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

MDSCON 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 Trendy 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 wonderful insights, instruments, and greatest practices to find out about.

In case you couldn’t make it, or in case you have been there and missed these classes, listed below 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 underneath an hour, we went from defining what information literacy actually is, to how analysts can go from glorified middlemen to leaders who empower their total group.

As Lauren Anderson from Okta put it, an organization’s information literacy is its capability to learn, talk, and derive insights from information clearly and effectively. And Gabi Steele, co-founder of Information Tradition, outlined what information 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 Trendy 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 information to creating dashboards to coordinating with different stakeholders to assist them entry and use information, analysts can typically really feel like middlemen. However they’ll additionally rework their organizations by main the way in which to information literacy.

Archer is aware of how Fivetran achieves this: by at all times asking the questions, are inside stakeholders really (a) adopting BI and information instruments and (b) utilizing them usually?

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

  • Including touchdown pages with prime 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 information merchandise
  • Champion self-service analytics inside their groups.

And to allow common use, they give attention to enabling insights by:

  • Demoing new instruments and providing focused trainings to related groups
  • Selling and incentivizing classes internally. For instance, via Workplace Hours that assist enterprise individuals suppose like information analysts via 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 information 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 information analysts” — a group of subject material consultants who socialize their findings and assist different stakeholders and groups — was an aspirational function and idea for a lot of.

Constructing a Tradition of Information Lovers

After a terrific first session on information literacy, we launched straight into one other session on information love — particularly, how will you get everybody in your group to like information?

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

Regardless of the session’s title, it wasn’t actually about getting your group to understand information. Somewhat, as Nelson stated, “Create a tradition the place nearly all of the persons are utilizing information to make selections, not simply utilizing information.” On this session, he talked about how constructing a tradition round information is basically about empowerment and belief. Information is in all places these days and information literacy is an important ability, however literacy isn’t nearly educating individuals how you can analyze information — it’s about giving them the flexibility and autonomy to make selections on that information.

Listed below are the highest 5 takeaways from this session:

  • Take into consideration information literacy when it comes to information participation, not information: At present, once we discuss constructing information literacy, we frequently 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 information literacy when it comes to the expertise adoption life cycle. In most organizations, the innovators and early adopters are utilizing information efficiently, but when we need to have conversations round information, we now have to penetrate previous that 16-18% of individuals to succeed in the early majority, late majority, and laggards. Solely then will nearly all of the group converse 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 information literacy wants each visualization and storytelling: Information dashboards and visualization are a gateway drug for getting individuals to dip their toe in information, nevertheless it’s necessary to not cease there. After we decide round information, we have to interact each components of our mind — each information visualization and information storytelling. The story is vital for serving to individuals perceive what they’re really seeing.
  • Start on the finish, not the start: After they need to begin utilizing information, many organizations soar to organizing and presenting it. That’s an necessary step, however Nelson defined how information isn’t the tip purpose — making a choice with information is. Somewhat than saying, “Give me your information and I’ll go determine issues out”, ask “What does success seem like?” and work backwards to determine what questions you’ll want to reply, what metrics or visualizations you really need, and what they need to seem 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 need. (From Nelson Davis’ session at Fivetran’s Trendy Information Stack Convention.)
  • Push authority downward, relatively than information upward: At present the everyday decision-making mannequin is that an analyst creates information, passes it up the chain of command, and a pacesetter makes the choice. This isn’t perfect, because it separates these with essentially the most context on the issue (leaders) from these with essentially the most information in regards to the information (analysts). As an alternative, leaders ought to flip this decision-making mannequin — discover the selections that they don’t must 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, relatively than pushing information upward.
  • Pair autonomy with guardrails: Creating authority and autonomy is necessary, however be sure that it comes with guardrails. Don’t simply let individuals obtain information from dashboards — it will probably 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, be sure that information instruments like dashboards are well-designed so individuals can get the insights they want. Allow individuals with the appropriate instruments to discover and innovate, however pair them with guardrails to make sure security and consistency.

As information 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 selections as decision-makers.

Nelson Davis, President & Founding Companion at Analytic Vizion

Methods 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 remark that making data-driven selections has by no means been tougher.

“It is advisable have trustable, usable information,” he stated, however that is exhausting to realize in actuality. To discover the way it can nonetheless be achieved, he launched a crew 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 information leaders shared their very own tales and ever-evolving challenges, and the way they overcame them — or didn’t.

Listed below are our important takeaways from the session:

  • It’s exhausting to be data-driven due to the very nature and significance of knowledge. Information informs varied varieties of choices at varied ranges, “from operational to strategic,” as Rashmi stated. 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 information.
  • David identified a widespread fallacy of so-called data-driven groups: Many use information to substantiate what they already suppose and reinforce present decision-making, however that isn’t what it means to be data-driven. As an alternative, really data-driven organizations are “really prepared to make use of information to make new selections and be agile” — not fall extra in love with their concepts, he stated. We should search for “uncomfortable” information as a result of being uncomfortable is what makes organizations develop.
  • Rashmi’s largest challenges typically revolve across the a number of variations of reality that get created as information proliferates in a company. Her crew 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 stated.
  • Working with pace is a vital problem that everybody surfaced. However David identified the irony of at all times in search of a technical resolution, “overlooking the ability of people speaking to one another”. Describing how the people of knowledge at WW Worldwide tackled this, he stated, “We determined to satisfy as soon as per week and simply speak in regards to the information”. They could not have had all the info they wanted, nevertheless it allowed them to align on what selections may and couldn’t be made.
  • Rashmi, however, described the advantages of the tooling and infrastructure that Oldcastle has adopted. One among their techniques to cope with information issues was to “purchase into new instruments relatively than leaping to constructing ourselves”. We liked that Kelly echoed her sentiment on this — all of us want information 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 ought to be 10% larger. What provides?” Groups in all places battle with these types of issues, however a terrific information 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 a terrific 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 a terrific analogy, evaluating the data structure for an information and analytics system to the ground plan for an IKEA showroom.

