High Signal Podcasts Evidence ledger
Method
Browse
← Back to evidence

Evidence receipt / belief

Published · transcript-backed

Austin Hay: belief

13 Aug 2023 Lenny's Podcast The ultimate guide to Martech | Austin Hay (Reforge, Ramp, Runway)

“I think that's a great move if you do not have a lot of engineering resources because you're not spending a ton of time and energy on a warehouse and all the modeling that comes with it, you're just spending time to implement one SDK.”

— Austin Hay

Source trail

Everything needed to verify it.

Speaker
Austin Hay
Attribution
Verified speaker
Claim type
belief
Recorded
13 Aug 2023
Publisher
Lenny's Podcast

Transcript context

…Great. Okay. So, B2C, if you back up to 2016, 2017, you have segment and the rise of the CDP. Consumer based businesses have to collect a user and tie a bunch of data to them and then track their actions to send it out to performance ad networks and email marketing tools and product analytics tools. And so you would see this very commoditized stack. It would be like CDP in the middle bunch of tools connected. The promise of the CDP was you integrate one SDK, your engineers don't hate you send all the data to the other tools you can create audiences. Great. Lasted for a long time. The thing about it though that I think really changed around 2020 is that the cost of ownership of warehousing became much cheaper. And so 2021, you start getting to the place where it actually makes a lot of sense and is really easy to store all your data in a warehouse model all your data in the warehouse, and to do it without needing a vast data team. I would say Airbnb was probably doing all this well before anybody else was, but they had the main advantage of a lot of money and a lot of resources. So, now come 2020, it's cost-efficient to have a data team with your own warehouse and to manage data centrally in something like Snowflake. So, now this question is like, okay, well we got to get data into the warehouse, but how do we move data around is totally different. And that's what really led to the rise of reverse ETLs. So, now you can actually build your own CDP and lots of businesses already have, I'm consulting with a well known financial trading platform a couple of months back, and they have a CDP, they have all this internal data in their warehouse, but they have not been able to activate it because it's pretty old architecture. Everything's batch based end of the day. What they need is a reverse ETL. They don't need to take that data and just get it out into the world. So, they need the reverse ETL component or the transformation component of a CDP. And so I'd say now today when we think about B2C businesses, you can either go to the traditional route, buy CDP, hook up all your tools, third party. I think that's a great move if you do not have a lot of engineering resources because you're not spending a ton of time and energy on a warehouse and all the modeling that comes with it, you're just spending time to implement one SDK. I think if simplicity is the name of the game for your business, CDP, Centralized Stack, great move. If you are an advanced engineering culture and you are cutting edge and you're going to do a bunch of modeling in DBT and you already have Snowflake, you should move towards a model of using a reverse ETL. What it means is that there's a way to get your data into the warehouse and then how you activate it is completely independent from the CDP. And so what that means is actually you can have lots of different variations of the stack. You could use Amplitude as your CDP, collect all your data, stream it into Snowflake. dependent from the CDP. And so what that means is actually you can have lots of different variations of the stack. You could use Amplitude as your CDP, collect all your data, stream it into Snowflake. They actually now have an integration with Snowflake that lets you feed data directly out of Snowflake, and then you could use a reverse ETL to just pipe that data wherever you want. There's a really good section though, again, sorry to self aggrandize, but there's a really good section in the Reforge module this fall that talks about what happens when you have multiple ways to move data. You buy amplitude for your CDP and you're moving data to your warehouse. Amplitude is a bunch of integrations, but you also have reverse ETL and you can move data out of your warehouse. Where do you choose? And I would say a lot of businesses get in trouble when they don't have a methodology or a system for how and when to move data from one place to the other, so they just do it haphazardly, right? And the key in systems management is you want to design a process for doing it some type of waterfall or mental model for when it makes sense to move data directly from Amplitude, which is the ingestion point of your data stream or from the warehouse where you can model it and make it better. I think the key is just having a philosophy and approach. There's not really one answer, but that's all B2C. So, B2B I would say ... Yeah, go ahead.…

Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.

Search evidence