Sep 14, 2015 · 24m · mad
Anthony Deighton, Qlik: Top 10 Requirements For Visual Analytics (Hosted by FirstMark Capital)
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
At DataDrivenNYC, Qlik CTO Anthony Deighton presents eleven essential requirements for modern visual analytics platforms and demonstrates Qlik Sense's self-service associative data capabilities.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Matt holds 1.1% of the talking time here. How this is scored →
Matt as informed peer 0.2 Guest teaching 0.1 Guest disagreement 2.1 Matt pushing back 0.1
speaking balance: gold is Matt, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 7:30 Stabbing yourself in the leg analogy
Deighton forcefully rejects visual tools without guardrails, comparing unguided user visualization directly to stabbing oneself in the leg.
Hardest push from Matt ▶ 22:08 Host intervenes to control Q&A flowMatt Turck steps in with 'Hold on, hold on' to enforce time limits and maintain order during audience question transitions.
Biggest teaching moment ▶ 21:54 Deighton reframes the NLP challengeIn response to an audience query, Deighton clarifies that processing language is not the real hurdle, but rather managing structured data once language is parsed.
Matt holds his own ▶ 21:22 Host steers session to audience Q&AMatt Turck efficiently transitions the presentation into audience Q&A while maintaining energy with humor about beer and pizza.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Matt as informed peer | Guest teaching | Guest disagreement | Matt pushing back | Why |
|---|---|---|---|---|---|---|
| Opening Remarks and BI Paradigm Shifts | 0 | 0 | 1 | 0 | Anthony Deighton opens the presentation with light banter about pizza and beer before explaining the ongoing shift from IT-driven reporting to user-driven cloud analytics. As this is a solo talk segment, the host does not participate. | |
| Visualization Beyond Pretty Pictures | 0 | 0 | 3 | 0 | Deighton challenges the industry assumption that pretty pictures solve BI problems, arguing that SQL queries fail to capture human questions accurately. Host scores remain zero during the guest presentation. | |
| Requirement #2: Unified Experience | 0 | 0 | 3 | 0 | Deighton uses a Macintosh analogy to advocate for tight UI-calculation integration and compares unguided visualization to stabbing yourself in the leg. Host participation is absent. | |
| Requirement #4: Flexible Data Models | 0 | 0 | 2 | 0 | Deighton criticizes traditional rigid schemas delivered to hapless users and advocates starting visual analytics with a search box rather than complex charts. Host scores are zero for this monologue. | |
| Requirements #6 & #7: Scalability and Governance | 0 | 0 | 2 | 0 | Deighton reframes the relationship between user empowerment and governance, contending users want trusted data frameworks rather than raw Excel exports. Host does not speak in this segment. | |
| Requirement #8: Open Platform Architecture | 0 | 0 | 3 | 0 | Deighton dismisses software vendors who build separate mobile or cloud versions of applications as silly, insisting users expect identical functionality across platforms. Host participation is zero. | |
| Requirements Summary and Live Demo Initiation | 0 | 0 | 2 | 0 | Deighton transitions into his live product demo while poking fun at executive reliance on red-green corporate dashboards despite common male colorblindness. Host does not speak. | |
| Live Demo: Natural Search and Dynamic Filtering | 0 | 0 | 2 | 0 | Deighton demonstrates Qlik's search and dynamic scatter plot clustering, contrasting automatic grouping against unhelpful Excel blue dots. Host scores remain zero during the demo monologue. | |
| Presentation Wrap-up and Stage Transition | 2 | 1 | 1 | 1 | Host Matt Turck takes stage to manage audience Q&A under strict time limits, while Deighton answers questions regarding NLP and predictive modeling. Host focus is purely on moderation rather than technical debate. |
Statements from this episode (18)
Deighton: BI industry is shifting from IT-driven to user-driven
“There's a, I think, again, a kind of fundamental shift in the BI industry that's, we're really right in the midst of, ah, today. And, ah, it's really a shift away from an IT-driven industry to a user-driven industry.”
Deighton: Vast majority of new startup data lives in cloud
“If you look at any new company today, and I bet any startup in this room, most, if, you know, if not a hundred percent, certainly the vast majority of the data that you're working with is sourced, begins, and lives in the cloud.”
Deighton: Platform providers will ultimately win over tech stack models
“We're moving away from this kind of stack model to a platform model, and this is not, ah, this is no big insight, this is something we see in every industry, a move away, ah, and where, where ultimately platform providers, ah, are the victors.”
Deighton: Key data insights often lie outside chosen query selections
“Oftentimes, ah, from a question perspective, the interesting insights in the data are actually found outside of your selection, ah, outside of your query.”
Deighton: Visual analytics value requires integrating interface and calculation engines
“And our view here is that no, in fact, the requirement here is really to bring these two worlds together, and you can't have one Without the other, or maybe slightly more generously, the real value comes when you bring the two together.”
Deighton: Analytics software must encode visualization best practices to prevent errors
“And with all due respect to people like Stephen Few who've written great books, the idea is to encode the book in the software. And so that it, when the user does the stupid thing, the result is something that's actually correct, and not rely on them to have r…”
Deighton: Data visualization without guide rails is like stabbing yourself
“So data visualization without guide rails, again, that's like stabbing yourself in the leg.”
Deighton: Hadoop's schema-on-read innovation hasn't reached front-end users
“I think one of the great innovations of Hadoop is this idea of schema unread, but that's not realized through to the front end, to the user itself”
Deighton: Visual analytics apps should start with search boxes, not charts
“The idea that users start most business questions from the perspective of a search is an important requirement, one that one should think very carefully about, especially for those in the room who are practitioners who are thinking about delivering application…”
Deighton: User activity plummets when response times exceed two seconds
“When a user waits more than about two seconds, two, two and a half seconds, they stop clicking. It's not a linear fall off. It, like, goes off a cliff.”
Deighton: Users export data to Excel only when lacking better software alternatives
“It's not that they don't want to, ah, you know, it's not that they desire export to Excel for some fundamental reason, they make good with that when they have no better alternative.”
Deighton: Expecting one analytics platform to satisfy all customer requirements is silly
“To believe that what we're going to solve from a platform perspective is going to meet every requirement of any customer who ever walked through the door is silly.”
Deighton: Software users want identical application experiences across PC and mobile
“No one wants a mobile version. They want the same version they had on their PC to be the one that works on their mobile device, to be the one that works on their big screen like this.”
Deighton: Users want identical software functionality on-premise and in the cloud
“No one wants a cloud version of their application. They want a version that runs on premise, and they want one that runs in the cloud, and there should really be no difference between those two.”
Deighton: Most business analytics work is storytelling rather than doing analysis
“And so much of what we do in the context of our work life isn't actually doing analysis. It's telling the story that the analysis, the results of that analysis.”
Deighton: Color saturation beats red-green coding for displaying differences in dashboards
“Ah, instead we use, ah, color saturation as a mechanism of showing difference, because it turns out that color saturation is A much better example, much better way of seeing difference.”
Qlik plans to integrate natural language processing without building underlying models
“So, natural language processing the simplest answer is yes. The longer answer is processing the language isn't the hard, I mean, that's the hard part. It's not one we're gonna do. Dealing with the language once you have it is something we're obviously very int…”
Deighton: Regulated industries lag retail in adopting modern visual analytics
“So you know, regulated industries are way behind. Even financial services, God forbid, Is a little bit further behind, ah, but you see other industries, ah, you know, retail is an example, pushing, you know, mobile decision making out to store managers who are…”