Beyond Measure

A database in essence does three things: write, store and read data.

DuckDB is small.

Open DevTools on LinkedIn and you’ll see hundreds of failed fetches pile up within seconds.

I wasted a morning and gained an opinion: Conductor is still the best tool out there if you want to orchestrate multiple coding agents.

The hubris is strong in this one.

People using this cell1 1101001k2k 4K UHDBuffering… Your throughput4K UHD48 Mbps the cell's capacity≈ 48 Mbps total Signal −95 dBm4 / 4 · unchanged QualityRSRQ −4.0 dByour slice: 100% Drag the slider to fill the cell.

For request analytics at the edge (using Cloudflare Workers here, but it’s similar for Vercel or AWS Lambda), the cheapest architecture is the one that batches writes at the right level.

If you have a website these days you might be missing out on a lot of valuable data.

So, you want real time and interactive, but handle large analytical data at the same time?

It feels like there’s a new tool, framework, or SaaS product every hour.

TLDR: Use microbatching for large, time-based tables where robustness and backfilling is important.

We talk a lot about whether AI is smart enough to take our jobs.

I’ll admit it, as a data engineer I still use Excel.

If you work with API data long enough you eventually inherit a directory or bucket full of tiny JSON files—one per product, event batch, log minute, user snapshot—because “it was easy at the time.”

Nietzsche reading Hacker News.

Deploying Streamlit apps with dbt on Snowflake using a custom dbt materialization: a practical tutorial and exploration If you want to jump straight into the code, you can find it on Github.

If you’re using Snowflake, you know that things can get expensive very quickly if you don’t watch out.

If you’ve spent more than three days as an analytics engineer, you’ve probably had that moment: your dbt model runs, but the numbers look… off.

If you come from an analytics background you might still know what it was like to have folders full of notes and SQL files to keep track of your transformation queries for dashboards and analyses for business questions.

If VS Code Copilot isn’t detecting your shell properly, it’s likely your right-hand prompt (RPROMPT) interfering.

Snowflake requires separate access permissions for each object type in a database.

Snowflake’s information schema uses ‘YES’/‘NO’ strings (like is_iceberg).

You can configure secrets directly in your dbt profile for MotherDuck.

To publish a remote MCP server to the registry, create a server.json with the schema: { "$schema": "https://static.modelcontextprotocol.io/schemas/2025-12-11/server.schema.json", "name": "com.example/my-server", "remotes": [{ "type": "streamable-htt…

In dbt, the + sign has different meanings depending on context.

Have you ever wasted time or money because you made a decision based on incorrect data?

This post has been expanded to appear in my book The Fundamentals of Analytics Engineering, a perfect handbook if you are looking to learn more about data modeling, data quality, data warehousing and data ingestion.

With Google Analytics’ sub-par performance over the last few years, you may have been on the lookout for alternatives or just overwhelmed with the amount of LinkedIn posts claiming to have the ultimate list of alternatives.

Have you ever seen certain small towns in your analytics reports with thousands of visitors?

Bots are a big part of the internet.

Imagine this: your website team has developed a great new feature.

With the recent update to Google Search Console (GSC) allowing us to export search console data to BigQuery it becomes very easy to do some fun stuff.

How hard can it be to measure how many people visited your site?

Incrementality is hard.

👋 Hey!

How do you expose data in BigQuery with an API?

Update June 2023: I have created a massive blog post on how to get started with Snowplow, Terraform, Docker and dbt that has some more insights on how to best run dbt in a container as well.

Over the last few years SQL has really started embracing its second adolescence —it’s nearly half a century old after all.

If you enjoy this post.

Server-side tracking is all the rage these days, but let me tell you about the uber-coolest kid on the blockchain: edge analytics.

If you’ve been around this blog before you might know that I know a thing or two about Google Tag Manager (GTM), the convenient tool that lets you run marketing and analytics scripts on your website.

If you’re just looking for the code and a quick start, you can find everything on Github and you can find a live demo at docs.dumky.net.

One thing that has bothered me over the years is the weirdness of not having any possibility for setting up a pipeline in Google Tag Manager for when new container versions are published.

“Google Page Speed says GTM is slowing down my site, can we remove it?”

The internet is a beautiful place.

Google Sheets, the online Excel alternative, is basically the Swiss army knife of any modern knowledge worker.

With terms like GDPR, Cookies, ePrivacy Directive, ePrivacy Regulation, ITP, ETP, most people’s heads will start spinning.

When we think of analytics we think of tools like Google Analytics, Mixpanel or even Tableau or PowerBI.

Just a few days ago Google released a new version of their tag management system: Google Tag Manager Server-Side.

Do you know if all your tags are used in Google Tag Manager or if you can remove some?