Secoda Community

Image from iOS.png

1_g61VL-RIQ3M7MPKTfyGx-w.png

A lot of food for thought here. Makes me think: are we increasing access to data to scale data scientists or data analysts? I would argue it's more on the analytics side, allowing data scientists to do what they're best at: complex workflows, feature engineering, ML decisioning, etc. More opportunity for Secoda to help with allowing those team members to focus on the projects where they bring the most value!

IMG_3208.jpeg

Congratulations. Really excited for you guys.

This is interesting and the spirit of discourse I know our data team is very much against the NPS score as an accurate measurement.

 I can’t remember why and I don’t jump into the CS world too much (where it is used a lot) but I’d be happy to find out and share the reasons why!

So this is what I got, more related to the CS settings:

```First, it's a 1-10 scale where the only acceptable answer is on an extreme of a range. Users don't often understand how NPS works and how it's not a normal scale where the bottom end is fail, mid-low is less than ideal, mid is neutral, mid-high is decent, high is very good, and the max is beyond amazing. 

It is also often collected at really low intervals so you never get a value that is truly representative of your users. 

Because it's a single-variable score, it also has flaws because you don't understand why your users rated you a 4 (was it customer service, price, internal adoption, or was the actual experience bad?).```
The other thing to add is ordinal scales ranging from 1-10 are difficult for users to seperate their feelings. What really is the difference between a 6 and 7 and how do you quantify that? I typically like using booleans for a proxy as a satisfaction rating (think how netflix adopted a thumbs up vs down) which can be better used for recommendations.

I think the way some of the automated NPS scores survey customers are misleading, asking if you are happy with the service and then would you recommend right after ..is a better indicator of the service you received and not an indicator of your users promoting your product. If every service I received asked me these following questions after providing me good service I would be recommending everyone, but I def do not.

I like how you guys talked about analyzing afterwards. I think thats an important aspect to the piece. Many people just use a NPS score and use that as a single metric to determine customer promotional activity without actually understanding it!

Thank you! We seriously could not have done it without the early vote of confidence from all of you :pray:

Thanks <@U01PM6182AX>! If anyone does find another part of networking confusing, lmk. I’m really hoping to make it more accessible through maybe a series of posts.

Unbundling_1_a37ce99b59.jpg

From our friends at Hightouch <https://hightouch.io/blog/unbundling-the-cdp/?utm_source=linkedin&amp;utm_content=lh2203016>

This is a great write up:fire::fire::sparkles:

Here's the link for those who missed it: <https://m.youtube.com/watch?v=nrIZROb0Byc|https://m.youtube.com/watch?v=nrIZROb0Byc>

Hey, does any one else feel like data quality is a harder problem to solve than code quality? Curious about what others think. I also listed some more and less popular advice about how to reduce data quality issues:

1. Move transformations to the right with ELT pipelines
2. Move transformations to the left with a metrics layer
3. Test and monitor data at the source and before being consumed
4. Reduce the number of intermediary models
5. Reduce the frequency of ingestion and processing jobs

<https://airbyte.com/blog/data-quality-issues>

<https://normconf.com/|NormConf>, a free conference "about all the mundane, behind-the-scenes, how-the-sausage-is-made, middlebrow, unsexy, normcore stuff in the data and ML parts of the tech world" is happening!
The live portion of the conference happens 2022-12-15, and the <https://www.youtube.com/playlist?list=PLYXaKIsOZBstGUTXZXp2azDk8UJhpVVq3|Lightning Talks>, 5 min pre-recorded talks were released today. Come for the memes, stay for the normie goodness!