Jeff Elderton

The thirst of data in all organizations is insatiable. While everyone argues over which LLM is the best, pragmatic data leaders are and should be concerned about how to serve data and semantic intelligence to these thirsty models. It’s time to talk about platform interoperability. That is the surest answer to this predicament that every CDAO finds him/her/them in. It’s a topic that does not get as much of a voice as it should.

41

6 Comments

🛑 ❗ Leaving technical names in your dashboard smacks of inconsideration for your audience. 👀 I’ve seen both inexperienced and more seasoned dashboard developers guilty of this underscore-toting offence and it makes me cringe every time! 😬 You can’t expect your business users to correctly guess what each graph is showing from alphanumeric-and-underscore soup. 🍜 Even for fairly straightforward terms like sales_num - the mature dashboard designer will ensure it is changed it to a properly capitalized and spaced business name label ---- before it ever rests on business users' eyes! 👩‍💼 👨‍💼 Check out my article for more info 👇

11

1 Comment

When I'm presenting on data storytelling, I typically start my talk by noting why it's important to use pie charts sparingly. This Datawrapper article helps make this point clear. https://lnkd.in/grdwGM3y And here's a Tableau summary of what charts to consider using instead of the pie, including three of the alternatives I like the most: tree map, donut and waffle (yes, there are a lot of chart types named after sugary foods!) https://lnkd.in/gUJ9jq5P And since we're talking about pie charts, I thought I'd share this fun one that I saw in Philip Bump's Saturday morning data viz newsletter, "How to Read This Chart."

7

I've been in the syndication space for a while, and understanding why syndication is so difficult is important to understanding how your company can get better at product syndication. Educating yourself can avoid choosing the wrong tool or strategy that can cost you sales and resources. Reach out to Trailbreakers.ai if you want an analysis of your Syndication needs.

6

Excellent essay on data pricing. It’s all about the application of the data, there’s no intrinsic price. https://lnkd.in/gWxRBr8h

3

1 Comment

{ "widget_block": { "value": { "raw_instance": { "content": "" } }, "type": "option", "user_id": 231528767, "date_modified_gmt": "2024-08-12 21:41:37" } }

2

1 Comment

I’m excited to share Molecule's first-ever ETRM/CTRM Transformation + Modernization Report! It has been seven months in the making, and it was no small feat to get to this point. This report shows that trading orgs are waking up to the reality that modernization is the key to their survival. Agility and speed are their top priorities in times of dramatic market shifts, and what companies expect from their #ETRM and #CTRM systems is changing to reflect that. A HUGE thanks to all who participated in our survey and provided expert commentary: Sandy Fielden, Geoffrey Davies, CMB, CMS, Geoffrey Cann, Steve Harwitz, Mark Bosse — couldn’t have done it without y’all!

48

1 Comment

Everybody Lies by Seth Stephens Premise: The book illustrates how big data can reveal our deepest secrets and highlights both the potential and the pitfalls of relying on massive datasets to understand complex human behaviors. My thoughts: As folks start in pattern recognition and data analytics it’s common to overvalue the data over intuition and personal insight. Seth Stephens makes the argument that while big data is important it’s only part of the conversation. That data needs to be balanced against our sense of reality and ethics. One example called out in the book is if we took a thousand coins and flipped them everyday and recorded which landed on heads and tails and compared it to the stock market where heads means it goes up and tails means it goes down, at the end of the year one coin will likely have done better at predicting the stock market than all the others. Does that mean we should then use that coin next year to inform our investments? This is a great resource for our big data works, where we can find big data, how we can use big data, and most importantly when we need to be careful with data and our conclusions from it! Insights from Everybody Lies: Good data science is less complicated than people think. The best data science is surprisingly intuitive. If you can't understand a study, the problem is probably with the study, not you. Don't trust what people tell you, trust what people do. If people constantly tell us what they think we'll consistently hear things more comforting than the truth. Collecting rich data on the world's problems is the first step towards fixing them. We're all so busy judging our own selves that there is little energy left to judge others. Neighbors of lottery winners are significantly more likely to go bankrupt. The more assertive the promise the more likely they will break it. You have to be open and flexible as to what counts as data. People who are going to be successful are going to find advantage in any situation. See comments for the do's and don'ts #BigData #DataScience #Analytics #MachineLearning #AI #EthicsInTech #BehavioralEconomics #PredictiveAnalytics #DataDriven #Technology #BookReview #SethStephens #DataAnalytics #DataEthics #DigitalTransformation #HumanBehavior #Privacy #PolicyMaking #TechEthics #Innovation #DataPrivacy #TechTrends

