The art of exploring and explaining data

[@colenussbaumerknaflic4724]

Thank you to everyone who joined live, as well as those who are watching the recording! I always enjoy watching Mike present. For additional insight, I'll share a tech challenge behind the scenes: at around minute 32, Mike's laptop started threatening to reboot. Every time he needed to advance a slide, he had to stand, close the window directing him to force-quit, quickly advance the slide, then sit back down-all while speaking coherently! I was proud to watch him masterfully maintain composure during this unexpected hiccup. This was only possible given how well he knew his material. When we are adequately prepared, we can respond gracefully-if not perfectly-to almost any challenge that arises!

[@МаксимГальченко-ч7к]

Extremely useful. It was great story from great storytellers. This is good story how to create interesting story for management from boring numbers and how to use different type of diagrams. Great! Thank you!

[@storytellingwithdata]

Thanks for your kind words! No number is boring if you have figured out how to make it relevant for your audience. That's part of the magic when data storytelling is done well.

[@clevercoworker]

This is a great session, totally beyond my expectation. Thank you Mike & Cole for sharing with us

[@sudhakarnote]

I bought your book in the Amazon kindle and glanced in 5 minutes....real work....congrats

[@storytellingwithdata]

Hello, Cole here. Happy to hear that you enjoyed the book!

[@sudhakarnote]

​@@storytellingwithdataHow to contact you

[@AdolfoWatanabe]

Thanks for the lessons! Crystal clear explanation and very informative.

[@mohamedamineelhaouari5281]

That's what I need ! Thank you so much for this shared knowledge 🙏🏼

[@storytellingwithdata]

Glad it was helpful!

[@AbdulVajid-fz3vs]

Wow , We need more video like this , it improves our analytical and other skill ,[sry for my english]

[@mishmohd]

I'm excited for this video I had it in my watch list for a week - glad I saved it.

[@storytellingwithdata]

That’s awesome! So glad you enjoyed it.

[@rolfkarlsson9439]

Very well presented. Good example session from beginning to end. Thanks!

[@storytellingwithdata]

Thank you!

[@SrinivasRao-fg9em]

This is an amazing and helpful session. Thanks for sharing.

[@storytellingwithdata]

Cole here. You're welcome and thanks for your kind comment! Mark your calendar for our next live event-I'll be presenting (and reading) to grown-ups and kids of all ages: www.daphnedrawsdata.com/

[@BAKANEKO99]

Very insightful ❤

[@YuhShanChang]

Thank you for this amazing mini-workshop! I am the type of person who's rewatching again🤣 Just have a quick question, how does Mike come up with 40% viewership on 20:27.

[@MikeCisnerosSWD]

Hi! Great question. In working through the analysis, we had come to understand that limiting our potential list of shows to keep to those with an audience of 50% millennial or more wasn't going to yield enough viable candidates. It would be ideal to get as big of a chunk of our audience in our target demographic as possible, but 50% turned out to be too high of a bar to clear. Given this, we had to drop our cutoff point some. Logically, we wouldn't want to go any lower than 33% millennial, since that's not really targeting our chosen demographic at all. For the purposes of exploration, we chose a threshold between 33% and 50%, and decided that 40% was broad enough to start from. As it turns out, even that was a bit too high-you can see in the sci-fi and drama genres, we wind up keeping a couple of shows that are JUST barely below 40%, but have audiences and/or ROIs that are so good that it seemed foolish to exclude them over a few tenths of a percentage point. We often say that visualizing data is a blend of art and science; so too is exploring and analyzing data. We can use objective, hard-line, quantitative values as cut-off points or filters for our data, but we also have to be thoughtful about the actual objective we're driving toward. Also, keep in mind that any objective, quantitative values we use as filters were at some point selected arbitrarily by human beings. The TL;DR here is: 40% is a boundary we chose because it seemed a reasonable compromise between audiences that were majority-millennial (50%) and random-chance millennial (33.3%). It was a guideline, not a rule, however: looking at options just outside that boundary helped ID other potentially valuable shows to keep.