At IKEA, every ground is laid out in a different way, with indicators above and arrows on the bottom guiding clients on pre-set paths.

  • The Showroom is prefer to an organization OKR dashboard, the place everyone seems to be led via 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 information warehouse or deep-dive dashboard, the place particular sorts 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 Trendy Information Stack Convention

As Callie and Jacob defined, creating a terrific 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 resolution for establishing the data structure for a brand new information system. As an alternative, it’s necessary to consider what several types of individuals want out of your organization’s information.

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


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


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

Callie and Jacob emphasised that whilst you suppose via 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 information.

  • 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 might be going to be interacting along with your information… Bake that into the planning of your info structure, relatively than attempting to face it up piece by piece as you go.

Callie White, Montreal Analytics

New Youngsters on the Trendy Information Stack Block

We kicked off Day 2 with Andrew from our crew talking in “New Youngsters on the Trendy Information Stack Block“, a session that includes 8 of the most well liked startups which might be altering the face of right this moment’s information stack. We’re honored to be have been included on this group, and we liked studying in regards to the wonderful issues that our fellow new children are constructing!

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

  • Shinji Kim, Founder & CEO @ Choose Star, an automatic information discovery platform that helps you perceive your information
  • 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 information accessible
  • Boaz Farkash, Chief Product Officer @ Firebolt, the cloud information warehouse for the quickest information 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 choice intelligence layer for insights discovery
  • Oz Katz, CTO & Co-Founder @ Treeverse, the crew behind the open-source information versioning software, lakeFS
  • Barry McCardel, CEO @ Hex, an analytics layer that’s the quickest solution to go from evaluation to affect
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 Youngsters on the Trendy Information Stack Block” at Fivetran’s Trendy 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 inside the information business. Sadiqah and Devina’s causes for founding the motion have been deeply private, based mostly on the exclusion they typically felt inside a number of teams within the information business.

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

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

Black in Information helps communities of individuals of shade via:

  • Group outreach: Conducting workshops on careers in information for colleges and faculties
  • Jobs board: Performing as a market between proficient individuals and organizations aiming for range and inclusion of their groups and hiring

Even with this rising consciousness, nevertheless, there isn’t any “silver bullet” or in a single day resolution to constructing a various and inclusive crew — an information crew of the longer term.

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

Listed below are the three systemic strategies the audio system prompt to modernize your hiring and construct information groups of the longer term.

  • Appeal to various expertise:
    • Work with specialised recruitment teams and develop a deliberate method to rising 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 information 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 replicate 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 via the recruitment course of.
    • Ensure your group is hiring certified individuals of shade and never simply any individual of shade to fill your quota.
    • Assembly the crew earlier than a candidate accepts a job provide. “There’s a retention drawback with individuals of shade,” stated Devina; assembly colleagues and understanding the house a candidate is about to enter may help them make their choice.
    • Guarantee hiring practices are embedded in enterprise methods and have accountable targets and success metrics.
  • Retain various expertise:
    • Think twice on the way you onboard workers members. “Folks suppose the largest drawback is attracting and hiring,” stated 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 targets.
    • Being open about ethnicity make-up and the group’s plans on bettering this.

Breaking Your Information Workforce Out of the Service Entice

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

As the primary information analyst at a number of corporations, Emilie has skilled the “cycle of unhappiness” that many information groups discover themselves in. That is the service entice, the place a crew is trapped in endless requests for creating stats and proving their work, relatively than specializing in driving affect.

If we spend all of our time answering questions, we’ll by no means ship insights.

Emilie Schario, Netlify

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

As Emilie defined, “The actual trigger — information 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 affect.

Somewhat than simply answering questions, listed below are the 5 sorts of work {that a} information crew really ought to be doing:

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

So how will you get out of the service entice and begin taking over significant, impactful information work? Cease transferring proper to left, and begin transferring left to proper:

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

Make information straightforward to entry, determine what information may be self-serve, floor insights publicly, experiment and educate to construct information information, after which serve the crew requests that haven’t been addressed by your information processes.

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

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

In the long run, you’ll know in the event you’re profitable if different individuals within the firm are advocating for the info crew as a result of they see the worth of its work. As Emilie defined, “Don’t fall into the entice of measuring your information crew by the variety of dashboards you’ve constructed. These are vainness metrics they usually’re not centered on affect. You need to measure affect, not output.”

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

Header picture courtesy of MDSCON 2021.

Leave a Reply

Your email address will not be published. Required fields are marked *