Tuesday, July 23, 2024
HomeBig DataMDSCON 2021: Learnings and Insights from Our Favourite Periods - Atlan

MDSCON 2021: Learnings and Insights from Our Favourite Periods – Atlan

We had the prospect to sponsor and attend the Fashionable Information Stack Convention (MDSCON) 2021, and it was a blast! With two days of occasions, and 4 talks occurring concurrently in each time slot, there have been tons of fantastic insights, instruments, and greatest practices to study.

In case you couldn’t make it, or in case you had been there and missed these classes, listed here are a few of our favourite talks and what we discovered from them.

How Analysts Are Reworking Information Literacy at Their Firm

This was one of many first classes at Fivetran’s MDSCON, and we couldn’t have been extra excited to take part and study from all the info veterans there. 4 analytics leaders from 4 totally different corporations — Okta, Asurion, Information Tradition, and Fivetran — shared their insights on How Analysts Are Reworking Information Literacy at Their Firm.

In beneath an hour, we went from defining what knowledge literacy actually is, to how analysts can go from glorified middlemen to leaders who empower their complete group.

As Lauren Anderson from Okta put it, an organization’s knowledge literacy is its potential to learn, talk, and derive insights from knowledge clearly and effectively. And Gabi Steele, co-founder of Information Tradition, outlined what knowledge tradition actually is—and what it isn’t. We liked it a lot, we made ourselves this mild reminder:

Inspired by Gabi Steele's session at Fivetran's Modern Data Stack Conference (MDSCON 2021)
Impressed by Gabi Steele’s session at Fivetran’s Fashionable Information Stack Convention.

Each Brittany Metropolis from Asurion and Archer Newell from Fivetran spoke in regards to the a number of hats analysts typically must put on, based mostly on their very own expertise.

From organizing knowledge to creating dashboards to coordinating with different stakeholders to assist them entry and use knowledge, analysts can typically really feel like middlemen. However they’ll additionally rework their organizations by main the way in which to knowledge literacy.

Archer is aware of how Fivetran achieves this: by all the time asking the questions, are inner stakeholders truly (a) adopting BI and knowledge instruments and (b) utilizing them usually?

Right here’s how Fivetran strikes the needle on each, in accordance with Archer’s discuss. For adoption, they construction their BI software so it’s straightforward to navigate:

  • Including touchdown pages with high stories, new releases, and necessary hyperlinks
  • Creating analytics-certified stories
  • Figuring out licensed stories by way of folder construction and naming conference.

In addition they accomplice with energy customers to:

  • Construct higher knowledge merchandise
  • Champion self-service analytics inside their groups.

And to allow common use, they concentrate on enabling insights by:

  • Demoing new instruments and providing focused trainings to related groups
  • Selling and incentivizing classes internally. For instance, by means of Workplace Hours that assist enterprise individuals assume like knowledge analysts by means of context and fast insights.

What we liked in regards to the session — like so many different attendees — was the audio system’ descriptions of the very actual issues knowledge groups and analysts face. Gabi talked in regards to the 4 totally different levels of knowledge maturity in an organization, with “dashboard purgatory” typically being a consequence of maturity for some. In the meantime, Lauren’s “citizen knowledge analysts” — a group of material specialists who socialize their findings and assist different stakeholders and groups — was an aspirational position and idea for a lot of.

Constructing a Tradition of Information Lovers

After an awesome first session on knowledge literacy, we launched straight into one other session on knowledge love — particularly, how are you going to get everybody in your group to like knowledge?

This session, “Constructing a Tradition of Information Lovers: A Plurality of Views“, featured Nelson Davis (President & Founding Accomplice at Analytic Vizion) and Veronica Zhai (Principal Product Supervisor, Analytics at Fivetran).

Regardless of the session’s identify, it wasn’t actually about getting your group to understand knowledge. Moderately, as Nelson mentioned, “Create a tradition the place the vast majority of the persons are utilizing knowledge to make choices, not simply utilizing knowledge.” On this session, he talked about how constructing a tradition round knowledge is de facto about empowerment and belief. Information is in every single place these days and knowledge literacy is a vital ability, however literacy isn’t nearly educating individuals the way to analyze knowledge — it’s about giving them the flexibility and autonomy to make choices on that knowledge.

