I Built a MILLION-Dollar Data Pipeline and You Can Too

0:00Okay, everyone. Have you ever wondered

0:02how could you build an entire data

0:05pipeline that connects your product

0:07through to your sales team through to

0:08your product team through to your

0:10marketing team through to every single

0:12operational team within your business

0:14and do it all seamlessly and automated?

0:17Well, I've done that and spoiler alert

0:20probably built close to 50 to 100

0:22different micro to macro automations.

0:25combination of agents, combination of

0:27using tools like triggery, clay, n10,

0:31segment, post hog, atio to name just a

0:35few. And what I'm going to do is I'm

0:37going to break this down. I'm going to

0:38start today with a very high level

0:40about, you know, what am I doing here?

0:42How does this flow work? How am I

0:44monitoring clients for signals? How am I

0:46taking data from my product from

0:48Postgress, from Superbase through to my

0:51CRM? How am I doing all of this? And

0:54then over the next five to maybe 10

0:56episodes, I'm going to go deeper into

0:59every single individual part of this

1:01entire web and then break down how I'm

1:04actually using these automations, how I

1:06set them up, and how I think about it.

1:08So, if you like what you're seeing,

1:10like, subscribe, do whatever you need to

1:12do. Stick around because over the next

1:14few weeks, I'm going to be dropping

1:15these videos, and I think they will

1:17hopefully be gold dust for you. So,

1:20let's dive into this kind of um data

1:23pipeline flow that we have going on here

1:25and explain how I'm doing everything.

1:29So, this is um I will preface this is at

1:32point of sign up. We've I've got a load

1:34of other stuff here. I've done videos on

1:36this in the past. I'm not going to go

1:37into that. So, this is at the point of

1:39where a user signs up. What we have is

1:42as they come in, they come in via Clerk.

1:45clerk is basically um syncing and

1:47speaking to two different systems

1:49essentially. One, it's speaking to this

1:52um clay table where I'm basically

1:53pulling through um trigger fire users

1:56and running a bunch of enrichment that's

1:58going to do a lot of different things

2:00and it's also speaking to Postgress

2:03which is hooked up to segment. So

2:05segment's a tool that allows you to do

2:06data pipelines. Segment is essentially

2:09syncing all of my data to Atio. any new

2:12users that come in, it syncs across to

2:15Atio, which is awesome. Um, at the same

2:18time, we're also syncing um some data

2:21from the Clay table that I have. So,

2:23this would be like, you know, who are

2:24they, the demographics, their LinkedIn

2:26URLs, like everything, right? Because

2:28they only give me the first name, last

2:29name, and email. And I take that and I

2:31go get a whole ton of information.

2:34From there, what we're doing is we're

2:36going to start to routt um these

2:38individuals depending on their plan. So

2:41the nice thing because we have segment

2:43hooked up to Postgress when their plans

2:45change i.e. if they go from a free trial

2:47to a paid or they go to a premium or

2:49whatever it is that's going to uh signal

2:53um to Atio to update

2:56um the life cycle and from there it's

3:00then going to update uh inside of the

3:02clay table because I'm constantly

3:04looking at the life uh cycle stage

3:06within the clay table as well. So here's

3:10what happens when they're on um a trial

3:12of some sort. They're coming through

3:14here. We've got the clay table. It's

3:16basically going to check ATIO to see

3:18what stage they're at. It's going to

3:20update Atio with any enrichment plus

3:22plan details. So this would be, you

3:24know, company size, URL, things like I

3:26just mentioned previously.

3:28Then what we have is we have an internal

3:30automation running inside of Atio which

3:32is updating a particular field they have

3:34in there called a life cycle stage. This

3:36is like are they a trial, are they

3:38ex-trial, are they customer, are they

3:40just a lead, whatever it is. And it does

3:42the same for a company level.

3:45Then what we're doing is we've hooked up

3:47triggery into this clay table, which is

3:49then also updating our post hog to see

3:52if they have engaged with us on social

3:54media or not, which is also updating our

3:56atio. So now we have loads of

3:59information. We have who they are,

4:01whether they've engaged with us on

4:02LinkedIn or Twitter or YouTube or Reddit

4:05or whatever it is before. We know we

4:08have the correct life cycle stage of

4:10where they are. And then we are also

4:12checking to see whether they're ICP or

4:14not. If they're ICP, this um triggers an

4:17NA10 workflow that basically sends them

4:20a Gmail inviting them to an onboarding

4:21call. Uh and we also get a Slack

4:24notification saying, "Hey, a new user

4:25signed up."

4:28Um what's happening uh here is we also

4:31create a secondary table within clay. So

4:34we have a master table which is called

4:37the um triggery users. Then we have the

4:40active trial table. Here in the active

4:42trial table we're getting their stripe

4:44information. We're getting their beehive

4:47information. And if they haven't signed

4:49up to our our beehive, then we're

4:50automatically adding them to our

4:52newsletter because that's one of the the

4:54tooss when you sign up with us is you

4:55get automatically added unless you don't

4:57wish to. So we add them automatically to

4:59the beehive and then what we're doing

5:02inside of this particular trial is we've

5:03created a new table as well called the

5:06beehive table inside of clay which then

5:08checks to see the life stage of the

5:11individual from Atio and updates their

5:14custom field within Beehive. So that we

5:16have four different types of newsletters

5:18within the same newsletter if that makes

5:19sense. So we have a newsletter for

5:21clients, we have a newsletter for

5:22exclients, we have a newsletter for

5:24people who are actively on trial and we

5:26can segment the type of content going

5:28forward as a result. And that's what's

5:30happening in our active trial table. So

5:33bear with me. What we've done so far is

5:35we've taken segment and we're syncing

5:37every single day the plan information to

5:39Atio. What's also happening is we have a

5:42master clay table with all of our

5:44trigger fire users where it then sprouts

5:46a new one called the active trial table

5:49and it also then sends all of this

5:50information through to the particular

5:53field that's been created from segment

5:54into all this other information around

5:57engagement um ICP all of that stuff as

6:00well.

