Edition #29 – March 2, 2020
Originally sent via Mailchimp
Good morning product friends 👋
We’ve talked about the democratization of data analysis before, with such examples as dbt’s IDE to make ELT accessible to analysts, the Meltano project from GitLab to create a single workflow for your data, data science notebooks, BI tools self-serve features, etc.
This idea is closely related to the Data > Insight > Action > Data loop. If an analyst can control all of those steps, then it becomes easier to go through that loop, accelerating learning and consequently product development.
There is something about that Superhuman+Coda.io story below that fits into that trend. It’s not very different than data science notebooks, but as it solves a common problem amongst product owners, it can be commoditized and distributed.
There is a lot of interesting projects that makes data engineering more accessible and allows all product owners to easily have access to a data warehouse with a clean and reliable single source of truth. It’s not a big leap to think that we might start seeing pre-formated analysis for common problems that will automatically plug into your data warehouse and deliver insights quickly.
I’d love to think that my very modest funnel contribution, which I’m presenting below, might be in alignment with that trend. And if you have other examples, I’d love to hear about them.
And with that, on with the 29th edition of the Product Analytics newsletter!
What has been my highlight?
So, um, my week’s highlight is actually something I wrote 😬But what can I say, I had fun working on this project 🤷♂️
It’s about building funnels with multiple paths, but structuring those within your data warehouse directly. The idea is for analysis to be possible through multiple interfaces (R, Tableau, whatever), all using the same configurations and results.
This project uses dbt (of course), but I’ve tried using concepts that are shared by tools that implement funnels. So hopefully those ideas could also be used within whatever other ETL/ELT tool you might be using.
Needless to say that I would very much appreciate any feedback you might have. I’d like to iterate on this idea to improve the overall concept and implementation. You can just drop me a line at email@example.com.
Growing your product with the help of data.
Back a few editions ago, I had shared the Superhuman Product-Market Fit Engine article, which became essential reading on the subject (the article, not my coverage of it…). Well, Coda.io collaborated with Rahul Vora (founder and CEO of Superhuman) to build an interactive document to guide founders on their own product-market fit journey.
In their own words:
This document contains a powerful engine to define, measure, and systematically increase product/market fit. It can even generate your roadmap for you!
The “engine” is essentially a framework that iterates around asking users how they would feel if your product wasn’t around anymore.
The idea behind that Coda.io doc is interesting as it makes the concepts behind the framework available to all. All you need is a dataset that is structured similarly and you can use that document to analyse your product-market fit. The analytics are a bit simplistic and you won’t be able to dig into those results, but it’s a novel approach to product analytics.
Factory operations to transform data into analytics.
I started experimenting with Great Expectations this sprint and one article I stumbled upon early on was this one on pipeline debt. As we care greatly about data quality and all DataOps tactics that can increase quality, it’s also important to understand how data problems creep in.
How does pipeline debt accumulates?
- Data systems naturally evolve to become more interconnected […] In a DAG context, “silo-free” is isomorphic to “tangled and messy.” Your pipelines wants to be a hairball, and your PMs, execs, and customers agree.
- Data pipelines cross team borders […] Most organizations are set up so that new models, dashboards, etc. require at least one analyst-to-engineer handoff before they make it to production. The difference in skill set and tools creates tons of surface area for pipeline debt to creep in.
Why should we care about pipeline debt?
This kind of bugginess erodes trust, sometimes to the point of putting the core usefulness of the data system in doubt. What good is a dashboard/report/prediction/etc if you don’t trust what it says?
Deriving insights from your product’s data.
This is probably not for everyone, but it’s still my analysis highlight of this edition 🙂 I love R and even though I don’t get to use it as often as I’d like, it’s still a treat whenever a project comes up which requires me to start RStudio and start wrangling, analysing and visualising data.
So I’m always a bit jealous when I see R conference videos with presenters that get to work extensively with R. Jealousy aside, this suite of videos is a treat.
There is one in particular which should be of interest to this newsletter’s readers. It’s called “Growth Hacking with R – Product Analytics at Scale using R and RStudio“. This is a bit salesforce-centric, but if you’re interested in calculating and visualizing retention in R, and doing a bunch of analysis on top of that data, it will be worth your time.
What’s happening the product analytics market.
After the “CRM is not enough” declaration, Segment now publishes a report on the state of the Customer Data Platform market. Not sure how intentional was the sequence, but it is driving the point home that customer analytics cannot rely solely on CRMs anymore. The introduction kinda gave me a hint 😉
Remember those halcyon days when companies had one piece of software – perhaps a CRM – to manage their customer data? Well, those days are well and truly behind us.
With the number of tools growing every year (the average SaaS company uses 80+), businesses are having to invest more and more resources to keep their tools and customer data in sync.
CDPs are more of a hub that captures data from multiple sources, unifies identity, processes/augment data and pipes the data to multiple destinations. Whereas a CRM is just one of many destinations for that unified/augmented data. The case against CRMs is really about continuing to deal with fragmented customer data. It leads to bad data and poor customer experiences.
Besides saying that this industry is growing rapidly (by analyzing vendors, data points processed, venture money invested, etc.), there is a bit that talks about the hype surrounding CDPs. As stated:
Gartner recently placed CDPs in the peak of “inflated expectations.” It means that CDP buyers are expecting this technology to do more than it’s actually capable of.
That stems from people expecting a bit too much of CDPs. That makes sense as it is a bit ill-defined and it sometimes feel like anything that has to do with customer analytics falls within that bucket.
For example, I receive the daily newsletter from the CDP Institute and the variety of vendors and use cases covered is quite vast. It’s sometime hard to identify what are the common traits of CDPs. But this is a recognized problem and this report is actually an effort to clarify that.
If only for this, this report is worth the read. But Segment also provides a glimpse into the data points it processes, the sources and destinations that are used by their users, etc. If you work with Segment, those insights are actually quite interesting in themselves.