Data Stacks and are we forgetting the main purpose of them? [4 mins]

[nelsonic]

A friend recently wrote this post and has RFC:
https://medium.com/@IDWT/data-stacks-and-are-we-forgetting-the-main-purpose-of-them-6f499bb2e4c9

Going to read it and capture my thoughts here.
Then will share with them.

[nelsonic]

Snapshot of content at the time of reading (reading anonymously to see UX of majority of readers):

Don't worry, I have logged into my Medium account and I am the author's first follower: https://medium.com/@IDWT

Full page screenshot courtesy of Firefox:

[nelsonic]

This is my line-by-line / paragraph-by-paragraph review:

Title:

Data Stacks and are we forgetting the main purpose of them?

I would either remove the word "and" e.g:

Data Stacks Are We Forgetting The Main Purpose Of Them?

or reformulate the title as:

Data Stacks Are We Forgetting Their Main Purpose?

or go with a more HN/Reddit (yes, potentially "clickbaity") title:

How A Data Stack Can Kill Your Startup

My reasoning for the sensationalist title is simple: it's true.
If you select a complex stack that then results in difficulty extracting/analysing data,
the team building the product will be less likely to do more interesting/advanced analysis
i.e. their product development will not be as data-driven meaning

• they won't achieve product-market-fit,
• find + fix the UX issues that are harming/preventing adoption
• ultimately run out of runway (cash) to keep building the product

and ... DIE!

The second biggest reason startups fail is the run out of cash
having a data stack that only a handful of people understand (the person who created it or other expensive experts) will rapidly drain your company's budget. You will get far more milage from a simpler solution that all engineers can easily query and everyone can use for anonymised Actionable Timely Metrics^TM. 😉
It's fine to have an experienced/competent person setup your stack, but if they are creating something with 20+ moving parts instead of just using a standard SQL (e.g. PostgreSQL / MySQL) service, they are setting you up for a very complex and expensive journey.

Note: on title capitalisation there's a whole article on this: https://en.wikipedia.org/wiki/Title_case#AP_Stylebook
I prefer to capitalise every word for consistency. You can do whatever you prefer. 👍

Subtitle

Why do you need a Data Stack or solution?

I don't know if this line adds much in terms of enticing the reader to want to give 4 minutes of their life to reading the article. The first 7 seconds must to answer the question:

"What's in it for me?"

Why should a busy person put down everything they are doing and read this article now?

[nelsonic]

Intro paragraph:

Having worked in a range of companies I want to share my findings, 
learnings and opinions about data stacks,  with the hope that organisations can 
frame their discussions on  focusing on why they are needed?

Read: https://insidegovuk.blog.gov.uk/2014/08/04/sentence-length-why-25-words-is-our-limit/

Consider rewording for brevity, e.g:

Having worked in a range of companies from 20-person startups to large enterprises,  
I want to share my findings about data stacks.
My hope is that organisations can frame their discussions on focusing on why they are needed.

Splitting the sentence does not lose any meaning but makes it easier and faster to read.

[nelsonic]

This sentence introduces the reader to the author as having "worked in a range of companies".
This is too vague and will not be enough for the reader to know the types of companies
and specifically if the experience of the author is relevant to the reader.

i.e. is this article relevant to me [the reader]?
does the author of this post have priceless wisdom about the topic that they can share with me if I read it?

Given the name "In Data We Trust", the first thing that should hook me in as a reader is some data.
Show me a graph or a 2x2 Matrix or a Table (last resort)! 📈
If no data is being presented to support the claims/content of the article, then we aren't trusting in any data ...
Also, having an image helps contextualise the content and makes it more "shareable".

[nelsonic]

Here is a sample post from AirBnB's Data Science / Engineering blog:
https://medium.com/airbnb-engineering/how-we-partitioned-airbnb-s-main-database-in-two-weeks-55f7e006ff21

The title explains exactly what the reader can expect when reading the article/post.
The first thing you read is a "tweetable" intro paragraph:

“Scaling = replacing all components of a car while driving it at 100mph” 
– Mike Krieger, Instagram Co-founder @ Airbnb OpenAir 2015

Followed by an contextual image:

The caption is:

"Airbnb peak traffic grows at a rate of 3.5x per year, with a seasonal summer peak."

Off the bat I have some insight into the seasonality of AirBnB's business and I want to keep reading.
The graph doesn't have a Y-axis label so it's definitely not scientific, but it is still insightful.

This is just a basic example. I could dig out others from my bookmarks.
But the idea is simple: make the intro paragraph tweetable, so under 140 characters.
Try to include a graph even if it's just a stylised one that conveys your message in less than 5 seconds
and adds a visual break between the title/intro and the main content.

[nelsonic]

The first sub-heading is a good one.

How can I help you on your data journey?

This feels like an opportunity to link to an "about me" page rather than in-line the bio.
Totally agree with building rapport and trust. 👍

I started my career as a Management Consultant 

As an engineer with Consulting experience (that is buried deep in the history of my LinkedIn but not on my CV!),
I am put off by the word "Management Consultant".
Not that "Management Consultants" are "bad", many of them are very smart and skilled people.
But the reputation in the tech community is:
people who superficially understand the industry
and have limited experience because they are fresh out of university
being parachuted into a company at high daily cost
to identify a problem - usually not from First Principals or through Root Cause Analysis -
and make a recommendation that leads to an up-sell of more consulting work
to "remedy" the problem they have identified.

I've never heard a Management Consultant tell the client "you don't need our help".
Why? because they don't get paid if they don't find something to "fix".
Even if they end up wasting the Client's time/money with a half-baked "solution"
that sets them back more than it moves them forward ...

e.g: https://madeintandem.com/blog/massive-hertz-accenture-lawsuit-warning-everyone-consulting-heres-can-learn/

[nelsonic]

Had to put this down because 👶 is 🤒 and I had to take care of ... ❤️

[nelsonic]

I started my career as a Management Consultant 
and then moved to become a hands-on data driven individual 
where I have experience of working across the full spectrum of Data. 
From engineering pipelines, designing and modelling 
data warehouses and building organisational wide analytics.

This is good for a bio. perhaps at the end of the post.
But at the beginning it just makes me think: get to the point!
(obviously I am interested in the Author's background as a friend,
but the reader who has no idea who they are ... not so much ...
just presenting my thought process as a random person reading this post ...)

Also, I don't understand the need for the word "individual" ... 🤷‍♂️

From my experience of working as a full stack data individual, 
I have learnt one key lesson. Business Value should remain the why 
of any decision when designing or building any data solution.

Again the word "individual" ... 💭
I understand the word, just don't feel like it adds anything to the sentence.
Would replace it with "Scientist", "Engineer", "Lead" or "Professional".
Any of these words add more authority/credibility to the post than the word individual.

At it’s core an effective data solution provides information 
into the hands of people quickly to make decisions. 
The more complicated the data stack, 
the longer the time is to generate this value. 
Hence, simplicity is key for data solutions which 
I will breakdown over a series of posts.

[nelsonic]

Diving back in ...

exceedingly complexity

should either be:

exceeding complexity

or

exceedingly complex architecture

[nelsonic]

Listened to and read the rest of the article. It's good. I feel it didn't really "hook" me though.
I'm only subscribed/following because I know the person. ❤️
If I was a stranger - interested in the topic - reading it I wouldn't feel compelled to "tune in next week" ...

Providing feedback directly. ✅

[nelsonic]

feedback sent. 👌🏻

[nelsonic]

Reminds me of: https://twitter.com/ThePracticalDev/status/800769209496170496