6:02So what happens if they don't um convert

6:06on their 14-day trial? Well, this

6:08triggers a new NA10 automation which is

6:11going to fire a custom event into loops

6:13which is our marketing automation email

6:15system that we use. Um, it updates at

6:18from free trial. Um, it basically says

6:21the free trial field that I have is

6:23false and the life cycle goes to X trial

6:25and it updates the person and company

6:27records appropriately. It also then is

6:31on constant lookout. We have an internal

6:33Attoio automation running here that's

6:36seeing if they have attempted to log

6:38back in. How are we tracking that? Well,

6:41we have Post Hog syncing to Atio as

6:43well, which tells us anytime someone

6:45logs in. So, if a user is on a premium

6:47account and attempts to log in, we get a

6:50notification. The notification goes from

6:52Postgog to Atio to Slack. um it

6:55automatically triggers us to find the

6:57mobile number as well and adds that

7:00mobile number into the field inside of

7:01postg. So then we can just get on the

7:03phone and call them and say hey saw you

7:05tried to log in etc which is pretty

7:07cool.

7:09if uh if they converted into a client,

7:13we are going to create again a new table

7:15inside of clay. And this is the closed

7:17one, you know, active clients um table.

7:20And here we're monitoring them and we're

7:23doing a few different things. Um we're

7:25monitoring them for kind of custom

7:26signals. We're understanding if these

7:28people ever change roles. We're

7:30understanding if they have engaged with

7:32X content or X people. Um and we're

7:35doing that via trigger. So, this is

7:37like, okay, are they engaging with

7:38competitors of ours? That's a that's a

7:40potential churn signal. Are they

7:42engaging with content that would suggest

7:43that there is an upsell opportunity? Are

7:46they engaging with content that would

7:47suggest there's a cross-ell opportunity?

7:49And so on and so on. Um, and we're also

7:52monitoring to see if they're posting

7:54about anything that's, you know,

7:55topically relevant to Triggery so that

7:58we can get involved, comment, and

8:00whatnot. And it's also a potential

8:01signal for us depending on the type of

8:03post. All of these signals feed through

8:05to a notification to the sales teams and

8:07updates at appropriately.

8:11Then what we have is as you come on

8:13board it triggers this onboarding flow.

8:16This onboarding flow is controlled from

8:17NA10 which triggers our loops automation

8:20that we have where it be day 1, day 3,

8:21day 6, day 10, you get an email. um as

8:24well. Not only that, we have an NA10

8:27automation flow running in the

8:28background which is querying the post

8:30hog data which is our product analytics

8:32to understand how active is someone,

8:34what features are they using and this

8:36then in turn updates at um and so on and

8:39so on and if anything enters into a

8:42threshold of redeem a churn risk the

8:44team gets instantly notified uh as well.

8:48Not only that, we then are updating the

8:51CL uh client based on month one, month

8:53two, three, 3, six, so on. Um, and we're

8:56basically sending custom content

8:58depending on the life cycle and stage

9:00that they're in.

9:02Now, if they churn and they go to closed

9:06uh lost, first of all, this triggers a

9:08NA10 flow monitoring cancellation which

9:10picks this up. What then happens with

9:13this uh agent is it triggers based on

9:16the stripe change. It looks up their

9:19data within postog to understand what's

9:20happened, what's happened over the last,

9:22you know, the course of their entire

9:24journey with us. We have a churn agent

9:26that's analyzing this data. This data is

9:28then sent to Slack. Um we add this to a

9:31clay table and the reason we add this to

9:34the clay table is we run some further

9:35enrichment that we deem necessary and

9:37then we update atio. A common question

9:40that I might get asked is why do you use

9:41NA10 and clay? Well, I use NA10 to do a

9:44lot of um infrastructure and

9:46organizational um flows and if the if

9:50the automation is more complex and

9:51requires better agent capabilities, I

9:54use clay primarily purely just for data

9:56enrichment um to to do things there. So

9:59that's kind of when I that's when I

10:00discern the difference between the two.

10:03After that, we then um the customer

10:06success team get notified by the agent

10:08that this person has churned along with

10:09all this information. We update the

10:11beehive field. Uh we update the custom

10:14cancellation reasons inside of Atio. And

10:16then we're doing the same thing. We're

10:17monitoring this close loss PLG signal,

10:20which is are they attempting to log in

10:21ever since they've churned? Uh if they

10:23do, we reach out off the back end uh of

10:26that. And that so far is everything that

10:30we have. Now, I'm actually continuously

10:32adding to this. Would you believe it or

10:33not? But what I'm going to do, like I

10:34said, this is high level what's going

10:36on. I'm breaking down every single

10:39section, every single agent, every

10:41single automation that I have over the

10:42course of multiple videos coming uh over

10:45the next um month or so. So, stay tuned.

10:48This is going to get really interesting

10:50and really exciting as I show what I'm

10:52doing. I've been building it solidly for

10:53the last couple of weeks, getting this

10:55all in place. And there are so much

10:56nuances from the amount of Atto

10:58workflows that we have, from the amount

11:00of NA10 workflows, from the amount of

11:02clay tables that we have, how we've

11:04created this data orchestration with the

11:05data pipeline. It's going to get

11:07technical. It's going to get good. It's

11:08going to get really interesting. So, if

11:10you're a founder, if you're revops, if

11:12you're ops, uh if you're marketing, if

11:14whatever you are, um GTM focused, this

11:18one is for