by Joe Reis. Is data modeling dead? Absolutely not, but it is evolving, and I’m looking forward to having Joe tell me exactly how. 6. Designing the AI Driven Foundations by Sanjeev Mohan. Sanjeev is one of the sharpest minds in our field, and has a wonderful skill for making the highly complex seem extremely approachable. I’m also looking forward to yet-to-be-released led books by Winfried Adalbert Etzel, Stacy Olson (MBA, MSIS), Ryan Dolley, and many others. What’s on your data reading list? #booknerd #data #ai
-
Malcolm Hawker liked thisMalcolm Hawker liked thisThe same vendors who sold you Single Source of Truth are now pivoting to Single Source of Semantics. Like its predecessor, it's a noble goal. It also isn't possible. Single Source of Truth tackled a real problem – two people ask the same question and get different answers, because their questions trace back to tables that don't agree with each other. That's worth solving. But in solving it, we often missed that the same question can have more than one right answer, depending on where you sit in the organization. And what's true today isn't necessarily true tomorrow. Business changes constantly. Keeping a single place that represents all of it is basically impossible. Now many vendors are saying 'Store all your context here, the one source for AI!' Raise the objections above and they wave them away with AI auto-curation. But AI accelerates the creation of data problems just as much as it accelerates their solution. Expect conflicting, scattered context that you're constantly managing. That's my read anyway. What's yours?
-
Malcolm Hawker liked thisMalcolm Hawker liked thisWoohoo, my debut book “Humanizing Data Strategy” achieved a new milestone today: 100 amazon reviews! Sure, there have been occasional negative reviews about me being “marxist”, having written “DEI cr*p” or it being AI slop (it’s 100% human written btw) - but the majority of feedback has been overwhelmingly positive. It’s been a little longer than 2 years since the book’s release and I’m still hearing about people quoting it, recommending it and sending me feedback about it. Huge thanks to all the readers and reviewers! #author #datastrategy #tiankaistuff
-
Malcolm Hawker liked thisThe remarkable thing about this panel is that each of the participants typically keynote major conferences in the field. First of all, credit to each of them for being willing to sit on a panel rather than headlining. For those that are attending the Semantic Layer Symposium, you're going to get the keynote energy and thought leadership of Noz Urbina, Malcolm Hawker, Amy Duncan, and Ben Clinch condensed into an hour of debate. I think Joseph Hilger will have his work cut out for him as moderator!Malcolm Hawker liked thisOnly 3 weeks to go before Semantic Layer Symposium! I'm looking forward to getting back in touch with great thinkers/speakers like Ben Clinch and Joseph Hilger and finally in-person meeting with Lulit T. and Benoit Gaussin after over 8 years collaborating with my friends at Enterprise Knowledge, LLC. You can get a sneak preview on the EK Knowledge Cast where Lulit and I discuss the state of the industry, the impacts of AI, and the challenges we our space needs to address: https://lnkd.in/gF8dNEQE (PS - And you can get a discount code...!) Get symposium tickets on https://lnkd.in/gQBxi-AP
Experience & Education
Licenses & Certifications
Projects
Recommendations received
8 people have recommended Malcolm
Join now to viewOther similar profiles
Explore more posts
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top content