Product Analytics Newsletter
Edition #7

  • In the News, it’s all about privacy, privacy and privacy – take me there
  • In Strategy, we look at how North Stars can go south – take me there
  • In Tools and Techniques, we move data around – take me there
  • In Privacy, we take steps to empower our privacy – take me there
  • In My Journey, we measure our first experiments – take me there

Do you consider yourself to be a product owner?
*|SURVEY: Yes|* *|SURVEY: No|*

Privacy, Privacy and Privacy
I assume we’ve all heard of that story where Amazon employees were listening on “some” of Alexa conversations. I get the tech reasons why Amazon would do it, but don’t they see how privacy is so freaking sensitive? Wired editor-in-chief Nick Thompson’s take on that story raises an interesting question: what should be done whenever Alexa picks up signs of violence in the room? Stay silent? Report it?There were also a few twists with GDPR into the extent of how that regulation is applied. Fist off, this has been stated in a Forbes article: “GDPR is often wrongly portrayed as a kind of privacy shield protecting EU citizens. In reality, […] businesses without an EU connection are free to harvest, mine, manipulate and monetize EU citizens to their heart’s content.”

We’ve also learned that tech groups are now told to stop nudging children into using apps. Btw, those may seem faraway concerns (for those outside of Europe), but it’s just a matter of time before all countries adopt similar regulations. Mights as well take notice now.

What else? Facebook of course… sigh… Our friend Zuckerberg’s “past comments on FB privacy are being examined and the FTC could try to send a message to corporate leaders that they could be held liable for their company’s mishaps.” Or how about this gem? While Facebook was claiming they were protecting your privacy… “We gave a bunch of stuff ‘for free’ historically and now we’re making you ‘pay’ for it via reciprocal value.”

Here’s a story of how the use of AI can serve authoritarian tendencies. China has started experimenting with AI to profile minorities. “The practice makes China a pioneer in applying next-generation technology to watch its people, potentially ushering in a new era of automated racism.”

On a somewhat related note, I recently watched People’s Republic of Desire, a “dystopian documentary that explores the people behind China’s craze for online livestreaming.” It’s a fascinating documentary, with a Black Mirror vibe to it. It essentially portrays the lives of Chinese people who are part of the social network YY and how their lives and social interactions are manipulated to serve this platform’s business interests.

North Stars Going South
In a previous edition, we had shared a video from Jon Noronha, Director of Product Management at Optimizely (at the time), called Experiment-Driven Products: Learning From 1M+ Experiments.One thing we had noted was…

“A struggle that caught my interest is the following: choosing the right North Start metric (your most important metric) is crucial as your experiments will make you progress towards improving that metric… but sometimes with unfortunate results.”

Well, we had a really interesting case in the last 2 weeks where the wrong North Star lead to serious flaws in a product. This story is about how YouTube’s algorithm ruined Katie Bouman’s black hole triumph by “bringing some of its most insidious content onto people’s front pages.”

Choosing the right North Star is crucial or it might lead you into a gigantic black hole (pow!). On that note, here’s a great piece from a16z that lists 16 startup metrics to lead your growth.

Moving Data Around
Modern analytical infrastructures requires moving data around. And lots of it. Historically, data engineers would build data pipelines that pulled data from source, transform it and push it to your data warehouse. That’s essentially what an ETL (extract, transform and load) process does.I did write an extensive guide on ELT (notice that load precedes transform) in general, and dbt (the data build tool) specifically.

What’s important in the context of moving data around, is that we now want to copy source data untransformed to our data warehouse. Data warehouse technologies such as Snowflake, AWS Redshift and Google BigQuery provide cheap storage and processing power, that there is no point in transforming your data within a custom pipeline. Might as well push your raw data to your data warehouse and transform it there with sql. Simple and freaking efficient!

So how do you move your raw data to your data warehouse? Of course, you could still maintain custom data pipelines to do this, but why? It’s error-prone and plainly just not fun to maintain.

There are some really cool and cheap SaaS tools out there to do it, such as Fivetran and StitchData. But even data warehouse providers have their own “solution”. AWS has their version of data pipelines and Snowflake has Snowpipes.

It’s so easy and cheap nowadays to centralise all your data within a data warehouse and transform it to your liking that there’s just no good reason to have custom python scripts that glue all your data and BI together.

As a newly-converted Snowflake user, I did stumble on that article not long ago and it just shows how easy moving your data around has become such a simple task – “Setting up a data pipeline using Snowflake’s Snowpipes in ’10 Easy Steps’”.

Empowering Our Own Privacy
I realize that I’m allocating a lot of space to discussing privacy issues in this newsletter, but it is a central concern of users that has to be taken seriously by product owners.As we’ve seen in the news section above, there are many stories that pops up each week about products we hold dear that take privacy not as seriously as should be. Regulations such as GDPR are there to force companies to properly manage our data, but we’re still in the far west when it comes to dealing with user privacy.

With heightened realization of the amounts of personal data being cumulated, shared and used by corporations, and seeing how regulations have limited impact for now, it’s no surprise that users wants to start actively have a say on how their data is being used.

In that context, the privacy assistant Jumbo has the ambition to allow users to easily control what data products have on them, what data can be captured and how they can use that data. It’s still limited to a few products for now (Twitter, Facebook, Google Search and Alexa) and only a few use cases, but I think this idea is spot on and comes at the perfect time.

How about going one step further, and have companies work pro-actively with Jumbo for users to manage data through Jumbo’s interface?

Measuring Experiments
In the last edition I mentioned that my North Star for the newsletter at this point is the number of links clicked per edition. In my opinion, this should indicate engagement and a clue that I’m on the right path to fulfilling this newsletter’s mission (to share data analytics’ best practices and new ideas, so product owners can improve their feedback loop and accelerate the development of their product).My approach towards influencing that metric is two-fold: first validate that my assumptions are correct in regards to this newsletter’s mission; shape the product to improve engagement.

The first type of experiments cares less about impacting a metric directly, but have more to do with validating/invalidating assumptions. For example, I have a poll running at the top of this newsletter to test my assumption that readers of this newsletter are product owners.

The second type of experiments have to do with making changes to the product and see how it will influence a targeted metric. For example, I ran a couple of experiments so far to see if that would incite users to click on links I share.

In the 6th edition, I had introduced the table of content and shared a bit more video links as I thought this would increase the number of links clicked. It doesn’t seem to have had a noticeable impact on the number of links clicked. Work in progress.

There’s A LOT to be improved here. The major caveat is that I do not have enough readers yet to do A/B testing. I also want to automate my product analytics process (I’ll share more on that in the next edition) and continue building more experiments.

Was this newsletter forwarded to you?

Subscribe Here

How can you leverage data to strategically grow your digital product? This newsletter’s mission is to share data analytics’ best practices and new ideas with product owners, so they can improve their feedback loop and accelerate the development of their product.

Copyright © *|CURRENT_YEAR|* *|LIST:COMPANY|*, All rights reserved.
*|IFNOT:ARCHIVE_PAGE|* *|LIST:DESCRIPTION|*Our mailing address is:

Want to change how you receive these emails?
You can update your preferences or unsubscribe from this list.