From our approach page: hashboard.com/approach

7. The data pipeline ends in the minds of people in your organization.

8. Your data team should be distributed and gain as much context on the domain as possible.

9. People who consume data should get context on business rules and get involved in how data is constructed.

5. Data is most powerful when it scales empathy: you can talk to a single user or talk to a single patient. You can't talk to a hundred thousand users. Data can let you talk to 100,000 users.

6. Think about the end-to-end value of your pipeline: from when data is collected to where it is consumed.

4. Only start to use data when you have enough of it. When you're early and you only have ten users, you can just ask them what they think.

The trick is that archiving is not deleting. You can restore from the trash later. But it doesn’t show up and is not discoverable and there is a red banner across the top. So best of both worlds: doesn’t clutter your project, but it’s there if you need to restore it later.

Yup, Hashboard has configurable rules for automatically archiving resources. The default is resources that aren’t verified and aren’t viewed in 30 days are archived. Want to keep skemthi g around? Just verify it. hashboard.com/changelog/au...

Sounds awesome, I'm always happy to chat through it. Maybe not analytics, but happy to point you in the right direction. DMs open

Nope, but happy to get a coffee while you’re in town

@imightbemary.bsky.social on building a business review process: www.datacouncil.ai/talks/buildi...

I'll also thread a couple of old talks.
@abhisivasailam.bsky.social on data trees: www.datacouncil.ai/talks/design...

Here's the form: www.datacouncil.ai/cfp-2025#cfp...

Text-to-sql is just being oversold today. Text-to-sql is being sold as a way for the CEO to replace the data analyst.

When in reality it's a copolit for sql writers (which is awesome and really helpful). All about expectations.

I haven't posted anything on this platform yet and got to 400 followers, not sure who all these people are or how they found me, but seem like real data people

Text-to-sql is just being oversold today. Text-to-sql is being sold as a way for the CEO to replace the data analyst.

When in reality it's a copolit for sql writers (which is awesome and really helpful). All about expectations.

Let data teams work on and incubate products, that's one killer AI usecase I've seen

Two examples:
- Notion's AI product was started by the data team trying to analyze embeddings in customer data without viewing sensitive data.
- Similar story at Replit.

A way to approximate this: some sort of propensity score matching. Take all of the accounts that the team took on, take the obvious predictors / correlates of churn and risk. Match each selected customer with a previous similar account for a year ago and track their future performance.

Naive approach: what was your selection criteria for these accounts? Back data up to a year ago, apply your selection criteria and see what performance is over the following quarter / year. That is there benchmark for that team.