4

2 Comments

Let's talk about CROSS JOIN. What it is. What it most definitely is not. It is useful. It's not a conditional many-to-many join. The main thing -- the dangerous thing, usually -- about cross joins is that they form the Cartesian product of two tables. That is, every row in the left-hand table is matched to every row in the right-hand table. The number of rows in the outcome is the product of the number of rows in the two tables. CROSS JOIN is syntactic sugar. You can accomplish this by using a different join type. These are the same: SELECT a.*, b.* FROM a FULL OUTER JOIN b ON TRUE; SELECT a.*, b.* FROM a CROSS JOIN b; I've seen cross joins proposed as a method of "finding all pairs of football matches played" or "finding all pairs of cities with flights between them". For little examples like this, the output looks correct, because you're wanting a dense outcome set. But imagine two tables with 1 million (1,000,000) rows each. The product of these will have 1 trillion (1,000,000,000,000) rows, right around 125 rows for each person alive today. If you want non-dense many-to-many matches, use an INNER JOIN with your many-to-many bridging table. If your rows have arrays of matches (instead of a bridging table), use a LATERAL JOIN first to expand these rows. (LATERAL JOIN is written as an UNNEST in some SQL variants.) So what's a CROSS JOIN good for? (1) Widening rows with additional information. (2) Providing an exhaustive list of widened rows. One of the tables should be tiny, and contain columns relevant to every row of your other table. Consider a 1-row table. All it contains is a pair of timestamps, and you want to apply these timestamps to every row of your main table. CROSS JOIN is great for this. Your number of rows is unchanged, and you get your extra columns. Or consider a case where you've moved from having 1 warehouse of products to having 3, and you want to add a warehouse location to each row of your product catalog. A CROSS JOIN of your product catalog and your (very short) list of warehouses will do this immediately.

3

In the health trifecta of sleep, nutrition, and exercise I have come to believe through my own experience that sleep is chief among them. (and I would put them in that order of importance) This excerpt resonated for me.   In early 2020, with pressures of the global pandemic, business challenges, and life challenges all converging for me, I started experiencing persistent insomnia for the first time in my life. It was jarring, and the experience of getting 4-5 hours of sleep a night over weeks and weeks consistently took a toll on me. Looking back, I'm glad I had great support networks at the time giving me some grace - spouse, kids, boss at work, team at work, and others. To try and understand and overcome it I dove in deep on the topic. Got a sleep coach, read several technical and non-technical books on the topic, built and improved habits and routines, and thankfully my sleep returned to normal after 5 or 6 months. However, that experience changed me forever. I now take sleep very seriously, and now sleep better than I ever had before that experience. A few things I learned that I keep top of mind: 1. If you want to sleep like a baby, have a routine like a baby. Consistent time, consistent environment, do something relaxing before bed, and so on. 2. Environment matters a ton, and there are lots of great lists and bits of wisdom there. Focus on your environment, but make sure it works for you. 3. Study and understand the 3 P's of Insomnia. There are Predisposing factors, Precipitating factors, and Perpetuating factors. For me, I've always been Predisposed to fall asleep easily, but if I woke up past 2 or 3 am I would have a hard time keeping my mind quiet. Pandemic and 2020 were my Precipitating factors. And then, new to me in my learning journey, I had developed Perpetuating factors of worry/anxiety about the bad sleep CAUSING me to continue to sleep poorly. 4. Sleep is one of those paradoxical things - the harder you try at it, the worse you'll do. Not thinking too hard about it, not obsessing over it is actually a key to success. Harder done than said. 5. Sleep aid medicines of any type don't work and can become a false crutch. 6. If you struggle with sleep, or just want to improve it, there is help out there! I'm glad I got over my worry about talking about it with others and sought some help with it. Business and life can get challenging, and we can too easily sacrifice our sleep (intentionally or not). Good sleep is the foundation for good "everything else" and it's worth the effort to improve it. Reach out to someone (a doctor, a coach, me, anyone) because the rewards are worth it!

