top of page
  • Writer's pictureMonica Kay Royal

Event Review: Semantic Layer Summit 2022

This was an event that not only included data, analytics, and data literacy, but also included some new topics of the semantic layer and data mesh. It was great to hear from all of the top data leaders, technologists, and industry experts, some of which I had the pleasure of meeting at prior events.


There were two stages, the main stage and a tech stage, as well as a networking chat area and expo hall. After stopping by the networking session to introduce myself both in the #introduceyourself and #hiring channels, I headed over the the main stage for the first Keynote.


Keynote 1 | The Semantics of a Semantic Layer


David, our moderator of the day, started us off with some definitions. This was great because I noticed the need for this right away in the chat, including myself. Consider this the background to get everyone up to speed 🤓


Firstly, the word semantic means relating to meaning in language or logic. Ok!

A Semantic Layer is used in the business world when dealing with data and brining together a common language so everyone knows what the heck everyone else is talking about. Great!



The layer itself is pretty unique in that it does not live in the applications or the visualization tools themselves, but sits alone and acts as a glue to be used by data professionals to build out models, analytics, and products.


The extra cool thing that was shared about this layer is that the user does not need to know or understand where the data is coming from, so it is easier to use. However, from my experience, I believe that having knowledge of where the data is coming from is actually quite valuable when performing analytics and building out analytical solutions because you are able to provide deeper insights if you know the origin story of it all.


IMO, what are your thoughts?

There is also the concept of a data catalog, where the documentation of this language mapping along with the metadata is documented. We all know that I love documentation!! (ex. auditor here) 😄


David shared 4 use cases for a semantic layer:

  • Cloud Analytics Optimization

  • Enterprise Metrics Store

  • Bridging AI and BI

  • OLAP modernization


The last term discussed was a Data Mesh, which is an organization principle that is used to deliver analytics to the organization but allows the business users / data stewards to control and create the data products.


If you cringed a little, you are not the only one. The problem with this model is that the data team is left out of the structure and they are not able to manage the interaction between different data domains which can result in the same chaos as with traditional self-service. David saves the day by explaining that this is where the semantic layer helps because of the glue and the data catalog! ‍🦸‍♂


Honestly, I still have some questions around this model because I have experienced similar structures where the data stewards are in charge of creating these business terms but in some cases they not complete, accurate, or frequency reviewed and updated as the business changes.

Again IMO, what are your thoughts?

I realize that I need to learn more here so I registered for an event focused on data mesh, Datanova: Taking Data Mesh from Theory to Practice. I will link to that review when it is ready.


Takeaways: Why invest in a semantic layer?

  • Everyone speaks the same (business) language

  • Everyone can ask questions (not just data engineers) using their preferred tools

  • Promotes a manageable 'data mesh' organizational style

  • Data governance is managed in one place for every query

  • 'Future proofs' your analytics stack


Keynote 2 | Designing and Executing a Scalable Enterprise Data Strategy


This was a panel, I love panels because of the relaxed setting and different perspectives shared.


Ram mentioned the importance of being data driven and technology enabled. You should apply analytics to drive intelligent data derived solutions and these solutions should be predictive and simple.


Vihdi gave some tips/tricks for organizing data teams. She said that you need to make sure that you are asking the right questions related to data and are using the right metrics to measure certain initiative. Half baked information leads to mistrust and the impact is not seen if the right metrics are not used.


Andrew shared how to drive the data driven culture by measuring and holding each person accountable and responsible. This accountability should be paired with an annual performance plan, such as 'protecting the privacy of your data' or else these things might get deprioritized. He also stressed the importance of people, processes, and technology. While technology is the easy part, people and processes are the things you have to really put in continued investments and focus. You want to make sure that you have well established processes that allow the people to be as efficient and leveraged as possible.


