[iamacyborg]

As a Marketing Ops guy, this gives me pause.

> nice and tidy. And then the marketers can do what they need in the platform, and they're able to iterate and learn at a much faster pace. It changes the whole game.

The reality is, marketers will frequently plug in any and every thing they can without thinking about the broader context. Sure you can plug all this stuff together, or you can spend a bit of time with ETL and rETL tools and develop a stack that isn't reliant on a bunch of third party silos.

[boc]

Is your goal to have a martech stack that doesn't rely on third party silos, or is your goal to have your lead-gen form populate your Airtable CRM for your new experimental acquisition campaign?

The marketer is being asked to acquire users, not make it look pretty behind the scenes. In the testing phase, it's way better to use Zapier to get things done fast. Then you can revisit and scale later with a proper solution if it turns out to be a productive campaign.

[iamacyborg]

My goal is to ensure some degree of process sanity exists to ensure data processing complies with the law and that systems interoperate effectively.

I don't think anyone using Airtable as a CRM has those sorts of concerns, however.

[boc]

Right - it sounds like you're processing data at a very different scale from the typical Zapier user.

[iamacyborg]

Yeah, potentially, though I think you’d be surprised at the size of teams still using it.

[cm2012]

It's still incredibly useful at billion dollar companies. Even more so since the development process is even harder to get through.

[ouEight12]

> revisit and scale later with a proper solution

But there in is the problem. Most Marketers I've had the pleasure of working with/for will never come back and implement the proper (scalable, secured, etc) solution. By the time this first ad-hoc creation has either failed horribly or taken off, they're off on the next new "idea".. there's no time to property revisit that last one, why would we? It's either dead and forgotten or working as I gloriously intended!

I've actually already seen Zapier go 'full lifecycle' in two separate environments. It went from "cautiously approved" to "banned by both name and by functionality" both times due to the marketing team zapping their collected data off to, lets call them.. overly public output storage locations.

[emteycz]

I would have no idea where to start with rETL and I'm a software engineer... Search turns up some medicine and other non-related stuff.

If it's a typo, then sorry. But ETL is the same beast. I have few ideas where to start thanks to my programming background, but I can't imagine the marketing department guys do any of it.

Non-ideal but less expensive is still better than nothing (or too expensive to stay profitable). Not all operations are huge companies that can afford IT teams working on their behalf.

[iamacyborg]

Basically, reverse ETL, ie get data from the warehouse to operational platforms. The two big players are Census and Hightouch.

https://www.getcensus.com/

https://hightouch.io/

Both offer no-code tools to help non-technical users make use of the data.

[maxerickson]

What does the L mean then?

[me_bx]

E - Extract

T - Transform

L - Load

[emteycz]

Thanks for the links, I'll have a look.

[borisjabes]

it refers to a somewhat poorly named concept known as reverse ETL, something we started working on back in 2018 with our product Census (getcensus.com). We launched it on HN in 2020, though it's come a long way since then :-) https://news.ycombinator.com/item?id=23034642

[emteycz]

Nice, definitely solves some problems I had in the past. I'll keep your product in mind, I like it very much.

[bduerst]

Think of rETL as an on-demand data lake that doesn't actually store any data (for the most part).

Yeah you can pull data and create snapshots to preserve views, but given the direction privacy and regulatory environments have been going with user data, it's incredibly sexy to be able to not have to dump data in yet another data warehouse just to use a new tool.

[eru]

Perhaps the proper programming term might be co-ETL or co-data-lake? (Where we use 'co' in the same sense as eg http://blog.ezyang.com/2012/10/duality-for-haskellers/ )

[bduerst]

>or you can spend a bit of time with ETL and rETL tools and develop a stack that isn't reliant on a bunch of third party silos

This is the real future of 'unbundling'.

I love reading this article because while it sets up the problem and solution proposition quite well, all I could think of was how the author is proving reverse-ETL engines (& not their own platform) are the best investment right now.

[georgesequeira]

At the end of the day, most business owners don't want to connect all these tools together. It just allows them to accomplish business outcomes. What makes their job hard is keeping track where all the data is and how to know if things are working.

Solutions to that problem can take shape as a CRM (and keeping it updated) or something as dumb as a dashboard of dashboards.

[ethbr0]

Agreed. It's a common trap of technologist thinking: if we gave people better access to primatives, they'd have more power to assemble what they wanted!

In fact, it's "assembling" that most people hate. Which is why products that sacrifice power to limit assembly tend to be successful.