Listed here are the highest 5 takeaways from this session:

  • Take into consideration knowledge literacy by way of knowledge participation, not data: In the present day, after we discuss constructing knowledge literacy, we regularly default to specializing in the info hierarchy of wants: warehousing, cleansing, analytics, ML/AI, and so forth. Truly, Nelson defined, it’s higher to consider knowledge literacy by way of the expertise adoption life cycle. In most organizations, the innovators and early adopters are utilizing knowledge efficiently, but when we wish to have conversations round knowledge, now we have to penetrate previous that 16-18% of individuals to succeed in the early majority, late majority, and laggards. Solely then will the vast majority of the group communicate the language of knowledge.
Data Science Hierarchy of Needs - MDSCON 2021
The Information Science Hierarchy of Wants Pyramid. (Supply: “The AI Hierarchy of Wants”, Monica Rogati.)
Technology Adoption Life Cycle - MDSCON 2021
Rogers’ Bell Curve of expertise adoption. (Supply: Wikipedia.)
  • Constructing knowledge literacy wants each visualization and storytelling: Information dashboards and visualization are a gateway drug for getting individuals to dip their toe in knowledge, nevertheless it’s necessary to not cease there. After we decide round knowledge, we have to interact each elements of our mind — each knowledge visualization and knowledge storytelling. The story is important for serving to individuals perceive what they’re truly seeing.
  • Start on the finish, not the start: Once they wish to begin utilizing knowledge, many organizations leap to organizing and presenting it. That’s an necessary step, however Nelson defined how knowledge isn’t the tip aim — making a call with knowledge is. Moderately than saying, “Give me your knowledge and I’ll go determine issues out”, ask “What does success appear like?” and work backwards to determine what questions it’s worthwhile to reply, what metrics or visualizations you really want, and what they need to appear like.
From Nelson Davis' session at Fivetran's Modern Data Stack Conference (MDSCON 2021)
Don’t begin with the bricks. Begin with the home and work backwards to determine what bricks you really want. (From Nelson Davis’ session at Fivetran’s Fashionable Information Stack Convention.)
  • Push authority downward, reasonably than knowledge upward: In the present day the everyday decision-making mannequin is that an analyst creates knowledge, passes it up the chain of command, and a frontrunner makes the choice. This isn’t excellent, because it separates these with essentially the most context on the issue (leaders) from these with essentially the most data in regards to the knowledge (analysts). As an alternative, leaders ought to flip this decision-making mannequin — discover the selections that they don’t have to make, and empower others to take them over. Get to a spot the place leaders push authority and context downward to those that even have the info, reasonably than pushing knowledge upward.
  • Pair autonomy with guardrails: Creating authority and autonomy is necessary, however guarantee that it comes with guardrails. Don’t simply let individuals obtain knowledge from dashboards — it could turn out to be a safety danger. Then individuals can take the info, put it in Excel, and create their very own numbers. As an alternative, guarantee that knowledge instruments like dashboards are well-designed so individuals can get the insights they want. Allow individuals with the proper instruments to discover and innovate, however pair them with guardrails to make sure security and consistency.

As knowledge turns into extra and ubiquitous, the selections are going to turn out to be extra data-driven, no matter whether or not we maintain onto these choices as decision-makers.

Nelson Davis, President & Founding Accomplice at Analytic Vizion

Find out how to be Information-Pushed When the Information Isn’t There

This one obtained very actual, in a short time. Kelly Kohlleffel, VP of GTM at Hashmap, opened it with the commentary that making data-driven choices has by no means been tougher.

“It’s essential to have trustable, usable knowledge,” he mentioned, however that is onerous to attain in actuality. To discover the way it can nonetheless be carried out, he launched a workforce of veteran panelists — David Cohen, VP of Information Analytics at WW Worldwide; Rashmi Agrawal, an analytics chief at Oldcastle APG; Paul Hardman, Senior Director, Information Operations at Arlo Applied sciences.

What was nice in regards to the panel dialogue was how these knowledge leaders shared their very own tales and ever-evolving challenges, and the way they overcame them — or didn’t.