[@yzhu2230]

Great video, i am pretty new to this, can I please ask, how do you get the proposed value when comparing the current and proposed viewership, target audience and overall cost? Is there a formula on how to get the proposed value? Thank you so much.

[@MikeCisnerosSWD]

Thanks for asking. For time and clarity, we simplified and condensed some of the exploratory analyses in this mini-workshop. We decided to simply set benchmarks: compared to the larger set of shows at the beginning, we wanted: our smaller lineup to be similar to our hit show (and we found out in our analysis, that meant "find shows popular with millennials"); to have about 15-20% as many shows to manage; to get more viewers per dollar spent; and to have shows that aren't as expensive. We weren't fully able to meet that last requirement, as it turns out, but like the traditional cliche of "fast, cheap, good - you can only pick two," it's sometimes not viable to meet all of the original goals. What we were able to recommend to our leadership team met all of the other goals, and we tried to make a case for why that was the direction to go. So there's no magic formula here that aggregates or puts a numeric index value for each show across all of the dimensions we considered. One could do that, but it seems to me that doing so is maybe more effort than is necessary. Logic and common sense found us a reasonable collection of shows that met all or most of our goals individually, and made sense collectively. Having said that, you do have access to the dashboard and all of the underlying data, so if you want to try to build a numeric model that weighs and balances all of our stated goals, and generates an index value for each show, that might be an interesting experiment. I'll close with this thought, one that my long-ish history in analysis has led me to believe: in any exploratory or predictive analysis, there's a danger of getting fixated on finding the absolute, no-doubt-about-it, bulletproof answer for what to do next. To find this elusive solution, it's easy to get sucked in, and spend weeks or months building models, running simulations, considering more and more dimensions (what if some of the shows had shorter seasons? are some of them growing in popularity? could we defray production costs by bringing in partners or sponsors? and so on), until you're so paralyzed by the analyses that no action ever seems like the optimal one. My wife and I have a running joke that comes from the Netflix show "Master of None." Aziz Ansari plays the lead character, and one night he is trying to use his phone to find the best taco truck in the city. He's having a hard time getting the info he wants, and and one point whines in frustration, "What am I supposed to do, eat at the SECOND-BEST taco truck?" The joke, of course, is that whether any given taco truck is the second-best, third-best, or tenth-best is irrelevant. It's (a) an arbitrary ranking, and (b) probably not worth worrying over. You just don't want to find the WORST taco truck. Your analysis should serve the purpose of understanding the challenges that we're facing, and finding a reasonable, positive action that will mitigate or overcome those challenges. Often, if you can find the 90th percentile answer in 30 minutes, you're better off being satisfied with that and moving on, rather than taking 3 weeks to find the 99th percentile answer. The second-best taco truck is still going to give you a pretty great meal.

[@yzhu2230]

@@MikeCisnerosSWD THANK YOU SO MUCH for your time to answer my questions. it is much appreciated. It is very helpful. thanks again :)

[@lastplce]

​@@MikeCisnerosSWDI can reread this over and over. It's a really brilliant answer. 🎉

[@OsamaAlam-t4t]

can you please share the dashboard link presented in the video ?

[@storytellingwithdata]

No problem, here you go: bit.ly/exploreandexplain

[@Jerry-o1j]

11:33 How do you determine the slope of that line?

[@pv0315]

fantabulous

[@graceamah996]

Please which softwares are used for graphing and presentation?

[@storytellingwithdata]

The graphs used in the final presentation (explanatory visuals) were created using Powerpoint. The dashboard and other exploratory visuals were built in Tableau.

[@spongebobby188]

I didn't win a book 😭

[@storytellingwithdata]

Sorry to hear that! You can still access plenty of free resources: as mentioned in the session, check out our other videos here on KZbin, and practice and exchange feedback in our online SWD community: community.storytellingwithdata.com/

[@emotionalmindedstate]

😂😂😂😂😂😂😂