Jun 8, 2018 · 25m · mad
10 Lessons from Building Data Driven Software // Raj De Datta, BloomReach (FirstMark's Data Driven)
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
Raj De Datta, CEO and Co-Founder of Bloomreach, presents ten core strategic lessons and architectural insights from building a data science-driven enterprise software platform at DataDrivenNYC, followed by an audience Q&A session with Matt Turck.
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 3.8% of the talking time here. How this is scored →
Matt as informed peer 0.8 Guest teaching 3.5 Guest disagreement 0.8 Matt pushing back 0.7
speaking balance: gold is Matt, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 13:20 Dismissing AI/ML Marketing Value
Raj forcefully dismisses market trends by stating that AI and ML have zero value in marketing and are merely noise that enterprise customers ignore.
Hardest push from Matt ▶ 18:36 Challenging the AI Marketing Value ClaimMatt Turck challenges Raj's premise that AI marketing holds no value, noting that AI positioning is actually crucial for engaging systems integrators and recruiting talent.
Biggest teaching moment ▶ 16:25 No Dedicated Data Scientists ReframeRaj reframes how software teams should be structured, educating the host on why Bloomreach employs zero standalone data scientists in favor of backend engineers who understand data science.
Matt holds his own ▶ 18:36 Synthesizing AI Marketing Utility for SIsMatt demonstrates high domain familiarity and sharp listening by synthesizing Raj's earlier talk points to catch a nuance regarding partner recruitment.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Matt as informed peer | Guest teaching | Guest disagreement | Matt pushing back | Why |
|---|---|---|---|---|---|---|
| Bloomreach Company Overview and Scale | 0 | 2 | 0 | 0 | This is a solo presentation segment where Raj De Datta introduces Bloomreach's scale and core data architecture. Because the host does not participate in this monologue section, host-side metrics are set to 0. Raj establishes authority by explaining how Bloomreach powers roughly 20% of US e-commerce transactions. | |
| Demonstration: AI-Powered Landing Pages | 0 | 3 | 0 | 0 | In this continuation of the monologue, Raj walks through AI landing page generation and autosuggest capabilities. The host does not speak, keeping host scores at 0. Raj educates the audience on technical trade-offs, noting that speed matters more than quality or algorithmic optimization. | |
| Demonstration: AI-Powered Search Results | 0 | 4 | 1 | 0 | Raj continues his presentation, explaining AI-powered search results and why AI models are inherently vertical-specific. Host participation remains 0 during this monologue. Raj offers mild contrarian criticism toward horizontal AI marketing terms like Einstein or Sensei. | |
| Demonstration: Sports Innovation with FC Bayern Munich | 0 | 3 | 0 | 0 | Raj outlines FC Bayern Munich's use of Bloomreach for sports innovation and stresses the importance of white-box tools for trust. The host does not speak during this section. Raj shares lessons about open ecosystems and building trust with enterprise clients. | |
| Stakeholders in the Digital Experience | 0 | 4 | 2 | 0 | Raj concludes his solo talk by listing final takeaways, including provocative claims about AI and ML having zero value in marketing. Host scores remain 0 due to the monologue format. Raj displays mild combativeness toward marketing hype and Silicon Valley talent wars. | |
| Fireside Q&A Session with Matt Turck | 5 | 5 | 2 | 4 | Matt Turck opens the fireside Q&A, asking targeted questions about team structure, open-source tech, and vertical expansion. Matt briefly pushes back by pointing out that AI marketing buzz does hold real value for partner ecosystems and recruiting. Raj provides detailed answers explaining why Bloomreach avoids hiring isolated data scientists and how they partner with systems integrators. |
Statements from this episode (20)
Bloomreach Touches 20% of All US E-Commerce Transactions
“We power almost 20% of e-commerce transactions in the U.S. At some point hit a Bloomreach server of, ah, of some kind.”
De Datta: Every enterprise requires three core digital experience platforms
“The way we see the world is more and more every enterprise has three core platforms that power their digital experience, or their digital business.”
Raj De Datta: Software applications and data platforms have no value without each other
“Bloomreach has, is always, you know, I struggle to describe it about whether it's a software company or whether it's a data company, because really one without the other has no value.”
Raj De Datta: Consolidated data sources create massive platform dependency risk
“There is arguably no bigger platform dependency than if your data largely comes from a consolidated set of sources”
Speed Matters More Than Quality or Optimization in Data Products
“Speed is more important than quality, which is more important than optimization.”
Cutting Autosuggest Latency Outperformed All Algorithm Accuracy Gains
“In, in the course of building the first versions of Autosuggest, they were, they returned data back at the level of about a 150 milliseconds. Today, they return data back at the level of 10 milliseconds. And the delta between that was more important than anyth…”
De Datta: Bloomreach algorithms determine search sort order without human intervention
“The algorithms are basically figuring out the sort order of the hoodies without any human intervention in the mix.”
De Datta: SaaS plus data science demands a new customer success model
“But today, because this is a handshake between our SaaS environment and the customer's data, that, that handshake requires a completely different, ah, model for customer success and support. Because what we have learned in being a data science plus SaaS compan…”
De Datta: All AI, machine learning, and data science is vertical-specific
“All data science And all AI and all ML is fundamentally vertically specific. Which is the dirty secret about AI and ML powered businesses. They are fundamentally vertical problems.”
Horizontal AI Suites Like Salesforce Einstein Are Just Marketing Fluff
“When we hear marketing terms like Einstein, or Sensei, or pick your other dead male inventor, That, that approach applied to a horizontal set of software problems simply doesn't make sense because the interesting stuff recruiting data or hiring data or our dat…”
Bloomreach powers FC Bayern Munich's digital apps and websites
“We power, ah, the, ah, soccer team Bayern Munich, and Bayern Munich, and this is not just the websites, this is the apps”
De Datta: Innovation mostly happens outside of your product
“And I think what I've learned from this is really that innovation mostly happens outside of your product.”
Algorithms Are Useless If Users Cannot Interpret and Trust the Outputs
“That at the end of the day, if people can't interpret it, if people can't make sense of it, if people can't trust it, then it really doesn't matter what you did.”
Raj De Datta: Data science software requires workflow or infrastructure integration
“At least in software, enterprise software, the value of data science is, is hugely valuable as a source of differentiation. But it's only valuable if presented along with either a workflow problem or an infrastructure problem.”
AI and Machine Learning Buzzwords Have Zero Value in Enterprise Marketing
“AI and ML has zero value in marketing.”
Raj De Datta: Data startups need not match tech monopoly salaries
“If you're an entrepreneur building a business around use of data and data science, You don't have to pay what the unnamed tech monopolists pay.”
Practical Enterprise ML is 20% Algorithms and 80% Systems Engineering
“So, you know, when you start getting into problem areas, you start to find that really that what you need to solve a practical problem is like, 20% algo and 80% systems, you know, in many cases.”
Bloomreach Employs Zero Dedicated Data Scientists
“Yeah, so first of all, we don't have any data scientists. We only have engineers who understand data science.”
Bloomreach Monetized Demands for AI Explainability as a New Analytics Product
“At a certain point, our customers kept asking us, how is the algorithm making a decision about what to do? And so what we did is we built an analytics tool and sold something to them. So we effectively turned the question into a product, and then monetized the…”
De Datta: Adobe is Bloomreach's largest competitor with a $1.5B business in their space
“Yeah, at this point, you know, the companies at a scale where I would say our largest competitor is the Adobe stack. So Adobe has about a one and a half billion dollar software business in our space, and they're the leader in our space, and we admire them a lo…”