Listed here are our important takeaways from the session:

  • It’s onerous to be data-driven due to the very nature and significance of knowledge. Information informs numerous sorts of choices at numerous ranges, “from operational to strategic,” as Rashmi mentioned. For Paul and Arlo Applied sciences — whose roughly 20 million gadgets generate greater than 1 GB of knowledge per thirty days every — the largest problem is to acquire well timed and related knowledge.
  • David identified a frequent fallacy of so-called data-driven groups: Many use knowledge to substantiate what they already assume and reinforce current decision-making, however that’s not what it means to be data-driven. As an alternative, really data-driven organizations are “truly keen to make use of knowledge to make new choices and be agile” — not fall extra in love with their concepts, he mentioned. We should search for “uncomfortable” knowledge as a result of being uncomfortable is what makes organizations develop.
  • Rashmi’s greatest challenges typically revolve across the a number of variations of fact that get created as knowledge proliferates in a company. Her workforce focuses on governance, however that comes with the added problem of retaining particular person autonomy and freedom. “We would like centralized governance however to maintain federated analytics,” she mentioned.
  • Working with velocity is a important problem that everybody surfaced. However David identified the irony of all the time in search of a technical answer, “overlooking the ability of people speaking to one another”. Describing how the people of knowledge at WW Worldwide tackled this, he mentioned, “We determined to satisfy as soon as every week and simply discuss in regards to the knowledge”. They could not have had all the info they wanted, nevertheless it allowed them to align on what choices might and couldn’t be made.
  • Rashmi, however, described the advantages of the tooling and infrastructure that Oldcastle has adopted. One in all their techniques to take care of knowledge issues was to “purchase into new instruments reasonably than leaping to constructing ourselves”. We liked that Kelly echoed her sentiment on this — all of us want knowledge merchandise, and we want them rapidly.

Why Don’t These Numbers Match?

“Finance says this income quantity is appropriate –– however Gross sales says the quantity must be 10% larger. What provides?” Groups in every single place battle with these types of issues, however an awesome knowledge system can resolve them.

Within the final slot of the day, Montreal Analytics blew our minds with an superior presentation on the significance of constructing an awesome info structure — “Why Don’t These Numbers Match? Why Data Structure Issues in Analytics“. This session featured two of the corporate’s Senior Analytics Consultants, Callie White and Jacob Frackson.

Callie and Jacob centered their session round an awesome analogy, evaluating the knowledge structure for an information and analytics system to the ground plan for an IKEA showroom.

At IKEA, every flooring is laid out in another way, with indicators above and arrows on the bottom guiding prospects on pre-set paths.

  • The Showroom is prefer to an organization OKR dashboard, the place everyone seems to be led by means of the identical info in the identical method.
  • The Markethall is sort of a division or persona-level dashboard. It has totally different metrics in it (e.g. totally different couches and lamps), that are contextualized to a particular area (e.g. the kitchen space).
  • The Self-Serve Furnishings Space is sort of a knowledge warehouse or deep-dive dashboard, the place particular forms of individuals (often analysts) have full flexibility to entry no matter they want, although they should know what they’re in search of upfront.
Callie White and Jacob Frackson's session at Fivetran's Modern Data Stack Conference (MDSCON 2021)
From Callie White and Jacob Frackson’s session at Fivetran’s Fashionable Information Stack Convention

As Callie and Jacob defined, creating an awesome info structure takes a bit of additional time to arrange, nevertheless it saves a ton of effort and time for analysts and different customers in the long term. There’s nobody answer for organising the knowledge structure for a brand new knowledge system. As an alternative, it’s necessary to consider what various kinds of individuals want out of your organization’s knowledge.

Listed here are some nice guiding questions that Callie and Jacob gave to assist anybody work out their excellent info structure:


  • What are your personas? How will these change over time?
  • Who’re your stewards / energy customers inside these personas? What personas must be prioritized given their energy customers?
  • What language do these personas use? Can we conform to that commonplace or does that battle with our bigger structure?
  • What’s the software of the info or info? Is the applying extra technical or extra basic utilization?


  • How are you going to point out timeliness or accuracy of your knowledge? How are you going to point out which dashboards and tables are updated and that are deprecated?
  • How can customers perceive the context round knowledge or visualizations? How are you going to guarantee they’re decoding knowledge appropriately? Is there language that you should utilize that’s contextualized sufficient to precisely convey info?

Callie and Jacob emphasised that when you assume by means of these questions, it’s necessary to recollect the broader consumer expertise, or the way in which that folks throughout your group will work together along with your knowledge.