Keynote 3 | The Evolution of Data Architecture & The Importance of a Semantic Layer


This was also a panel, yay! And it was great to see some familiar faces 😊


I first met Bill Inmon just last week at a meetup event hosted by Joe Reis. Bill is the father of the data warehouse and he talked about the progression of the traditional data warehouse and the newer concept of a data lakehouse. It was an interesting discussion about IOT, textual data, and Bill shared some notable stories about the development (or lack there of) the IT / Business relationship over the years.


I first met Kirk Borne at the DATAcated Conference 2020. Kirk is an amazing person with a fascinating background in astronomy and loves talking about image and time series data. Kirk supported the semantic layer as it gives us the ability to use all these different technologies and tools together along with the ability to communicate with each other.


Benn enjoyed the idea of the semantic layer because you can forget the distinction of what type of data you are working with and it will still operate like it is a regular database, regardless of the hot mess that is under the hood. He related it to a UI (user interface).


Between these panels, at the end of the day, it all boils down to communication between people, processes, and technology! 😎


Tech Stage | Turbocharging your semantic layer with better data: behavioral data

Snowplow

As the leader in Data Creation, Snowplow empowers more than 10,000 organizations, including Strava, Autotrader, and Software.com to purposefully create behavioral data to unlock transformative AI and advanced analytics directly from their warehouse, lake or in a real-time stream.


With our open-core technology, teams can generate, govern and model high-quality, granular behavioral data within their own cloud. Equipped with AI and BI-ready data, teams can focus on creating pioneering data applications across their business.



Buckle in, this was my favorite session, I am a huge fan of behavioral data 🧠



The unique thing about this session was that they went into the concept of creating behavioral data, with the focus of usefulness, at scale.


Typical Semantic Layer Data:

  • Customer (demographic)

  • Sales (transactions)

  • Product (category)

  • Finance (revenue)


This is just the tip of the iceberg of the possibilities with the semantic layer. We also can find out the data that describes what individual people are doing minute by minute / second by second. We can even see what decisions people are making and how they are making these decisions. Sounds a bit creepy, but also super exciting and opens up enormous possibilities!


More data beats clever algorithms, but better data beats more data ~ Peter Norvig

If we focus on the customer, aside from just demographic data we can also look at engagement over time by product, by channel, by content, or by campaign. This is powerful data because we can take this and further make predictions for customer lifetime value models.


Check out these simple questions turned powerful:

  • Are customers making decisions based on price vs. How sensitive are customers to price changes

  • How many customers do we have vs. What is the likelihood we can retain our existing customers

  • What products are customers buying vs. What are customers likely to buy together as a bundle


Million Dollar Question: If creating behavioral data drives so much value, why don't more organizations do it?

Sadly, it comes down to the 80/20 rule. Data teams are still spending 80% of the work upfront in the preparation phases of the lifecycle and only 20% producing and sharing insights.


Additionally, as I eluded to previously, this data can be considered PII (personally identifiable information). This means there are privacy policies/procedures/regulations that you must adhere to in order to appropriately collect, use, and store this data. And sometimes companies rather not deal with formality of it all, thus missing out on all the cool things they can do.

What are your thoughts about behavioral data?

Main Stage | Scaling Data Literacy


Data literacy is the ability to read, work with, analyze and argue with data

Jordan is another cool dude, the 'Godfather of Data Literacy' and author of an amazing set of books: Be Data Literate, Be Data Driven, and Be Data Analytical. Both he and I can talk about data literacy until our faces turn blue, but here are some highlights:


Data is just data, it is not a panacea. We need something that brings it to life and transform it into something powerful (i.e. insights). You may ask how can I get better at these things, or how can that translate into making better decisions. It all boils down to data literacy and a data culture.


Data culture includes:

  • Data Fluency: you must speak a common language with data

  • Iteration: lose the mindset that data and analytics is a pot of gold at the end of the rainbow

  • Data DNA: become the data, use it more

  • Data Community: do things with data together

  • Learn Fast: data moves fast

  • Data Skepticism: this is NOT meant in bad way. You need to understand that there is more to everything. As an ex auditor, I was taught to be skeptical. It's the same with data, and Jordan agrees 🙂


Megan provided some great facts. Did you know that only 21% of employees reported confident in their data literacy. That's hecka low! Organizational data literacy only happens when employees work together to learn about, access, work with, analyze, and communicate data in context.


