Stephen McDaniel | YakData

Netflix's first data scientist. Founder, YakData. Building production AI, forecasting and decision systems. Pilots impress. Production pays.

[Stephen McDaniel | YakData @ThatDataGuy007]

𝗖𝗮𝘀𝗲 𝗔𝟬𝟬𝟭: 𝗧𝗮𝗯𝗹𝗲𝗮𝘂 𝗞𝗲𝘆𝗻𝗼𝘁𝗲 𝟮𝟬𝟮𝟲 𝗗𝗮𝘀𝗵𝗯𝗼𝗮𝗿𝗱, 𝟵𝟳% 𝗗𝗲𝘀𝗶𝗴𝗻, 𝟳𝟵% 𝗜𝗻𝘁𝗲𝗴𝗿𝗶𝘁𝘆

AI is making dashboards faster and prettier. Someone still has to catch the integrity issues before the CFO, CEO, or board acts on the chart. Dashboard integrity audits for executive decisions at link in comment.

Yesterday, Tableau showed a polished executive dashboard during the conference keynote.

I ran it through the 77 Rules Engine.

The results: Design Score: 97% Integrity Score: 79% Combined Score: 85%

This is not a knock on Tableau. The dashboard looks excellent. But it illustrates the central risk in BI today: a dashboard can look executive-ready while still leaving decision risk exposed.

In this first Dashboard Forensics teardown, we look at 3 integrity issues: 1. LIVE PULSE without a freshness timestamp. 2. Missing time ranges on critical KPIs. 3. Percentages presented without base counts.

#Tableau #TableauConference #DataAnalytics #DataIntegrity #BusinessIntelligence #CFO

[Stephen McDaniel | YakData @ThatDataGuy007]

If you aren’t writing in machine language you are all losers! 😂

No, if I solve the problem before you and my client can understand what I left behind that is called winning…

Only very specific high value problems need to be super optimized.

Optimize time to value and ease of handoff usually wins for most client engagements.

[Stephen McDaniel | YakData @ThatDataGuy007]

@LoganARobison I can relate.

3 different highly rated dentists in Orlando. 3 cavities at one. 4 at another. 3 at the 3rd. Only one tooth identified across all 3 overlapped… got that one filled.

[Stephen McDaniel | YakData @ThatDataGuy007]

No

First replace the system most critical to the flow of your biz that is the worst fit.

Build momentum

Optimize the replacement as you fade out system 1 and scope replacing system 2

My guess is the most gains are not annual spend but rather optimizing flow/efficiency but both matter.

[Stephen McDaniel | YakData @ThatDataGuy007]

I think the nature of the problem is this…

$500-$1500 standard profit
3% test drives buy

$6,000-$7,000 marked up
1% test drives buy

So, 1,000 test drives a week at either $30k profit or at $65k profit

You have to find the massive volume dealer out a ways to get best deal.

Basing this on my father running a wholesale used car business in the 1990’s and what I hear from friends today.

[Stephen McDaniel | YakData @ThatDataGuy007]

@santiagomed This is why unit & feature testing under several common scenarios is critical.

You can ask it to map the system & then ask another AI to write unit tests. You should step in between the system map & the unit tests to see that they meet your common use cases.

Vibe vs Prod

[Stephen McDaniel | YakData @ThatDataGuy007]

@chamath I hate to inform you but the most advanced AI models write bad code as well…

The best part! “Your code has bugs!” Is the response 10 prompts later…

[Stephen McDaniel | YakData @ThatDataGuy007]

@spakhm My experience managing a Chinese remote dev team was quite positive.

They were careful to commit. Asked methodical questions and never overpromised. Quality was quite good and would definitely work with a well-chosen team from China again.

[Stephen McDaniel | YakData @ThatDataGuy007]

@VCBrags I lived Silicon Valley twice.

Reed Hastings is the nicest person in the Valley.

@pmarca and @bhorowitz were actually very nice to work for as well. Only one a-hole exec in the whole company, pretty amazing with the rate they grew.

[Stephen McDaniel | YakData @ThatDataGuy007]

Some of his portfolio companies? Also at Opsware/Loudcloud. I was one of the few preserved when it went from 800 to about 120 at the dot com implosion of 2001. Carnage as many contracts were lost by client implosions.

So he and Ben lived it. So did I. But since my wife lost her job in Santa Clara, we left the Valley for the East Coast.

[Stephen McDaniel | YakData @ThatDataGuy007]

Nearby Florida State Park

95 degrees but gorgeous!

Big gator in that lake! 14 footer!

[Stephen McDaniel | YakData @ThatDataGuy007]

@racerdogjack Stuck?

Or just optimizing their property values? Covert profiteering masked as compassion.

You find this throughout history…

See the $70B train to nowhere that barely made it out of LA. “You will kill the triple spotted beetle which only exists in this 4 square mile area!”

[Stephen McDaniel | YakData @ThatDataGuy007]

@Jason 😂😂😂

[Stephen McDaniel | YakData @ThatDataGuy007]

@anthony4ohio @unusual_whales Correct.

In fact, big business will be much slower to adapt. Resistance by entrenched ways.

It is why the Yellow Pages lost to Google and Blockbuster to Netflix.

[Stephen McDaniel | YakData @ThatDataGuy007]

@pmarca That was Loudcloud when I joined and when it was down to 100 in the implosion of all dot coms.

Netflix was the same in 2008.

[Stephen McDaniel | YakData @ThatDataGuy007]

The best AI models one day say smart things and two days later forget half the purpose of the threads I built with them.

Smart people rarely do that.

This is dangerous to think AI is like a smart person. It is a talking probabilistic model that can regress without warning to your strategy 2 quarters ago.

Just a warning. Seen it multiple times now.

[Stephen McDaniel | YakData @ThatDataGuy007]

@Jeff_2542 @Stephen19718352 Smaller towns in Italy we found amazing ex Michelin chefs running family restaurants that were stellar. Entire meals for 2 with wine and dessert for $100!

[Stephen McDaniel | YakData @ThatDataGuy007]

@awscloud 💯💯💯💯💯💯💯💯💯💯

[Stephen McDaniel | YakData @ThatDataGuy007]

@FreddyLA7 Many wonderful college nights there…

[Stephen McDaniel | YakData @ThatDataGuy007]

@AdrianDittmann It is so safe it will prevent you from asking remedial Biology questions.

Ridiculous. Super intelligence! 😂