20

The Bygone Era of Legacy BI - Dashboards are Dead! If you've ever felt like you’re stuck in a time loop every time you open a dashboard, you're not alone. Dashboards, those flashy displays of KPIs and metrics, were supposed to be the knight in shining armor for data-driven decisions. Yet, here we are, years later, still shackled by their limitations. A stagnant adoption rate of 30%, an average report turnaround time of 4.5 days, and poor data experiences reported by 84% of frontline business workers make it painfully clear: DASHBOARDS are DEAD! ☠ Unpacking the Hidden Costs of Legacy BI Let’s get real for a moment. When was the last time your data team celebrated a dashboard? More likely, they’re cursing the operational nightmare it has become. Reports show that 92% of data workers spend their time on tasks outside their roles, while 68% of data teams lack time to implement profit-driving ideas. Your team isn't just buried in data—they’re drowning in it. Consider this: 50% of your data team’s headcount budget is wasted on remedial tasks instead of actual data analysis. Imagine hiring a Michelin-star chef only to have them spend their days peeling potatoes. The hidden costs of maintaining these dashboards extend far beyond just financials—they’re a productivity black hole. Read more in the link to CDO TIMES independent research on how you can leverage AI to achieve productivity gains and ultimately hand the keys over to your business teams to ask questions about your data and get petabyte scale data insight on an individual level. #aianalytics #aidriveninsights #cdo #datastrategy #cdotimes #AIstrategy #digitalstrategy

1

1 Comment

When I am learning about a company's #Operations and/or #Analytics, I take A LOT of notes. ⚡ Sometimes these notes are facts (like run sizes, MOQ's, locations of inventory, etc.) ⚡ Sometimes these notes are themes (missing process for x, confusion around y) ⚡ And sometimes these notes are direct quotes that capture the essence of a company Later, when I go back through these notes, it paints a pretty comprehensive contextual picture which helps to locate the data parts in the overall story of a company's Supply Chain. A while ago, I found something really special in my notes. Time and again - there were quotes about the leader from various members of the Team, both internal and external, that conveyed a deep admiration, respect, appreciation and .... care for this person. These people clearly expressed...and I do not think this is stretch..... love for this person. I have been very lucky to work with extremely high functioning Teams since I left corporate work. Teams that respect and enjoy each other and have AMAZING leadership. But this stood out. This was deeper. We all know great leaders are very rare. And this Team clearly had a special one and, even more impressive, THEY KNEW IT. Pretty freaking breathtaking to see. #Leadership #SupplyChain #Teamwork #Gratitude #CPGBrands What are your thoughts on leadership and what makes a great leader/Team?

15

4 Comments

Your Data Infrastructure (or lack thereof) Is Costing You BIG TIME This simple adjustment led to 110% more purchases and a 33% reduction in Cost Per Purchase. What was the change? Was it new creatives? Nope. Was it new copy? Nope. Was it simplifying their offer? Nope. The real game changer? Hiring our team of specialists to optimize the D A T A infrastructure & strategy Your Data Infrastructure is the foundation of ALL your marketing and growth activities & a robust first-party data strategy is crucial for your growth efforts. Here's the issue. . . Marketing and growth teams are stretched thin— managing clients, creating content, testing creatives, developing strategies—the list goes on. And what inevitably happens is that the very foundation of all of their efforts is left unmanaged. The consequences of neglecting your data infrastructure and data strategy include: > Erratic ad performance > Rising acquisition costs > Missed growth opportunities All of which directly impact your bottom line. After the release of iOS 14.5, brands have adopted a variety of strategies to overcome tracking and data challenges, typically falling into three main categories: Bucket 1: Ignore the need for effective first-party data systems & endure rollercoaster ad performance. Bucket 2: Use standard CAPI through platforms like Shopify. Better than nothing, but still limited and will never deliver optimal performance. Bucket 3: Rely on tracking, CAPI or attribution tools just because a well known podcast or thought leader recommends it. Spoiler alert. . .most of these tools are just glorified tag managers. Most of our clients come to us from Bucket #3—using tracking, CAPI, or attribution tools recommended by influencers or podcasts. When they put these popular solutions to the test against us, we've never been outperformed. Upgrade your data strategy today with audienceOS & join our clients who experience: ✅ More consistent ad performance ✅ Lower costs per acquisition ✅ Enhanced ability to scale Bottom line, an optimized data strategy & infrastructure are no longer 'nice-to-haves'—they’re essential for consistent, predictable growth. Don’t let your data infrastructure be your Achilles’ heel. Invest in your data infrastructure today & unlock massive growth and stability for your brand or clients. Comment below or send me a DM & let's see how much more profit we can help you squeeze out of your paid acquisition efforts.