Megan actually experienced the impact of data literacy during the pandemic. She worked in an environment with high in-person productivity. So when she couldn't pop into someone's office real quick to ask a question, things rapidly changed. She has an awesome perspective of data, her best quote:


Data are magic and I want it now!

Chad had a very interesting perspective, he said that data is a network composed of semantic elements. The more entities, events, relationships, and attributes that are created, the more network connections are formed within our data and the more questions we can answer. He was taking us through the old school concept of an ERD (entity relationship diagram) and states that this is the fundamental principle to data literacy.


Entities = Nouns (the state of the entity right now)

Events = Verbs (details about when something meaningful occurred)

Attributes = Adjectives (describes something about the noun or verb)

and ... Relationships help us understand how nouns and verbs interact

(this is where the questions are answered)


Main Stage | Business Intelligence & Analytics


Jon gave us some practical keys to success:

  • Lead with empathy

  • avoid shiny objects

  • think like a business owner

  • proof of concepts are good, but always think about scale

  • set expectations for test and learn, it is a continuous evolution

Biggest takeaway: make sure you know what you are trying to solve, then bring the solutions; not the opposite direction.


#1 driver of success is willingness to work through adversity


Andrea provided us with a some strategies for winning:

  • Avoid False Starts

  • Analytics transformation should be gradual

  • Connect investments with generated value


Book Recommendation: Data Analytics Made Easy, Packt, 2021

Tech Stage | Competitive Advantage Through Self-Service BI at Enterprise Scale

Tableau

As the market-leading choice for modern business intelligence, the Tableau platform is known for taking any kind of data from almost any system, and turning it into actionable insights with speed and ease.


Did you know that there is a cost of curiosity? No, it is not lives 🐱


The cost of curiosity is the time, steps, effort, and complexity involved with asking a question of your data in order to derive an insight.


This was not covered during the presentation in detail, but if you are curious about the difference between visual analytics and spreadsheet analytics, check out The Beautiful Science of Data Visualization by Ashley Howard Neville. Spoiler alert! visual analytics are better 🤭


Tech Stage | Building the best analytics team in times of uncertainty

DIGITAL ANALYTICS ASSOCIATION

Fostering community, advocacy, and professional development that empowers you and your team to deliver value through analytics.

  • Be transparent

  • Encourage feedback

  • Recognize contributions

  • Prioritize engagement

This was an extremely short session, only 15 minutes! But I was able to grab the key points, and my takeaway from it all is to make sure to COMMUNICATE!


Closing Keynote | Actionable Insights for Everyone


One last panel to close out the day 😃


David asked each participant to bust out their crystal balls and predict what is to come in the future 🔮


Colin sees ML and AI moving into more areas of the business, and the partnership with the business will become strong 💪🏻


Greg sees a lot more of the data sharing capabilities, using the right data at the right time to the right people. He also hopes for an open source explosion in the data space for data engineering world in particular, and that data literacy stays hot 🔥


Jenni sees the progression of descriptive, prescriptive, ML/AI capabilities and assures us that black boxes will disappear and be replaced with explainable AI. ⬛


Mark actually doesn't like the word data literacy, but does see that it will increase in fame. He also hopes to see more data product management and help with the data lifecycle 🚴🏻‍♀️


David officially closes things out by agreeing with Colin's ML/AI prediction and expects that no one will even realize they are using ML or AI 🎩🐇


In Conclusion

Kudos to all those that were involved in providing such a wonderful and educational event!


Brought to you by:


Participating partners:


Many of the speakers today were featured in Making Insights Actionable with AI and BI - collective advice from 20+ data leaders. You can download it here.


All sessions are available on demand here.



Happy Learning!!

Recent Posts

See All

コメント


bottom of page