  • What’s the consumer expertise?
  • What’s the maintainer expertise?
  • How can we simplify or enhance these with info structure?

Take into consideration the totally different flooring and personas which can be going to be interacting along with your knowledge… Bake that into the planning of your info structure, reasonably than making an attempt to face it up piece by piece as you go.

Callie White, Montreal Analytics

New Children on the Fashionable Information Stack Block

We kicked off Day 2 with Andrew from our workforce talking in “New Children on the Fashionable Information Stack Block“, a session that includes 8 of the most popular startups which can be altering the face of at present’s knowledge stack. We’re honored to be have been included on this group, and we liked studying in regards to the superb issues that our fellow new youngsters are constructing!

Right here’s everybody who spoke on this session and the way they’re remodeling how we work with knowledge. Don’t overlook to take a look at their websites and tech!

  • Shinji Kim, Founder & CEO @ Choose Star, an automatic knowledge discovery platform that helps you perceive your knowledge
  • Andrew Ermogenous, Head of North America @ Atlan, the collaborative workspace for contemporary data-driven groups
  • Nicholas Handel, Co-Founder & CEO @ Remodel, the primary “metrics retailer” to make knowledge accessible
  • Boaz Farkash, Chief Product Officer @ Firebolt, the cloud knowledge warehouse for the quickest knowledge experiences
  • Arjun Narayan, CEO & Co-Founder @ Materialize, a streaming database for real-time analytics that delivers incremental view updates
  • Ajay Khanna, Founder & CEO @ Tellius, an augmented determination intelligence layer for insights discovery
  • Oz Katz, CTO & Co-Founder @ Treeverse, the workforce behind the open-source knowledge versioning software, lakeFS
  • Barry McCardel, CEO @ Hex, an analytics layer that’s the quickest strategy to go from evaluation to impression
Andrew Ermogenous from Atlan speaking at the "New Kids on the Modern Data Stack Block" at Fivetran's Modern Data Stack Conference (MDSCON 2021)
Andrew talking on the “New Children on the Fashionable Information Stack Block” at Fivetran’s Fashionable Information Stack Convention

Modernizing Your Hiring to Construct Information Groups of the Future

Day 2 began with this highly effective and extremely related session by Sadiqah Musa and Devina Nembhard, each analysts and co-founders of Black in Information. It’s a collaborative motion for individuals of shade that goals to advertise equality of alternative and illustration throughout the knowledge business. Sadiqah and Devina’s causes for founding the motion had been deeply private, based mostly on the exclusion they typically felt inside a number of teams within the knowledge business.

Black in Data - Fivetran's Modern Data Stack Conference
Black in Information goals to raise individuals of shade by means of studying and improvement, whereas offering a supportive community.

Sadiqah believes the way forward for an business will likely be present in its individuals, not simply within the business itself. Variety in groups has repeatedly confirmed to be a driver of fascinating, profitable enterprise outcomes — from higher groups to raised monetary efficiency.

Black in Information helps communities of individuals of shade by means of:

  • Neighborhood outreach: Conducting workshops on careers in knowledge for faculties and faculties
  • Jobs board: Appearing as a market between proficient individuals and organizations aiming for variety and inclusion of their groups and hiring

Even with this rising consciousness, nonetheless, there isn’t any “silver bullet” or in a single day answer to constructing a various and inclusive workforce — an information workforce of the long run.

Sadiqah Musa and Devina Nembhard speaking at Fivetran’s Modern Data Stack Conference
Sadiqah Musa and Devina Nembhard talking at Fivetran’s Fashionable Information Stack Convention

Listed here are the three systemic strategies the audio system instructed to modernize your hiring and construct knowledge groups of the long run.

  • Appeal to various expertise:
    • Work with specialised recruitment teams and develop a deliberate method to growing your pool of various expertise.
    • Firms within the UK mandatorily publish their gender hole, however some additionally publish their ethnicity pay hole voluntarily. This begins with collating knowledge about ethnic illustration and compensation inside your organization, and whether or not constructive efforts are being made to fill a spot which may emerge.
    • Perceive your model notion, and take into consideration how individuals of shade see it. Do your web site and supplies mirror inclusivity and illustration?
    • Construct and preserve transparency. She recommends promoting the wage bands all through the corporate, and being clear about them with potential hires too.

