timo dechau 🕹🛠

We will be launching a new website version for datatasks this week. See the preview below. The initial design I cluttered together by using Tailwind UI. It was ok, but not really great. Now someone with design skills has worked on it - and it's far better now. #buildinpublic

@cronsholt @cronsholt - this would be a cool use case. If you like to chat about it - dm me

@julianajaxx I was allowed to peek and it is really good.

@JohnKutay I think this is the best analogy for CDC I read so far.

An analogy for raw log-based change data capture: Imagine you want to survey a household on what they've been eating for dinner every night. Knock on the door and ask? No they're too busy. Instead, analyze their garbage and other waste I won't speak of.

@ergestx +1 for JTBD - I can recommend you some books @ergestx

One huge realization I’ve had recently: Whether you’re doing marketing, product development, analytics, etc. asking users what their problems are, what they’re struggling with doesn’t always lead to the core problem. Often you’ll just get symptoms but you won’t know it.

@sspaeti Awesome. Will check it out!

We've open-sourced an "Open Enterprise Data Platform", integrating the Modern Data Stack into a single portal. It features state-of-the-art tools like dbt for SQL data modeling, Airflow for task orchestration, and Superset for BI dashboards, all on a Postgres database.

@julianajaxx Let me know when it is ready

@julianajaxx The gclid lady. Improving your gclid performance by +20%

If vendors can't stop talking about cookieless. Ignore them. There is no cookieless. And yes, fingerprints are not technically cookies but are legally the same stuff.

@laurenbalik @acemoney21 Another famous German word made it into English

@acemoney21 Prop traders need to know Marxist theory, Ace. Marx is the best writer on capitalism, hands down... ...even if his solutions didn't work out well in practice.

Happy New Year. I would like to apologize to all of X. In late 2022 (~1 yr ago) I was offered the opportunity to interview for a Twitter (now X) data executive role by Andreessen Horowitz VCs who sought to bribe me off so that I'd stop writing blog posts about dumb a16z data infra and SaaS investments. (a16z participated in the Twitter LBO and has/had sway over Twitter, had people there after Twitter fired many employees.) I didn't take the X role. I was too spooked by the sketchy a16z behavior -- if me writing a few viral blog posts about how companies shouldn't use Fivetran + dbt to solve all their data problems pissed off an a16z GP enough that he'd be willing to offer me something to shut me up, what did that say about X/Twitter at the time? I know Elon is smarter than a16z mid-tier GPs and all the other VC toadies. I really do believe that. 1) Every time there's downtime or an outage on this platform I cringe, because I know I could have limited this. 1a) Every time there's another story about advertising on X (somewhat overblown, IMO) I know I could have helped do monetization better. 2) X is more alive than ever. I love it as a user. A lot of the nonsense has left the platform. The lumpenproletariat class of low-level, low-intellect people trying to create "gotchas" for minor rent extraction has mostly disbanded. The pseudo-patrician class of technocrats who can't write a for-loop is declining. This is all good. I've learned SO MUCH on this platform in the last year. 3) I am looking forward to X in 2024. I am very bullish on this platform. 2024 is an election year in the US and there will be nonsense everywhere. However, X is still the best way to learn new things and explore ideas, hands down. X is how the wheat and chaff separate. Always happy to exchange ideas @elonmusk. I think you're doing an above average job, certainly better than the prior administration, but there's always room for improvement and happy to help (no VCs allowed though 😜).

@VicVijayakumar That’s how we do it. Zipbag with the bricks and instructions.

My wife tells the kids to : - put their broken down sets into ziplock bags (I support this) - and put those bags in a Tupperware container under the bed (okay) and throw away the box (uhhhhhhhh) However I’m unsure what I would do with the boxes.

Wife and I are having a heated disagreement. What do you with the LEGO box after you or your kids build a set?

@fastbloke @julianajaxx @PiwikPro Oh - that is weird

@julianajaxx @PiwikPro @timdechau <sad face>

Charcoal Gray: #36454F - Offers a professional and sophisticated feel. Bold Purple: #673AB7 - Implies wisdom and forward-thinking. Sunset Orange: #FF5722 - Radiates enthusiasm and energy. Now - I can work with it from here.

And the response: Classic Blue: #007AFF - Represents trustworthiness and reliability. Vibrant Teal: #32BAAE - Suggests creativity and analytical thinking. Deep Sky Blue: #00BFFF - Conveys a sense of innovation and clarity. Crisp Green: #4CAF50 - Symbolizes growth and renewal.

Crazy prompts, you would never thought about before: "I am building a membership program for people to learn product analytics - and I need a primary brand color - can you suggest ten different well-designed brand colors as hex code."

@Ldnbox These lists are always misleading. They are alternatives for you but no alternatives for a lot of other setups.

GA 4 is just a terrible product. 😭 Here are 10 Google Analytics alternatives to consider. fathom analytics Plausible Analytics umami Simple Analytics Overtracking Seal Metrics Koko Analytics Matomo Analytics Wide Angle Analytics Pirsch Analytics My favorite is umami.

@julianajaxx Looking good.

@ben_brandwood @ergestx In general I do agree with the two. But these are output metrics and the data team often don’t have direct influence on them. So you need to break it down.

Focusing on insights is another data dark hole that is impossible to climb out of. Just like questions, insights are quite insidious… Here are some of the problems with insights: - Insights are subjective. What’s insightful for me could be common sense for you and vice versa - Insights don’t automatically lead to decisions. What I consider actionable might not be actionable for you and vice versa - The value of insights (from data) is unknown and maybe even unknowable until some action is taken - Many so called insights end up confirming existing biases or reinforcing existing assumptions and prodesss due to how data is produced - Interpretation of insights is relative. We could be even looking at the same exact results and still interpret them differently based on our knowledge, experience and beliefs - Just because you think an insight is actionable doesn’t mean action will be taken. This alone makes the work of data analysis and scientists incredibly frustrating - Measuring the performance of the data team by the number of insights produced then becomes nearly impossible So what’s the alternative? Here are my suggestions: 1. Redefine success in analytics as producing curated and validated data sets that represent the business as accurately as possible. 2. Adopt standard ways of measuring business performance and get buy in from executives. Review them periodically to suggest new ones. 3. Instead of providing insights, take on research questions that directly impact the bottom line. The idea is to produce implementatable changes not just insights.

@HoloMarkeD That is an interesting idea.

Will the embeddings for a book and all its references be a new medium? Plug it into your favourite chatbot, ask questions and explore over its content.

@beeonaposy Highly recommend Olga’s Substack: dataanalysis.substack.com - it is a great library how to approach product analytics tasks like funnel or retention analysis with SQL.

Has anyone found good open source libraries (Python or R) or playbooks for product data science? Particularly interested in case studies of scaling analyses (which aren’t paid tools!)

@ergestx It definitely gets better over time. What helped me was to accept that I can waste time and resources and can throw away the result. This makes computer work so great. Just delete it.

I have this thing that if I don’t know how to do something, I’d rather not do it than look incompetent. Fixed mindset? Maybe, but that’s just an explanation. What I need are algorithms to overcome this limitation.