Libra + Newsletter Analytics + Experimentation At AirBnb + Stitch’s Mailchimp Integration (PAN #12)

This newsletter’s goal is to cover digital products and how analytics helps them grow. Now here’s a product that will be interesting to watch grow in the following weeks and months: the cryptocurrency Libra by Facebook and a consortium of founding member giants. Their goal? They “want to make Libra the evolution of PayPal”.

Why is it interesting? The promise of stable value in cryptocurrencies, the launch of a privately-owned currency, a whole new sort of product with what it entails in terms of adoption, analytics, etc. 🤯

Like me, are you also running a newsletter? This article is a gold mine. It goes through many important dynamics observed in newsletter and the outcomes that could/should be measured.

Read about metrics such as the CPEEA (Cost per engaged email acquired) – “To understand the true efficiency of your marketing spend, you should understand whether or not you’re investing in signing up engaged readers.”. Also read of laws such as Andrew Chen’s Law of Shitty Clickthroughs. Packed with insights.

They also have a section on their data stack at the end, which feels rather incomplete to me, but still interesting 🤓

“Be smart: How Axios drives engagement with its email newsletters through user-level data“, by NiemanLab

If you’re into Lean, then you’re most probably managing your product through a cycle of Measure, Learn and Shape. Analytics is foundational to those 3 parts: Measure is where you transform raw data into important metrics about your product’s dynamics, Learn is how you consume that information and Shape is how you act on it.

But how do you act on it? 🧐

Hopefully, experiments is a big part of how you approach that phase. You wouldn’t want to come up with big product changes, based on learnings that you haven’t validated through experimentation. But how do you go about this?

For ourselves, our approach is very simplistic where we just document our experiments on Asana and analyze results in R. But, experimentation can be much more sophisticated. And this article on experimentation at AirBnb gives you a glimpse of how the big players are doing it.

Experiment Reporting Framework, by AirBnb Engineering & Data Science Team

Alright, so this is a follow up on last week’s story on how to automatically transfer Mailchimp data from source to your own data lake. With the help of the good folks at Fishtown Analytics, creators of the most excellent dbt ETL (and much more) tool, I got into Stitch’s Mailchimp integration Beta program.

Win! 🥳

If you’re familiar with Stitch, this is a thing of beauty! You get to replicate 8 source tables (automations, campaigns, list_members, list_segment_members, list_segments, lists, reports_email_activity, unsubscribes). I’ve started adding some of those to my ETL, which feeds my data warehouse and it’s all the data I was hoping for!

Of course, there are bugs along the way (as you can see in the screen capture above), but it’s a Beta program after all. And the guys at Stitch have been really responsive so far.

In next week’s edition, I’ll show you how my ETL is looking so far. It’s starting to become very interesting with the addition of Mailchimp. 🤩