We have to know that we’re not going to be paid any lower than anybody else.

Devina Nembhard, Black in Information

  • Rent various expertise:
    • Incorporate various interview panels by means of the recruitment course of.
    • Be certain that your group is hiring certified individuals of shade and never simply any individual of shade to fill your quota.
    • Assembly the workforce earlier than a candidate accepts a job provide. “There’s a retention drawback with individuals of shade,” mentioned Devina; assembly colleagues and figuring out the house a candidate is about to enter may also help them make their determination.
    • Guarantee hiring practices are embedded in enterprise methods and have accountable aims and success metrics.
  • Retain various expertise:
    • Consider carefully on the way you onboard employees members. “Individuals assume the largest drawback is attracting and hiring,” mentioned Devina. However the onboarding is equally, if no more, necessary.
    • Assemble and work on constructing really inclusive environments and areas.
    • Conduct genuine anti-racism coaching.
    • Measurable diversity-focused enterprise aims.
    • Being open about ethnicity make-up and the group’s plans on enhancing this.

Breaking Your Information Crew Out of the Service Lure

We wrapped up our MDSCON 2021 with a tremendous discuss by Emilie Schario, Director of Information at Netlify, referred to as “Breaking Your Information Crew Out of the Service Lure“. And it wasn’t simply us — individuals had been raving within the chat in regards to the session.

As the primary knowledge analyst at a number of corporations, Emilie has skilled the “cycle of unhappiness” that many knowledge groups discover themselves in. That is the service lure, the place a workforce is trapped in unending requests for creating stats and proving their work, reasonably than specializing in driving impression.

If we spend all of our time answering questions, we are going to by no means ship insights.

Emilie Schario, Netlify

The “cycle of unhappiness” or service lure. From Emilie Schario’s session at Fivetran’s Fashionable Information Stack Convention 2021

As Emilie defined, “The actual trigger — knowledge groups are seen as service organizations… They’re not creating insights. They’re answering questions.” Servicing requests is a part of an information groups’ work, nevertheless it’s not their core mission — to create insights and drive impression.

Moderately than simply answering questions, listed here are the 5 forms of work {that a} knowledge workforce truly must be doing:

  • Operational analytics: Rising individuals’s entry to knowledge throughout the group.
  • Metrics administration: Serving to different groups take possession over their knowledge and making knowledge extra self-serve.
  • Information insights: Serving to individuals study customers from the corporate’s knowledge.
  • Experimentation reporting: Assessing whether or not experiments are resulting in elevated activation, conversion, and retention.
  • Servicing different individuals: Serving to different groups with knowledge requests and analyses

So how are you going to get out of the service lure and begin taking up significant, impactful knowledge work? Cease transferring proper to left, and begin transferring left to proper:

From Emilie Schario’s session at Fivetran’s Fashionable Information Stack Convention 2021

Make knowledge straightforward to entry, establish what knowledge could be self-serve, floor insights publicly, experiment and train to construct knowledge data, after which serve the workforce requests that haven’t been addressed by your knowledge processes.

If restructuring your knowledge workforce isn’t doable, there are nonetheless loads of issues you are able to do to construct a greater knowledge tradition.

  • Use Slack to construct knowledge literacy and share knowledge insights — e.g. create a #data-reads channel for related knowledge articles, and publicly share hyperlinks to new knowledge insights and findings.
  • Discover companions in different elements of the enterprise to construct visibility and achieve help — e.g. collaborate with the UX workforce on analysis or add an introduction to firm knowledge into onboarding.
  • Prioritize proactive insights by dedicating 2-3 hours weekly to deep knowledge work — e.g. if you happen to’re analyzing the pricing web page for a request, flip it into a possibility to have a look at customers’ paths to the pricing web page.

Ultimately, you’ll know if you happen to’re profitable if different individuals within the firm are advocating for the info workforce as a result of they see the worth of its work. As Emilie defined, “Don’t fall into the lure of measuring your knowledge workforce by the variety of dashboards you’ve constructed. These are self-importance metrics and so they’re not targeted on impression. You wish to measure impression, not output.”

This weblog was co-authored by Prukalpa Sankar, Sanjana Ramachandran, and Christine Garcia.

Header picture courtesy of MDSCON 2021.



Please enter your comment!
Please enter your name here

Most Popular

Recent Comments