3

Unstructured Data is Useless! During a recent interview on The MarTech Matrix Podcast, Obele (Brown-West) Hinsley, President of Tracer, emphasized the importance of data normalization in realizing the full potential of AI. She noted, "AI does not work unless you have structured data." Tracer empowers brands by organizing their data, providing the structure needed to leverage machine learning effectively. (Show links are in the comments) During the Super Bowl, Tracer harnessed the power of Google AI to analyze what makes an ad impactful. By normalizing the data and feeding it into AI, Tracer revealed that the most effective Super Bowl spots didn’t feature celebrities. They even uncovered patterns like word count, enabling clients to optimize their creative strategies. Tracer has spent nine years perfecting its technology, making structured data accessible to brands so they can fully harness the capabilities of AI for smarter, data-driven decision-making. #AI #MarketingData #DataNormalization #DigitalTransformation Cogent Collective Tom Barbaro Sean Simon

4

https://lnkd.in/gSAcfbxa Part 10 of my series Why Data Science Projects Fail is available today on INFORMS Analytics Magazine. Tom Davenport once said that "Models make the enterprise smarter, but models embedded in mission-critical business processes and systems make the enterprise more economically efficient." That is the end game of data science models--- generating significant business value and economic impact, and to achieve that, repeatably, reliably, and sustainably, requires that models are deployed as systems. Today's installment addresses the most challenging part of implementing model-based systems: Getting from Sandbox Model to Production System. Part 10 encapsulates many aspects of data science project complexity that were addressed in prior installments such as data availability, project management, and change management, and introduces new dimensions that determine how long it will take and how much it will cost to get from a sandbox model to a fully deployed and operational enterprise-grade production system. Levels of Dynamism, Integration, Mission-Criticality, and Problem and Model Complexity are addressed as four of the key determinants of how easy or difficult it will be to get your model from the sandbox to being a fully deployed, enterprise system, and integrated as part of the overall corporate technology ecosystem. The article provides examples of projects that span the continuum of relatively easy and inexpensive to deploy to extraordinarily complex and expensive to deploy. Successfully deploying and operating data science models as part of enterprise systems requires a holistic, comprehensive team of not only data scientists, but also data engineers and analysts, business process (domain SMEs) analysts, managers and executives, cloud engineers, ML Pipeline & Ops engineers, change management pros, software engineers and system integrators to fully realize the benefits such models promise.

21

2 Comments

I've seen a lot of complicated analyses looking to uncover the reality of whether PMAX is good or bad. None of them can fully answer the question, because the analysis misses the core principle of the product: PMAX is entirely context dependent. The answer is simple: PMAX effectiveness is a function of your context.

12

"Data has a massive credibility problem. You owned the P&L of a $100MM data business, you need to put that front and center so people know you've actually done something real and aren't full of shit. There are countless 'Data & AI Experts' out there who don't actually know anything." There was a very popular post yesterday from Vin Vashishta about how too many people put "Ex-Company" on their profiles, and should stop living in the past. Since my profile has "ex Nielsen P&L owner | ex unicorn SaaS VP of Data" I thought I'd provide some perspective from the other side. The advice at the top of this post was given to me by a fellow data founder when I first started posting online. Since then, for those that have followed me, I hope my content has made it clear that I've spent many years in the trenches actually doing things with data. But when people first look at my profile - knowing that I owned the P&L of a data business at Nielsen and that I was a VP of data at a unicorn startup gives me WAY more credibility than the generic "Data Leader" headline I had before. So if you see credibility as one of the most important things you need to quickly convey - pointing to the fact that you worked at highly credible companies is a good way to do that. Yes it's only one signal... There are a lot of idiots that work at all these great companies, and having a job doesn't mean you were actually good at it. But it's far less likely you're a complete fraud.

34

5 Comments