📚 Cedric Chin on LinkedIn: Open Source XmR Charts (2024)

This is exactly what needs to be done.Don’t want to educate for data literacy? Fine. Then build process into your tools, such that the only way to get business value is to learn the embedded process, or churn out.If you’re a data tool vendor, this is likely the high order bit.Accounting software does not have this issue because they have GAAP. What’s the GAAP for operational data?

15

1 Comment

This was an eye opening observation for me.The context here was that I was telling Crystal what I’ve learnt from using process behaviour charts to grow an info product business.“Process behaviour charts are used to reduce variation in manufacturing, but they are more useful outside of manufacturing as a method to identify when variation has changed.” Which I think is still broadly correct, and why they’ve not spread so broadly.Ahh, but then I said “typically in manufacturing / continuous improvement, the advice is to reduce the variation before improving the process but I don’t think that applies as much outside” at which point Crystal leapt in and said “no, that’s true in tech also!”This post lays it out wonderfully:

16

3 Comments

So, I, uh, spent maybe 30 minutes ranting about data topics at the end of my Analytics Power Hour episode with Tim Wilson, Michael Helbling and Julie (Shallman) Hoyer, and Tim turned one of my analogies into a full blown parable!(To be fair, I think he already had a similar analogy of his own ;)The truth is that there’s a rotten stink at the heart of the data industry.All everyone seems to want to talk about are the tools, or perhaps the techniques ABOUT the tools — there are a gajillion articles about ELT and how to build a data warehouse and how to become an analytics engineer and bla bla modern data stack is alive or dead or alive again but nearly nothing if you are a business operator, like me, and you want to learn how to become more data driven.Don’t get me wrong: I clearly enjoyed reading (and in some cases writing!) such navel-gazing articles in the past.But I quickly tired of them. They are counter-productive.The reason? It was becoming clear to me that churn in the data space is so high because the vast majority of companies who buy such tools are NOT data driven.And so they can’t get business value out of their tools. What happens? Simple: you churn. You switch from one tool to another. Most companies get sold this nice shiny narrative about better tools, about how buying this new stack will lead better data driven outcomes, and of course no such thing happens.In the parable below, Tim extends my throwaway analogy about kids being sold nicer pencils, when what they really need is to be TAUGHT TO WRITE.But life is fair — for as long as nobody teaches customers to become data driven, BI tool companies will enjoy, at best, 2% monthly churn, and watch 22% of their recurring revenue walk out the door each year. Good luck trying to reach venture scale on such fundamental economics!

35

11 Comments

I was on the Analytics Power Hour podcast recently to talk about becoming data driven, from first principles!(Actually, about using dead simple, 90-year old techniques to become data driven, heh ;-) But this was quite intimidating because Michael, Tim, and Julie were Actual Analytics Professionals(tm) — and had more years in the industry combined than I have years on earth. I had listened to a number of their episodes before our call, and so kinda over-prepared for the episode.But it was SUPER fun, and if you listen till the very end, you’ll notice a credits segment where I rant about how hard it was to learn about ACTUALLY using data in a business context from the data vendors.

7

I used to think that Continuous Improvement is valuable because it's about continuous ... improvement. I no longer think this.I now think that the real secret with continuous improvement is that it gives you a rigorous methodology for N=1 studies. (Alternatively: 'single subject studies').The what now?Well, if you're running a business, you can't really do randomised controlled trials for everything. RCTs take too long, and are too expensive — plus it's not clear that what you learn from them, across many companies, applies to YOUR company.What you want is a way to verify that something works FOR YOU. That something could be anything — a new sales process, a new marketing campaign, some new software engineering practice, etc.The body of work that is Continuous Improvement gives us one rigorous way to do this.There are really just 3 load bearing ideas:1. The PDSA loop2. The Three Questions of Continuous Improvement ... and3. The Process Behaviour Chart.To find out more, read this week's Commoncog essay below:https://lnkd.in/g9cnZmqS

34

6 Comments

I'll be giving a talk at the NUS Hackers tomorrow, titled 'How to ACTUALLY become Data Driven'.The goal is to talk about the stuff I've been testing for the past 1.5 years — how to become data driven, yes, but from first principles, and building all the way up from scratch to a practice like the Amazon-style Weekly Business Review.My co-speaker is Shawn Tan, who will be talking about the limitations of Transformer Models!Come down to NUS if you'd like to ask questions, and hear me go on (and on, and on) about this data thing that I can't shut up about!https://lnkd.in/gUZSXbJF

26

2 Comments

I'm pleased to announce a tool that we've been working on for around 9 months now.If you've been reading my stuff, you've probably heard of XmR charts.XmR charts tell you what to pay attention to in your data.They are dead simple. Today we launch something to make them EVEN simpler.I'm pleased to introduce Xmrit: https://lnkd.in/ggkx8BbR Xmrit is a free, browser-based tool to help you create, modify, experiment with and share XmR charts. It's actually a little nuts that nobody has built this before.We've been using it operationally for about 5 months now. And we're confident that it's ready for broad use.We want XmR charts to take off. Which means making it dead easy to create XmR charts for your company's data.With Xmrit, this is trivial: treat the table as Excel, paste in some data, and you're good to go.Xmrit comes with features that you might not think is necessary until you've been using XmR charts operationally for a bit.For instance:- You can insert dividers to characterise process change.- We give you a way to lock limit lines. - You can programmatically generate new Xmrit share links.We've also created a free online course to get you started!Go on, give it a go! I hope you'll start sharing your Xmrit charts publicly!As I joke, crudely, in the announcement post: “the goal is not to talk; the goal is to use these methods and get rich!” (Where ‘rich’ is a proxy for achieving your goals, becoming successful, etc).Let me know what you think!

101

14 Comments